Human Computer

Interaction:

Concepts, Methodologies,
Tools, and Applications
Panayiotis Zaphiris
City University of London, UK
Chee Siang Ang
City University of London, UK

Volume I

Information Science reference

Hershey New York

Acquisitions Editor:
Development Editor:
Senior Managing Editor:
Managing Editor:
Typesetter:

Cover Design:
Printed at:

Kristin Klinger
Kristin Roth
Jennifer Neidig
Jamie Snavely
Michael Brehm, Jeff Ash, Carole Coulson, Elizabeth Duke,
Jennifer Henderson, Chris Hrobak, Sean Woznicki
Lisa Tosheff
Yurchak Printing Inc.

Published in the United States of America by

Information Science Reference (an imprint of IGI Global)
701 E. Chocolate Avenue, Suite 200
Hershey PA 17033
Tel: 717-533-8845
Fax: 717-533-8661
E-mail: [email protected]
Web site: http://www.igi-global.com/reference
and in the United Kingdom by
Information Science Reference (an imprint of IGI Global)
3 Henrietta Street
Covent Garden
London WC2E 8LU
Tel: 44 20 7240 0856
Fax: 44 20 7379 0609
Web site: http://www.eurospanbookstore.com

Copyright 2009 by IGI Global. All rights reserved. No part of this publication may be reproduced, stored or distributed in any form or by
any means, electronic or mechanical, including photocopying, without written permission from the publisher.
Product or company names used in this set are for identification purposes only. Inclusion of the names of the products or companies does not
indicate a claim of ownership by IGI Global of the trademark or registered trademark.

Library of Congress Cataloging-in-Publication Data

Human computer interaction : concepts, methodologies, tools, and applications / Panayiotis Zaphiris and Chee Siang Ang, editors.
p. cm.
Includes bibliographical references and index.
Summary: "This reference book penetrates the human computer interaction (HCI) field a wide variety of comprehensive research papers
aimed at expanding the knowledge of HCI"--Provided by publisher.
ISBN 978-1-60566-052-3 (hardcover) -- ISBN 978-1-60566-053-0 (ebook)
1. Human-computer interaction. I. Zaphiris, Panayiotis. II. Ang, Chee Siang.
QA76.9.H85H8563 2008
004.01'9--dc22
2008035756

British Cataloguing in Publication Data

A Cataloguing in Publication record for this book is available from the British Library.
All work contributed to this book set is original material. The views expressed in this book are those of the authors, but not necessarily of
the publisher.
If a library purchased a print copy of this publication, please go to http://www.igi-global.com/agreement for information on activating
the library's complimentary electronic access to this publication.

Editor-in-Chief

Mehdi Khosrow-Pour, DBA

Editor-in-Chief
Contemporary Research in Information Science and Technology, Book Series

Associate Editors
Steve Clarke
University of Hull, UK
Murray E. Jennex
San Diego State University, USA
Annie Becker
Florida Institute of Technology USA
Ari-Veikko Anttiroiko
University of Tampere, Finland

Editorial Advisory Board

Sherif Kamel
American University in Cairo, Egypt
In Lee
Western Illinois University, USA
Jerzy Kisielnicki
Warsaw University, Poland
Keng Siau
University of Nebraska-Lincoln, USA
Amar Gupta
Arizona University, USA
Craig van Slyke
University of Central Florida, USA
John Wang
Montclair State University, USA
Vishanth Weerakkody
Brunel University, UK

Additional Research Collections found in the

Contemporary Research in Information Science and Technology
Book Series
Data Mining and Warehousing: Concepts, Methodologies, Tools, and Applications
John Wang, Montclair University, USA 6-volume set ISBN 978-1-60566-056-1
Electronic Business: Concepts, Methodologies, Tools, and Applications
In Lee, Western Illinois University 4-volume set ISBN 978-1-59904-943-4
Electronic Commerce: Concepts, Methodologies, Tools, and Applications
S. Ann Becker, Florida Institute of Technology, USA 4-volume set ISBN 978-1-59904-943-4
Electronic Government: Concepts, Methodologies, Tools, and Applications
Ari-Veikko Anttiroiko, University of Tampere, Finland 6-volume set ISBN 978-1-59904-947-2
Knowledge Management: Concepts, Methodologies, Tools, and Applications
Murray E. Jennex, San Diego State University, USA 6-volume set ISBN 978-1-59904-933-5
Information Communication Technologies: Concepts, Methodologies, Tools, and Applications
Craig Van Slyke, University of Central Florida, USA 6-volume set ISBN 978-1-59904-949-6
Intelligent Information Technologies: Concepts, Methodologies, Tools, and Applications
Vijayan Sugumaran, Oakland University, USA 4-volume set ISBN 978-1-59904-941-0
Information Security and Ethics: Concepts, Methodologies, Tools, and Applications
Hamid Nemati, The University of North Carolina at Greensboro, USA 6-volume set ISBN 978-1-59904-937-3
Medical Informatics: Concepts, Methodologies, Tools, and Applications
Joseph Tan, Wayne State University, USA 4-volume set ISBN 978-1-60566-050-9
Mobile Computing: Concepts, Methodologies, Tools, and Applications
David Taniar, Monash University, Australia 6-volume set ISBN 978-1-60566-054-7
Multimedia Technologies: Concepts, Methodologies, Tools, and Applications
Syed Mahbubur Rahman, Minnesota State University, Mankato, USA 3-volume set ISBN 978-1-60566-054-7
Virtual Technologies: Concepts, Methodologies, Tools, and Applications
Jerzy Kisielnicki, Warsaw University, Poland 3-volume set ISBN 978-1-59904-955-7

Free institution-wide online access with the purchase of a print collection!

Information Science reference

Hershey New York

Order online at www.igi-global.com or call 717-533-8845 ext.100

MonFri 8:30am5:00 pm (est) or fax 24 hours a day 717-533-7115

List of Contributors

Aaltio, Iiris / Lappeenranta University of Technology, Finland....................................................... 2030

Abraham, J. R. / University of the West of England, UK................................................................... 376
Abutarbush, Sameeh M. / University of Prince Edward Island, Canada....................................... 1656
Adam, Alison / University of Salford, UK........................................................................................ 2435
gerfalk, Pr J. / University of Limerick, Ireland & rebro University, Sweden........................... 2341
Ahuja, Manju / Indiana University, USA......................................................................................... 1736
Ala-Rmi, Katariina / University of Oulu, Finland......................................................................... 2037
Albert, Sylvie / Laurentian University, Canada............................................................................... 1915
Aldiri, Khalid / University of Bradford, UK..................................................................................... 2235
Almeida, Hyggo / Federal University of Campina Grande, Brazil.................................... 228, 254, 279
Almendros-Jimenez, Jesus M. / Universidad de Almeria, Spain...................................................... 404
Alpert, Sherman R. / IBM T. J. Watson Research Center, USA......................................................... 177
Altenhofen, Michael / SAP Research, CEC Karlsruhe, Germany................................................... 2144
Anagnostopoulos, Christos / University of Athens, Greece............................................................. 1096
Andersen, Peter Bgh / University of Aarhus, Denmark................................................................... 552
Andrade, Stephen C. / Johnson & Wales University, USA.............................................................. 2590
Andrea, Carugati / Catholic University of Lille, France................................................................... 489
Anke, Juergen / Dresden University of Technology, Germany.......................................................... 168
Ardissono, Liliana / Universit de Torino, Italy................................................................................ 791
Avery Gomez, Elizabeth / New Jersey Institute of Technology, USA.............................................. 1334
Baber, Chris / The University of Birmingham, UK.................................................................... 80, 1134
Barbosa, Nadia / Federal University of Campina Grande, Brazil..................................................... 279
Beckwith, Laura / Oregon State University, USA................................................................................ 25
Bellotti, Francesco / University of Genoa, Italy............................................................................... 1234
Ben Zakour, Amel / FSJEGJ, Tunisia................................................................................................ 132
Bernsteiner, Reinhard / University for Health Sciences, Medical Informatics
and Technology, Austria..................................................................................................................... 1402
Berta, Riccardo / University of Genoa, Italy.................................................................................... 1234
Bharati, Pratyush / University of Massachusetts, Boston, USA...................................................... 1956
Bhattacherjee, Anol / University of South Florida, USA................................................................. 1608
Bias, Randolph G. / The University of Texas at Austin, USA........................................................... 2404
Bieber, Michael / New Jersey Institute of Technology, USA............................................................. 1334
Blocher, J. Michael / Northern Arizona University, USA................................................................. 1547
Blomqvist, Kirsimarja / Lappeenranta University of Technology, Finland.................................... 2181
Bondarouk, Tanya / University of Twente, The Netherlands........................................................... 2214

Bonk, Curtis J. / Indiana University, USA....................................................................................... 1262

Bornhvd, Christof / SAP Research, Research Center Palo Alto, USA.......................................... 2144
Bourguet, Marie-Luce / Queen Mary University of London, UK........................................................ 95
Bowen, Jonathan P. / London South Bank University, UK.............................................................. 1629
Boyle, Randall J. / University of Utah, USA.................................................................................... 1970
Brey, Philip / University of Twente, The Netherlands......................................................................... 154
Bruegger, Pascal / University of Fribourg, Switzerland................................................................... 1015
Brynskov, Martin / Center for Interactive Spaces, University of Aarhus, Denmark......................... 552
Bublitz, Frederico / Federal University of Campina Grande, Brazil................................................ 279
Burnett, Gary / University of Nottingham, UK.................................................................................. 532
Burnett, Margaret / Oregon State University, USA............................................................................. 25
Bushey, Robert G. / AT&T Laboratories, USA................................................................................ 2404
Bygstad, Bendik / Norwegian School of Information Technology, Norway..................................... 2090
Caronia, Letizia / University of Bologna, Italy................................................................................ 2655
Castelfranchi, Cristiano / Institute for Cognitive Sciences and Technology, National Research
Council of Italy, Italy......................................................................................................................... 2316
Catarci, Tiziana / University of Rome La Sapienza, Italy...................................................... 872, 928
Chaudhury, Adhijit / Bryant University, USA................................................................................. 1956
Chen, Chaomei / Drexel University, USA........................................................................................ 1678
Chen, Yen-Jung / National University of Tainan, Taiwan................................................................ 1330
Cheuk, Bonnie Wai-yi / Improvement Service for
the Scottish Local Government, Scotland.......................................................................................... 1905
Cigognini, M. E. / University of Florence (IT), Italy.......................................................................... 961
Clark, Chris / University of Saskatchewan, Canada........................................................................ 1656
Cohn, Jonathan / UCLA, USA......................................................................................................... 2530
Compagna, Luca / SAP Research, France....................................................................................... 2424
Cox, Anna L. / University College London, UK............................................................................... 1154
Crombie, David / DEDICON S.r.l., The Netherlands....................................................................... 1350
Cyr, Dianne / Simon Fraser University, Canada.............................................................................. 1761
da Cunha Borelli, Walter / State University of Campinas, Brazil................................................... 1037
Dai, Honghua / DePaul University, USA............................................................................................ 732
Daniel, Ben K. / University of Saskatchewan, Canada...................................................................... 979
Das, Sajal / University of Texas at Arlington, USA........................................................................... 1173
Davenport, Glorianna / The Media Laboratory, USA..................................................................... 1521
Davis, Donald D. / Old Dominion University, USA.......................................................................... 2023
de Ftima Queiroz Vieira Turnell, Maria / Universidade Federal de
Campina Grande (UFCG), Brazil..................................................................................................... 2573
De Gloria, Alessandro / University of Genoa, Italy......................................................................... 1234
de Queiroz, Jos Eustquio Rangel / Universidade Federal de
Campina Grande (UFCG), Brazil..................................................................................................... 2573
de Sousa Ferreira, Danilo / Universidade Federal de Campina Grande (UFCG), Brazil............. 2573
Deng, Li / Microsoft Research, USA................................................................................................... 997
Deng, Liqiong / University of West Georgia, USA............................................................................. 850
Diaper, Dan / Middlesex University, UK.............................................................................................. 70
Dix, Alan / Lancaster University, UK................................................................................................. 928
Doyle, D. John / Cleveland Clinic Foundation, USA......................................................................... 483

Doyle, Linda / University of Dublin, Ireland.................................................................................... 2266

Edenius, Mats / Stockholm School of Economics, Sweden............................................................... 2376
El Louadi, Mohamed / University of Tunis, Tunisia.......................................................................... 123
Elias, Hadzilias / Catholic University of Lille, France....................................................................... 489
Erlandson, Benjamin E. / Arizona State University, USA................................................................. 698
Esteves, Jose / Universitate Politecnica de Catalunya, Barcelona, Spain....................................... 2132
Evangelou, Christina E. / Research Academic Computer Technology Institute, Greece................ 1044
Falcone, Rino / Institute for Cognitive Sciences and Technology, National Research
Council of Italy, Italy......................................................................................................................... 2316
Fan, Haiyan / Texas A&M University, USA........................................................................................ 850
Farmer, Rod / The University of Melbourne, Australia..................................................................... 899
Ferreira, Glauber / Federal University of Campina Grande, Brazil......................................... 228, 279
Filippini-Fantoni, Silvia / The University of Paris I Sorbonne University, France........................ 1629
Fischbach, Kai / University of Cologne, Germany........................................................................... 2191
Fitzgerald, Robert / University of Canberra, Australia................................................................... 1459
Fletcher, Gordon / University of Salford, UK.................................................................................. 1871
Floridi, Luciano / Universit degli Studi di Bari, Italy and Oxford University, UK........................ 2450
Foth, Marcus / Queensland University of Technology, Australia....................................................... 589
French, Sheila / Manchester Metropolitan University, UK.............................................................. 2296
Fu, Guohong / The University of Hong Kong, Hong Kong.............................................................. 1595
Furtado, Elizabeth / University of Fortaleza, Brazil......................................................................... 448
Fuschi, David Luigi / GIUNTI Labs S.r.l., Italy............................................................................... 1350
Fussell, Susan R. / Carnegie Mellon University, USA..................................................................... 1801
Gabrielli, Silvia / University of Rome La Sapienza, Italy.............................................................. 928
Gallagher, Kevin / Northern Kentucky University, USA.................................................................. 2052
Gefen, David / Drexel University, USA............................................................................................ 2644
Gena, Cristina / Universit de Torino, Italy....................................................................................... 791
George, Joey F. / Florida State University, USA.............................................................................. 1970
Georgiadis, Christos K. / University of Macedonia, Thessaloniki, Greece....................................... 393
Geri, Nitza / The Open University of Israel, Israel.......................................................................... 2644
Gkotsis, George / Research Academic Computer Technology Institute, Greece.............................. 1044
Gold, Stuart S. / DeVry University, USA.......................................................................................... 1864
Goldkuhl, Gran / Linkping University, Sweden & Jnkping International
Business School, Sweden......................................................................................................... 2071, 2341
Greenhill, Anita / The University of Manchester, UK............................................................ 1871, 2259
Greitzer, Frank L. / Pacific Northwest National Laboratory, USA................................................. 2693
Griffith, Douglas / General Dynamics Advanced Information Systems, USA.................................. 2693
Gruba, Paul / The University of Melbourne, Australia...................................................................... 899
Guo, Huaqun / Institute for Infocomm Research
and National Univerity of Singapore, Singapore............................................................................... 1087
Gurevych, Iryna / Technische Universitt Darmstadt, Germany.......................................................... 1
Gurram, Rama / SAP Research, USA................................................................................................ 102
Hadjiethymiades, Stathes / University of Athens, Greece............................................................... 1096
Hall, Jon G. / The Open University, UK............................................................................................. 713
Hanrahan, Michael / Bates College, USA....................................................................................... 1251
Heinemann, Andreas / Technical University of Darmstadt, Germany.............................................. 333

Henttonen, Kaisa / Lappeenranta University of Technology, Finland............................................. 2181

Her, One-Soon / National Chengchi University, Taiwan.................................................................. 2499
Herdin, Thomas / University of Salzburg, Austria........................................................................... 1055
Herring, Susan C. / Indiana University, USA.................................................................................. 1736
Hirsbrunner, Bat / University of Fribourg, Switzerland................................................................ 1015
Hobbs, Dave / University of Bradford, UK....................................................................................... 2235
Hofkirchner, Wolfgang / University of Salzburg, Austria................................................................ 1055
Hohmann, Claus / Bergische Universitaet Wuppertal, Germany.................................................... 2462
Hornik, Steven / University of Central Florida, USA...................................................................... 1885
Horowitz, Ellis / University of Southern California, USA.................................................................. 659
Hsu, HY Sonya / Southern Illinois University, USA......................................................................... 1649
Hsu, Jeffrey / Fairleigh Dickinson University, USA........................................................................ 1308
Hsu, Shang Hwa / National Chiao Tung University, Taiwan............................................................. 516
Huang, Haiyan / The Pennsylvania State University, USA.............................................................. 1753
Huang, Rui / University of Texas at Arlington, USA........................................................................ 1173
Hubona, Geoffrey S. / Georgia State University, USA.................................................................... 2473
Huffaker, David A. / Northwestern University, USA........................................................................ 1484
Hrst, Wolfgang / Albert-Ludwigs-Universitt Freiburg, Germany................................................ 1436
Hwang, Gwo-Jen / National University of Tainan, Taiwan............................................................. 1330
Iribarne, Luis / Universidad de Almeria, Spain................................................................................. 404
Isomki, Hannakaisa / University of Lapland, Finland.................................................................... 605
James, Frankie / SAP Research, USA................................................................................................ 102
Jaspers, M.W.M. / University of Amsterdam, The Netherlands......................................................... 622
Jefsioutine, Marie / University of Central England - Birmingham, UK............................................ 432
Jin, Dal Yong / Simon Fraser University, Canada............................................................................ 1200
Jordan, Robert / U.S. Bureau of Labor Statistics, USA..................................................................... 290
Jos Luzn, Mara / University of Zaragoza, Spain........................................................................ 1291
Julien, Christine / The University of Texas at Austin, USA.............................................................. 2708
Kabadayi, Sanem / The University of Texas at Austin, USA............................................................ 2708
Kacmar, Charles J. / The University of Alabama, USA................................................................... 1970
Karali, Izambo / University of Athens, Greece................................................................................ 1096
Karousos, Nikos / Research Academic Computer Technology Institute, Greece............................. 1044
Khattab, Ishraga / Brunel University, UK....................................................................................... 1569
Kidd, Terry T. / University of Texas School of Public Health, USA................................................ 1468
Kimani, Stephen / Jomo Kenyatta University of Agriculture and Technology, Kenya, &
University of Rome La Sapienza, Italy.................................................................................... 872, 928
Kirakowski, Jurek / University College Cork, Ireland...................................................................... 361
Klassen, Myungsook / California Lutheran University, USA.......................................................... 1729
Kleinschmidt, Joo Henrique / State University of Campinas, Brazil........................................... 1037
Klinker, Gudrun / Technische Universitt Mnchen, Germany........................................................ 629
Klobas, Jane / Bocconi University, Italy & University of Western Australia, Australia.................. 1283
Knight, James / The University of Birmingham, UK....................................................................... 1134
Knight, John / University of Central England - Birmingham, UK.................................................... 432
Knott, Benjamin A. / U.S. Air Force Research Laboratory, USA.................................................... 2404
Kolomvatsos, Kostas / National & Kapodistrian University of Athens, Greece.............................. 1503
Kortum, Kortum / Rice University, USA......................................................................................... 2404

Kosonen, Miia / Lappeenranta University of Technology, Finland.................................................. 2181

Koukia, Spiridoula / University of Patras, Greece............................................................................ 992
Kulviwat, Songpol / Hofstra University, USA.................................................................................. 1649
Kumar, Nanda / Baruch College, The City University of New York, USA......................................... 212
Lasnik, Vincent E. / Independent Knowledge Architect, USA......................................................... 2106
LeBrasseur, Rolland / Laurentian University, Canada.................................................................... 1915
Lee, Cha-Hoang / National Chiao Tung University, Taiwan.............................................................. 516
Lee, Suet Chun / BUSINEX, Inc., USA.............................................................................................. 760
Leow, Winnie C. H. / Singapore Polytechnic, Singapore................................................................... 807
Lin, Meng-Fen Grace / University of Houston, USA....................................................................... 1262
Lind, Mikael / University College of Bors, Sweden, & Linkping University, Sweden................. 2071
Link, Lisa / Flensburg University of Applied Sciences, Germany...................................................... 316
Lippert, Susan K. / Drexel University, USA.................................................................................... 2118
Liu, Hugo / The Media Laboratory, USA......................................................................................... 1521
Liu, Tong / Massey University, New Zealand................................................................................... 2675
Lobo, Natasha / Drexel University, USA.......................................................................................... 1678
Lodi, Stefano / University of Bologna, Italy....................................................................................... 872
Lohmann, Katharina L. / University of Saskatchewan, Canada.................................................... 1656
Loi, Daria / RMIT University, Australia........................................................................................... 1388
Long, Yuan / Colorado State University-Pueblo, USA....................................................................... 645
Lotz, Volkmar / SAP Research, France............................................................................................ 2424
Loureiro, Emerson / Federal University of Campina Grande, Brazil............................... 228, 254, 279
Love, Steve / Brunel University, UK................................................................................................. 1569
Luke, Kang-Kwong / The University of Hong Kong, Hong Kong................................................... 1595
Ma, Will Wai-kit / Hong Kong University of Science and Technology, China................................ 2287
Machwe, A. / University of the West of England, UK......................................................................... 376
Maes, Pattie / The Media Laboratory, USA...................................................................................... 1521
Maier-Rabler, Ursula / University of Salzburg, Austria.................................................................. 1055
Major, Debra A. / Old Dominion University, USA........................................................................... 2023
Malaka, Rainer / Bremen University, Bremen................................................................................. 2627
Margarone, Massimiliano / University of Genoa, Italy.................................................................. 1234
Markellou, Penelope / University of Patras, Greece & Research Academic and Computer
Technology Institute, Greece.................................................................................................... 1079, 2164
Mason, Hilary / Johnson & Wales University, USA......................................................................... 2590
Mason, Robert M. / University of Washington, USA....................................................................... 2052
Mawson-Lee, Kym / University of New South Wales, Australia...................................................... 1707
McNamara, Niamh / University College Cork, Ireland..................................................................... 361
Melliar-Smith, P. M. / University of California, Santa Barbara, USA............................................ 2556
Michael, Ian / Zayed University, UAE.............................................................................................. 1673
Mishra, Sanjaya / Indira Gandhi National Open University, India................................................. 1777
Mitchell, Stella / IBM T. J. Watson Research, USA.......................................................................... 1213
Mobasher, Bamshad / DePaul University, USA................................................................................ 732
Mohamed, Khaireel A. / Albert-Ludwigs-Universitt Freiburg, Germany..................................... 1436
Morgan, Konrad / University of Bergen, Norway........................................................................... 2274
Morgan, Madeleine / University of Bergen, Norway....................................................................... 2274
Moser, Louise E. / University of California, Santa Barbara, USA.................................................. 2556

Muhle, Steffen / University of Cologne, Germany........................................................................... 2191

Mhlhuser, Max / Technische Universitt Darmstadt, Germany........................................................ 1
Murphy, Elizabeth D. / U.S. Census Bureau, USA............................................................................ 192
Naylor, Jonathan M. / Ross University School of Veterinary
Medicine, West Indies........................................................................................................................ 1656
Ngoh, Lek-Heng / Institute for Infocomm Research, A*STAR, Singapore....................................... 1087
Nousia, Dora / Research Academic Computer Technology Institute, Greece................................... 1044
Numerico, Teresa / London South Bank University, UK.................................................................. 1629
OBrien, David / University of Saskatchewan, Canada..................................................................... 979
OConnell, Theresa A. / Humans & Computers, Inc., USA............................................................... 192
OHare, Gregory M. P. / University College Dublin, Ireland......................................................... 2614
Oberle, Daniel / SAP Research, CEC Karlsruhe, Germany............................................................. 2144
Ocker, Rosalie J. / The Pennsylvania State University, USA........................................................... 1745
Ogan, Christine / Indiana University, USA...................................................................................... 1736
OGrady, Michael J. / University College Dublin, Ireland............................................................. 2614
Olaniran, Bolanle A. / Texas Tech University, USA............................................................... 1423, 1849
Oliveira, Loreno / Federal University of Campina Grande, Brazil........................................... 254, 279
Ong, Bee / University of Leeds, UK.................................................................................................. 1350
Ostermann, Herwig / University for Health Sciences,
Medical Informatics and Technology, Austria................................................................................... 1402
Oulasvirta, Antti / Helsinki Institute for Information Technology, Finland......................................... 20
Owens, Emiel / Texas Southern University, USA.............................................................................. 1468
Pallotta, Vincenzo / University of Fribourg, Switzerland................................................................ 1015
Panayiotaki, Angeliki / University of Patras, Greece...................................................................... 1079
Paravastu, Narasimha / Central State University, USA.................................................................. 2644
Parchoma, Gale / University of Saskatchewan, Canada.................................................................. 1656
Parmee, I. C. / University of the West of England, UK....................................................................... 376
Passerini, Katia / New Jersey Institute of Technology, USA............................................................ 1334
Pastor, Joan / Universitat Internacional de Catalunya, Barcedona, Spain...................................... 2132
Pavlovski, Christopher J. / IBM Corporation, Australia................................................................ 1213
Pearson, Will / Sheffield Hallam University, UK................................................................................ 220
Peebles, David / University of Huddersfield, UK.............................................................................. 1154
Perkusich, Angelo / Federal University of Campina Grande, Brazil................................ 228, 254, 279
Pettenati, M. C. / University of Florence (IT), Italy........................................................................... 961
Polley, Lydden / University of Saskatchewan, Canada.................................................................... 1656
Polovina, Simon / Sheffield Hallam University, UK........................................................................... 220
Porterfield, Sharon / University of Saskatchewan, Canada............................................................ 1656
Qahwaji, Rami / University of Bradford, UK................................................................................... 2235
Quah, Jon T. S. / Nanyang Technological University, Singapore...................................................... 807
Rahman, Hakikur / SDNP Bangladesh, Bangladesh...................................................................... 2014
Rahman, Lutfor / Association for Advancement of Information
Technology (AAIT), Bangladesh........................................................................................................ 1591
Rahman, Nusrat / University of Dhak and Association
for Advancement of Information Technology (AAIT), Bangladesh.................................................... 1591
Rapanotti, Lucia / The Open University, UK..................................................................................... 713
Reddick, Christopher G. / The University of Texas at San Antonio, USA...................................... 1687

Richter, Kai / Computer Graphics Centre (ZGDV), Germany........................................................... 323

Rigou, Maria / University of Patras, Greece & Research Acaademic Computer Technology
Institute, Greece......................................................................................................................... 992, 2164
Rittgen, Peter / University College of Bors, Sweden...................................................................... 2390
Robinson, Jean C. / Indiana University, USA.................................................................................. 1736
Roth, Volker / OGM Laboratory LLC, USA....................................................................................... 323
Ruzic, Fjodor / Institute for Informatics, Croatia............................................................................ 2324
Saddler, Paula F. / Association of Certified Fraud Examiners, USA............................................... 2023
Sajjapanroj, Suthiporn / Indiana University, USA......................................................................... 1262
Salah Hamdi, Mohamed / University of Qatar, Qatar.................................................................... 1118
Salovaara, Antti / Helsinki Institute for Information Technology, Finland.......................................... 20
Snchez-Franco, Manuel Jess / University of Seville, Spain........................................................ 1995
Sandberg, Karl W. / Mid Sweden University, Sweden......................................................................... 60
Sandor, Christian / Technische Universitt Mnchen, Germany....................................................... 629
Sanford, Clive / Aalborg University, Denmark................................................................................ 1608
Sarkar, Asit / University of Saskatchewan, Canada........................................................................... 979
Sater Carstens, Deborah / Florida Institute of Technology, USA.................................................... 1769
Sawasdichai, Napawan / King Mongkuts Institute of Technology, Thailand.................................... 816
Schilling, Albert / University of Fortaleza, Brazil............................................................................. 448
Schmitt, Christian / University of Cologne, Germany..................................................................... 2191
Schoder, Detlef / University of Cologne, Germany.......................................................................... 2191
Schwarz, Kathy / University of Saskatchewan, Canada.................................................................. 1656
Selgrade, Katherine A. / Old Dominion University, USA................................................................ 2023
Seok, Soonhwa / The University of Kansas, USA............................................................................. 2542
Serenko, Alexander / Lakehead University, Canada....................................................................... 1069
Serour, Magdy K. / University of Technology, Sydney, Australia.................................................... 1937
Seta, Kazuhisa / Osaka Prefecture University, Japan........................................................................ 950
Setlock, Leslie D. / Carnegie Mellon University, USA..................................................................... 1801
Shah, Abad / R & D Center of Computer Science, Pakistan.............................................................. 659
Sharma, Ramesh C. / Indira Gandhi National Open University, India........................................... 1777
Sharples, Mike / University of Nottingham, Jubilee Campus, UK..................................................... 677
Shirah, Gregory W. / National Aeronautics and Space Administration, USA................................. 2473
Shmon, Cindy L. / University of Saskatchewan, Canada................................................................ 1656
Siau, Keng / University of Nebraska-Lincoln, USA............................................................................ 645
Singhera, Zafar / ZAF Consulting, USA............................................................................................ 659
Sirmakessis, Spiros / Technological Institution of Messolongi, Greece & Research
Academic Computer Technology Institute, Greece.................................................................... 992, 2164
Song, Holim / Texas Southern University, USA................................................................................ 1468
Sorte, Shraddha / Oregon State University, USA................................................................................. 25
Sousa, Kenia / University of Fortaleza, Brazil................................................................................... 448
Spedding, Paul / University of Salford, UK...................................................................................... 2435
Staudinger, Roland / University for Health Sciences, Medical Informatics
and Technology, Austria..................................................................................................................... 1402
Stockard, Russell / California Lutheran University, USA............................................................... 1729
Straub, Tobias / Fraunhofer Institute for Secure Information Technology, Germany....................... 333
Sundaram, David / University of Auckland, New Zealand................................................................ 168

Switzer, Jamie S. / Colorado State University, USA........................................................................ 1837

Tabachneck-Schijf, Hermina / Utrecht University, The Netherlands............................................... 465
Tarsiero, Rosanna / Gionnethics, Italy............................................................................................. 1817
Taylor, Susan M. / University of Saskatchewan, Canada................................................................ 1656
Toprani, Kaushal / Drexel University, USA..................................................................................... 1678
Trauth, Eileen M. / The Pennsylvania State University, USA.......................................................... 2309
Tsakalidis, Athanasios / University of Patras, Greece..................................................................... 1079
Tsounis, Athanasios / University of Athens, Greece......................................................................... 1096
Tupchiy, Anna / University of Central Florida, USA....................................................................... 1885
Tzagarakis, Manolis / Research Academic Computer Technology Institute, Greece...................... 1044
Valiulis, Maryann / University of Dublin, Ireland........................................................................... 2266
Van Dyke, Thomas P. / The University of North Carolina Greensboro, USA................................. 2355
van Nimwegen, Christof / Utrecht University, The Netherlands....................................................... 465
van Oostendorp, Herre / Utrecht University, The Netherlands......................................................... 465
van Riemsdijk, Maarten / University of Twente, The Netherlands................................................. 2214
Vergo, John G. / IBM T. J. Watson Research Center, USA................................................................. 177
Wagner, Daniela / University of Hildesheim, Germany..................................................................... 316
Waldner, Cheryl / University of Saskatchewan, Canada................................................................. 1656
Wallace, J. D. / Lubbock Christian University, USA.......................................................................... 299
Wang, Yingxu / University of Calgary, Canada........................................................................... 33, 685
Wanless-Sobel, Colette / University of Minnesota & Inver Hills Community College, USA.......... 2278
Weber, Ian / Texas A&M University, USA........................................................................................ 1298
Weng, Ming-Hui / National Chiao Tung University, Taiwan............................................................. 516
Whitworth, Brian / Massey University, New Zealand..................................................................... 2675
Wong, Wai-Choong / National University of Singapore, Singapore................................................ 1087
Woodfield, Ruth / University of Sussex, UK.................................................................................... 1583
Wrona, Konrad / SAP Research, France......................................................................................... 2424
Wu, Dezhi / Southern Utah University, USA.................................................................................... 1334
Wu, Ting-Ting / National University of Tainan, Taiwan.................................................................. 1330
Xia, Yun / Rider University, USA...................................................................................................... 1786
Yan, Pan / Lule University of Technology, Sweden............................................................................. 60
Yong, K. L. / Nanyang Technological University, Singapore............................................................. 807
Yoshikawa, Masataka / Hakuhodo Inc., Japan................................................................................ 1223
Yu, Dong / Microsoft Research, USA.................................................................................................. 997
Yuen, Allan Hoi-kau / The University of Hong Kong, China.......................................................... 2287
Zruba, Gergely V. / University of Texas at Arlington, USA........................................................... 1173
Zhang, Daqing / Institute for Infocomm Research, Singapore......................................................... 1087
Zhang, Qiping / Long Island University, USA................................................................................. 1801
Zhang, Zhijun / University of Phoenix, USA..................................................................................... 262
Zheng, Song / Institute for Infocomm Research, Singapore............................................................. 1087

Contents
by Volume

Volume I.

Section I. Fundamental Concepts and Theories

This section serves as the foundation for this exhaustive reference tool by addressing crucial theories essential to the understanding of human-computer interaction. Chapters found within these pages provide
an excellent framework in which to position human-computer interaction within the field of information
science and technology. Individual contributions provide overviews of ubiquitous computing, cognitive
informatics, and sociotechnical theory, while also exploring critical stumbling blocks of this field. Within
this introductory section, the reader can learn and choose from a compendium of expert research on the
elemental theories underscoring the research and application of human-computer interaction.
Chapter 1.1 Introduction to Ubiquitous Computing /
Max Mhlhuser, Technische Universitt Darmstadt, Germany;
and Iryna Gurevych, Technische Universitt Darmstadt, Germany........................................................ 1
Chapter 1.2 Ubiquitous Computing and the Concept of Context /
Antti Oulasvirta, Helsinki Institute for Information Technology, Finland;
and Antti Salovaara, Helsinki Institute for Information Technology, Finland...................................... 20
Chapter 1.3 Gender and End-User Computing / Laura Beckwith, Oregon State University, USA;
Margaret Burnett, Oregon State University, USA;
and Shraddha Sorte, Oregon State University, USA.............................................................................. 25
Chapter 1.4 The Theoretical Framework of Cognitive Informatics /
Yingxu Wang, University of Calgary, Canada....................................................................................... 33
Chapter 1.5 Human Factors in Public Information Systems /
Karl W. Sandberg, Mid Sweden University, Sweden;
and Pan Yan, Lule University of Technology, Sweden......................................................................... 60

Chapter 1.6 Task Analysis at the Heart of Human-Computer Interaction /

Dan Diaper, Middlesex University, UK................................................................................................. 70
Chapter 1.7 Evaluating Mobile Human-Computer Interaction /
Chris Baber, The University of Birmingham, UK.................................................................................. 80
Chapter 1.8 An Overview of Multimodal Interaction Techniques and Applications /
Marie-Luce Bourguet, Queen Mary University of London, UK............................................................ 95
Chapter 1.9 Multimodal and Federated Interaction / Frankie James, SAP Research, USA;
and Rama Gurram, SAP Research, USA............................................................................................. 102
Chapter 1.10 The Arab World, Culture and Information Technology /
Mohamed El Louadi, University of Tunis, Tunisia.............................................................................. 123
Chapter 1.11 Information Technology Acceptance across Cultures /
Amel Ben Zakour, FSJEGJ, Tunisia..................................................................................................... 132
Chapter 1.12 Is Information Ethics Culture-Relative? /
Philip Brey, University of Twente, The Netherlands............................................................................ 154
Chapter 1.13 Personalization Techniques and Their Application /
Juergen Anke, Dresden University of Technology, Germany;
and David Sundaram, University of Auckland, New Zealand............................................................. 168
Chapter 1.14 User-Centered Evaluation of Personalized Web Sites: Whats Unique? /
Sherman R. Alpert, IBM T. J. Watson Research Center, USA;
and John G. Vergo, IBM T. J. Watson Research Center, USA.............................................................. 177
Chapter 1.15 The Usability Engineering Behind User-Centered Processes
for Web Site Development Lifecycles / Theresa A. OConnell, Humans & Computers, Inc., USA;
and Elizabeth D. Murphy, U.S. Census Bureau, USA.......................................................................... 192
Chapter 1.16 Personalization Systems and Their Deployment as Web Site Interface
Design Decisions / Nanda Kumar, Baruch College, The City University of New York, USA.............. 212
Chapter 1.17 Communication + Dynamic Interface = Better User Experience /
Simon Polovina, Sheffield Hallam University, UK; and Will Pearson, Sheffield Hallam
University, UK...................................................................................................................................... 220
Chapter 1.18 Pervasive Computing: What Is It Anyway? /
Emerson Loureiro, Federal University of Campina Grande, Brazil;
Glauber Ferreira, Federal University of Campina Grande, Brazil;
Hyggo Almeida, Federal University of Campina Grande, Brazil;
and Angelo Perkusich, Federal University of Campina Grande, Brazil............................................. 228

Chapter 1.19 Pervasive and Grid Computing Merging /

Loreno Oliveira, Federal University of Campina Grande, Brazil;
Emerson Loureiro, Federal University of Campina Grande, Brazil;
Hyggo Almeida, Federal University of Campina Grande, Brazil;
and Angelo Perkusich, Federal University of Campina Grande, Brazil............................................. 254
Chapter 1.20 Leveraging Pervasive and Ubiquitous Service Computing /
Zhijun Zhang, University of Phoenix, USA.......................................................................................... 262
Chapter 1.21 Service Provision for Pervasive Computing Environments /
Emerson Loureiro, Federal University of Campina Grande, Brazil;
Frederico Bublitz, Federal University of Campina Grande, Brazil;
Loreno Oliveira, Federal University of Campina Grande, Brazil;
Nadia Barbosa, Federal University of Campina Grande, Brazil;
Angelo Perkusich, Federal University of Campina Grande, Brazil;
Hyggo Almeida, Federal University of Campina Grande, Brazil;
and Glauber Ferreira, Federal University of Campina Grande, Brazil............................................. 279
Chapter 1.22 Preparing Participants for Computer Mediated Communication /
Robert Jordan, U.S. Bureau of Labor Statistics, USA......................................................................... 290
Chapter 1.23 Computer-Mediated Communication Research /
J. D. Wallace, Lubbock Christian University, USA............................................................................. 299
Chapter 1.24 Computer-Mediated Communication in Virtual Learning Communities /
Lisa Link, Flensburg University of Applied Sciences, Germany;
and Daniela Wagner, University of Hildesheim, Germany.................................................................. 316
Chapter 1.25 Human-Computer Interaction and Security /
Kai Richter, Computer Graphics Centre (ZGDV), Germany;
and Volker Roth, OGM Laboratory LLC, USA.................................................................................... 323
Chapter 1.26 Security for Ubiquitous Computing /
Tobias Straub, Fraunhofer Institute for Secure Information Technology, Germany;
and Andreas Heinemann, Technical University of Darmstadt, Germany............................................ 333
Section II. Development and Design Methodologies
This section provides in-depth coverage of conceptual architectures, frameworks and methodologies
related to the design and implementation of interfaces and technologies that promote successful human-computer interaction. Throughout these contributions, research fundamentals in the discipline are
presented and discussed. From broad examinations to specific discussions on particular frameworks
and infrastructures, the research found within this section spans the discipline while also offering detailed, specific discussions. Basic designs, as well as abstract developments, are explained within these
chapters, and frameworks for designing and personalizing interfaces for medical equipment, televisions,
and automobiles are discussed.

Chapter 2.1 Measuring the Human Element in Complex Technologies /

Niamh McNamara, University College Cork, Ireland;
and Jurek Kirakowski, University College Cork, Ireland.................................................................... 361
Chapter 2.2 Human-Centric Evolutionary Systems in Design and Decision-Making /
I. C. Parmee, University of the West of England, UK;
J. R. Abraham, University of the West of England, UK;
and A. Machwe, University of the West of England, UK..................................................................... 376
Chapter 2.3 Adaptation and Personalization of User Interface and Content /
Christos K. Georgiadis, University of Macedonia, Thessaloniki, Greece........................................... 393
Chapter 2.4 User Interaction and Interface Design with UML /
Jesus M. Almendros-Jimenez, Universidad de Almeria, Spain;
and Luis Iribarne, Universidad de Almeria, Spain.............................................................................. 404
Chapter 2.5 Design Methods for Experience Design /
Marie Jefsioutine, University of Central EnglandBirmingham, UK;
and John Knight, University of Central EnglandBirmingham, UK................................................... 432
Chapter 2.6 Integrating Usability, Semiotic, and Software Engineering into a Method
for Evaluating User Interfaces /
Kenia Sousa, University of Fortaleza, Brazil;
Albert Schilling, University of Fortaleza, Brazil;
and Elizabeth Furtado, University of Fortaleza, Brazil...................................................................... 448
Chapter 2.7 Guidance in the Interface: Effects of Externalizing Information
During Problem Solving / Christof van Nimwegen, Utrecht University, The Netherlands;
Hermina Tabachneck-Schijf, Utrecht University, The Netherlands;
and Herre van Oostendorp, Utrecht University, The Netherlands...................................................... 465
Chapter 2.8 Ergonomic User Interface Design in Computerized Medical Equipment /
D. John Doyle, Cleveland Clinic Foundation, USA............................................................................ 483
Chapter 2.9 Development Of E-Government Services For Cultural Heritage:
Examining the Key Dimensions / Carugati Andrea, Catholic University of Lille, France;
and Hadzilias Elias, Catholic University of Lille, France................................................................... 489
Chapter 2.10 An Activity-Oriented Approach to Designing a User Interface for Digital Television /
Shang Hwa Hsu, National Chiao Tung University, Taiwan;
Ming-Hui Weng, National Chiao Tung University, Taiwan;
and Cha-Hoang Lee, National Chiao Tung University, Taiwan.......................................................... 516
Chapter 2.11 Designing and Evaluating In-Car User-Interfaces /
Gary Burnett, University of Nottingham, UK...................................................................................... 532

Chapter 2.12 The Semiotics of Smart Appliances and Pervasive Computing /

Peter Bgh Andersen, University of Aarhus, Denmark;
and Martin Brynskov, Center for Interactive Spaces, University of Aarhus, Denmark....................... 552
Chapter 2.13 Analyzing the Factors Influencing the Successful Design
and Uptake of Interactive Systems to Support Social Networks in Urban Neighborhoods /
Marcus Foth, Queensland University of Technology, Australia.......................................................... 589
Chapter 2.14 Different Levels of Information Systems Designers
Forms of Thought and Potential for Human-Centered Design /
Hannakaisa Isomki, University of Lapland, Finland......................................................................... 605
Chapter 2.15 The Think Aloud Method and User Interface Design /
M.W.M. Jaspers, University of Amsterdam, The Netherlands............................................................. 622
Chapter 2.16 Lessons Learned in Designing Ubiquitous Augmented Reality User Interfaces /
Christian Sandor, Technische Universitt Mnchen, Germany;
and Gudrun Klinker, Technische Universitt Mnchen, Germany...................................................... 629
Chapter 2.17 Social Network Structures in Open Source Software Development Teams /
Yuan Long, Colorado State University-Pueblo, USA;
and Keng Siau, University of Nebraska-Lincoln, USA........................................................................ 645

Volume II.
Chapter 2.18 A Graphical User Interface (GUI) Testing Methodology /
Zafar Singhera, ZAF Consulting, USA; Ellis Horowitz, University of Southern California, USA;
and Abad Shah, R & D Center of Computer Science, Pakistan.......................................................... 659
Chapter 2.19 Socio-Cognitive Engineering /
Mike Sharples, University of Nottingham, Jubilee Campus, UK......................................................... 677
Chapter 2.20 On the Cognitive Processes of Human Perception
with Emotions, Motivations, and Attitudes / Yingxu Wang, University of Calgary, Canada.............. 685
Chapter 2.21 Sociotechnical System Design for Learning: Bridging
the Digital Divide with CompILE / Benjamin E. Erlandson, Arizona State University, USA............. 698
Chapter 2.22 Problem Frames for Sociotechnical Systems / Jon G. Hall, The Open University, UK;
and Lucia Rapanotti, The Open University, UK.................................................................................. 713
Chapter 2.23 Integrating Semantic Knowledge with Web Usage Mining for Personalization /
Honghua Dai, DePaul University, USA; and Bamshad Mobasher, DePaul University, USA............ 732

Chapter 2.24 Modeling Variant User Interfaces for Web-Based Software Product Lines /
Suet Chun Lee, BUSINEX, Inc., USA................................................................................................... 760
Chapter 2.25 A User-Centered Approach to the Retrieval of Information in an Adaptive Web Site /
Cristina Gena, Universit di Torino, Italy; and Liliana Ardissono, Universit di Torino, Italy......... 791
Chapter 2.26 Auto-Personalization Web Pages /
Jon T. S. Quah, Nanyang Technological University, Singapore;
Winnie C. H. Leow, Singapore Polytechnic, Singapore;
K. L. Yong, Nanyang Technological University, Singapore................................................................. 807
Chapter 2.27 A Qualitative Study in Users Inforamtion-Seeking Behaviors on Web Sites:
A User-Centered Approach to Web Site Development /
Napawan Sawasdichai, King Mongkuts Institute of Technology, Thailand........................................ 816
Chapter 2.28 Developing and Validating a Measure of Web Personalization Strategy /
Haiyan Fan, Texas A&M University, USA; and Liqiong Deng, University of West Georgia, USA..... 850
Chapter 2.29 User Interface Formalization in Visual Data Mining /
Tiziana Catarci, University of Rome La Sapienza, Italy;
Stephen Kimani, University of Rome La Sapienza, Italy;
and Stefano Lodi, University of Bologna, Italy.................................................................................... 872
Chapter 2.30 Understanding the Nature of Task Analysis in Web Design /
Rod Farmer, The University of Melbourne, Australia;
and Paul Gruba, The University of Melbourne, Australia.................................................................. 899
Chapter 2.31 Designing for Tasks in Ubiquitous Computing: Challenges and Considerations /
Stephen Kimani, Jomo Kenyatta University of Agriculture and Technology, Kenya;
Silvia Gabrielli, University of Rome La Sapienza, Italy;
Tiziana Catarci, University of Rome La Sapienza, Italy;
and Alan Dix, Lancaster University, UK............................................................................................. 928
Chapter 2.32 Task Ontology-Based Human-Computer Interaction /
Kazuhisa Seta, Osaka Prefecture University, Japan............................................................................ 950
Chapter 2.33 Social Networking Theories and Tools to Support Connectivist Learning Activities /
M. C. Pettenati, University of Florence (IT), Italy;
and M. E. Cigognini, University of Florence (IT), Italy...................................................................... 961
Chapter 2.34 User-Centered Design Principles for Online Learning Communities:
A Sociotechnical Approach for the Design of a Distributed Community of Practice /
Ben K. Daniel, University of Saskatchewan, Canada;
David OBrien, University of Saskatchewan, Canada;
and Asit Sarkar, University of Saskatchewan, Canada........................................................................ 979

Chapter 2.35 Content Personalization for Mobile Interfaces /

Spiridoula Koukia, University of Patras, Greece;
Maria Rigou, University of Patras, Greece & Research Academic Computer Technology
Institute, Greece; and Spiros Sirmakessis, Technological Institution of Messolongi, Greece
& Research Academic Computer Technology Institute, Greece.......................................................... 992
Chapter 2.36 Speech-Centric Multimodal User Interface Design in Mobile Technology /
Dong Yu, Microsoft Research, USA; and Li Deng, Microsoft Research, USA.................................... 997
Chapter 2.37 Kinetic User Interfaces: Physical Embodied Interaction
with Mobile Ubiquitous Computing Systems / Vincenzo Pallotta, University of Fribourg, Switzerland;
Pascal Bruegger, University of Fribourg, Switzerland;
and Bat Hirsbrunner, University of Fribourg, Switzerland............................................................. 1015

Section III. Tools and Technologies

This section presents extensive coverage of specific tools and technologies that humans interact with
and react to in their daily lives. These chapters provide an in-depth analysis of devices and tools such
as portable music players, mobile phones, and even blogs. Within these rigorously researched chapters,
readers are presented with countless examples of the technologies that support and encourage societal
development and their resulting impact.
Chapter 3.1 Enabling Technologies for Pervasive Computing /
Joo Henrique Kleinschmidt, State University of Campinas, Brazil;
and Walter da Cunha Borelli, State University of Campinas, Brazil................................................. 1037
Chapter 3.2 Augmenting Collaboration with Personalization Services /
Christina E. Evangelou, Research Academic Computer Technology Institute, Greece;
Manolis Tzagarakis, Research Academic Computer Technology Institute, Greece;
Nikos Karousos, Research Academic Computer Technology Institute, Greece;
George Gkotsis, Research Academic Computer Technology Institute, Greece;
and Dora Nousia, Research Academic Computer Technology Institute, Greece.............................. 1044
Chapter 3.3 Culture and Technology: A Mutual-Shaping Approach /
Thomas Herdin, University of Salzburg, Austria;
Wolfgang Hofkirchner, University of Salzburg, Austria;
and Ursula Maier-Rabler, University of Salzburg, Austria............................................................... 1055
Chapter 3.4 Importance of Interface Agent Characteristics from End-User Perspective /
Alexander Serenko, Lakehead University, Canada............................................................................ 1069
Chapter 3.5 Web Mining for Public E-Services Personalization /
Penelope Markellou, University of Patras, Greece;
Angeliki Panayiotaki, University of Patras, Greece;
and Athanasios Tsakalidis, University of Patras, Greece.................................................................. 1079

Chapter 3.6 Context-Aware Service Discovery in Ubiquitous Computing /

Huaqun Guo, Institute for Infocomm Research and National Univerity of Singapore, Singapore;
Daqing Zhang, Institute for Infocomm Research, Singapore;
Lek-Heng Ngoh, Institute for Infocomm Research, A*STAR, Singapore;
Song Zheng, Institute for Infocomm Research, Singapore;
and Wai-Choong Wong, National University of Singapore, Singapore............................................. 1087
Chapter 3.7 Ontological Engineering in Pervasive Computing Environments /
Athanasios Tsounis, University of Athens, Greece;
Christos Anagnostopoulos, University of Athens, Greece;
Stathes Hadjiethymiades, University of Athens, Greece;
and Izambo Karali, University of Athens, Greece............................................................................. 1096
Chapter 3.8 MASACAD: A Multi-Agent System for Academic Advising /
Mohamed Salah Hamdi, University of Qatar, Qatar......................................................................... 1118
Chapter 3.9 Human Factors Problems of Wearable Computers /
Chris Baber, The University of Birmingham, UK;
and James Knight, The University of Birmingham, UK.................................................................... 1134
Chapter 3.10 Modelling Interactive Behaviour with a Rational Cognitive Architecture /
David Peebles, University of Huddersfield, UK;
and Anna L. Cox, University College London, UK............................................................................ 1154
Chapter 3.11 Device Localization in Ubiquitous Computing Environments /
Rui Huang, University of Texas at Arlington, USA;
Gergely V. Zruba, University of Texas at Arlington, USA;
and Sajal Das, University of Texas at Arlington, USA...................................................................... 1173
Chapter 3.12 Socio-Cultural Interpretations to the Diffusion and use of Broadband Services in a
Korean Digital Society / Dal Yong Jin, Simon Fraser University, Canada....................................... 1200
Chapter 3.13 Mobility and Multimodal User Interfaces /
Christopher J. Pavlovski, IBM Corporation, Australia;
and Stella Mitchell, IBM T. J. Watson Research, USA....................................................................... 1213
Chapter 3.14 Foreseeing the Future Lifestyle with Digital Music: A Comparative
Study Between Mobile Phone Ring Tones and Hard-Disk Music Players Like iPod /
Masataka Yoshikawa, Hakuhodo Inc., Japan.................................................................................... 1223
Chapter 3.15 Widely Usable User Interfaces on Mobile Devices with RFID /
Francesco Bellotti, University of Genoa, Italy;
Riccardo Berta, University of Genoa, Italy;
Alessandro De Gloria, University of Genoa, Italy;
and Massimiliano Margarone, University of Genoa, Italy................................................................ 1234

Chapter 3.16 Plagiarism, Instruction, and Blogs / Michael Hanrahan, Bates College, USA............ 1251
Chapter 3.17 Twin Wiki Wonders? Wikipedia and Wikibooks as Powerful Tools
for Online Collaborative Writing / Meng-Fen Grace Lin, University of Houston, USA;
Curtis J. Bonk, Indiana University, USA; and Suthiporn Sajjapanroj, Indiana University, USA..... 1262
Chapter 3.18 Wikis as Tools for Collaboration / Jane Klobas, Bocconi University, Italy &
University of Western Australia, Australia........................................................................................ 1283
Chapter 3.19 Academic Weblogs as Tools for E-Collaboration Among Researchers /
Mara Jos Luzn, University of Zaragoza, Spain............................................................................ 1291
Chapter 3.20 Assessing Weblogs as Education Portals / Ian Weber, Texas A&M University,
USA.................................................................................................................................................... 1298
Chapter 3.21 Innovative Technologies for Education and Learning: Education
and Knowledge-Oriented Applications of Blogs, Wikis, Podcasts, and More /
Jeffrey Hsu, Fairleigh Dickinson University, USA............................................................................ 1308
Chapter 3.22 Ubiquitous Computing Technologies in Education /
Gwo-Jen Hwang, National University of Tainan, Taiwan;
Ting-Ting Wu, National University of Tainan, Taiwan;
and Yen-Jung Chen, National University of Tainan, Taiwan............................................................. 1330
Chapter 3.23 Utilizing Web Tools for Computer-Mediated Communication to Enhance
Team-Based Learning / Elizabeth Avery Gomez, New Jersey Institute of Technology, USA;
Dezhi Wu, Southern Utah University, USA; Katia Passerini, New Jersey Institute of Technology,
USA; and Michael Bieber, New Jersey Institute of Technology, USA................................................ 1334

Volume III.
Chapter 3.24 How Technology Can Support Culture and Learning /
David Luigi Fuschi, GIUNTI Labs S.r.l., Italy;
Bee Ong, University of Leeds, UK;
and David Crombie, DEDICON S.r.l., The Netherlands................................................................... 1350
Chapter 3.25 Tangible User Interfaces as Mediating Tools within
Adaptive Educational Environments / Daria Loi, RMIT University, Australia................................. 1388
Chapter 3.26 Facilitating E-Learning with Social Software: Attitudes and Usage from
the Students Point of View / Reinhard Bernsteiner, University for Health Sciences,
Medical Informatics and Technology, Austria; Herwig Ostermann, University for
Health Sciences, Medical Informatics and Technology, Austria; and Roland Staudinger,
University for Health Sciences, Medical Informatics and Technology, Austria.....................................1402

Section IV. Utilization and Application

This section introduces and discusses the ways in which particular technologies have impacted humanity
and proposes new ways in which IT-related innovations can be implemented within organizations and
in society as a whole. These particular selections highlight, among other topics, ubiquitous computing
applications in an educational setting and computer-mediated communication. Contributions included
in this section provide excellent coverage of todays world and insight into how our interaction with
technology impacts the fabric of our present-day global village.
Chapter 4.1 Human Computer Interaction and the Best Mix of Face-to-Face
and E-Interactions in Educational Settings / Bolanle A. Olaniran, Texas Tech University, USA...... 1423
Chapter 4.2 Human Computer Interaction for Computer-Based Classroom Teaching /
Wolfgang Hrst, Albert-Ludwigs-Universitt Freiburg, Germany;
and Khaireel A. Mohamed, Albert-Ludwigs-Universitt Freiburg, Germany................................... 1436
Chapter 4.3 Wikis as an Exemplary Model of Open Source Learning /
Robert Fitzgerald, University of Canberra, Australia....................................................................... 1459
Chapter 4.4 Re-Examining the Socioeconomic Factors Affecting Technology Use
in Mathematics Classroom Practices / Emiel Owens, Texas Southern University, USA;
Holim Song, Texas Southern University, USA;
and Terry T. Kidd, University of Texas School of Public Health, USA.............................................. 1468
Chapter 4.5 Let Them Blog: Using Weblogs to Advance Literacy in the K-12 Classroom /
David A. Huffaker, Northwestern University, USA............................................................................ 1484
Chapter 4.6 Ubiquitous Computing Applications in Education /
Kostas Kolomvatsos, National & Kapodistrian University of Athens, Greece.................................. 1503
Chapter 4.7 Unraveling the Taste Fabric of Social Networks /
Hugo Liu, The Media Laboratory, USA; Pattie Maes, The Media Laboratory, USA;
and Glorianna Davenport, The Media Laboratory, USA.................................................................. 1521
Chapter 4.8 Computer-Mediated Communication: Enhancing Online Group Interactions /
J. Michael Blocher, Northern Arizona University, USA.................................................................... 1547
Chapter 4.9 Mobile Phone Use Across Cultures: A Comparison Between
the United Kingdom and Sudan / Ishraga Khattab, Brunel University, UK;
and Steve Love, Brunel University, UK............................................................................................. 1569
Chapter 4.10 Gender and Computing at University in the UK /
Ruth Woodfield, University of Sussex, UK......................................................................................... 1583
Chapter 4.11 Gender and Information Technology in Rural Bangladesh /
Lutfor Rahman, Association for Advancement of Information Technology (AAIT), Bangladesh;
and Nusrat Rahman, University of Dhak and Association for Advancement of Information
Technology (AAIT), Bangladesh........................................................................................................ 1591

Chapter 4.12 Chinese POS Disambiguation and Unknown Word Guessing

with Lexicalized HMMs / Guohong Fu, The University of Hong Kong, Hong Kong;
and Kang-Kwong Luke, The University of Hong Kong, Hong Kong................................................. 1595
Chapter 4.13 IT Implementation in a Developing Country Municipality:
A Sociocognitive Analysis / Clive Sanford, Aalborg University, Denmark;
and Anol Bhattacherjee, University of South Florida, USA.............................................................. 1608
Chapter 4.14 Personalization Issues for Science Museum Web Sites and E-Learning /
Silvia Filippini-Fantoni, The University of Paris I Sorbonne University, France;
Jonathan P. Bowen, London South Bank University, UK;
Teresa Numerico, London South Bank University, UK...................................................................... 1629
Chapter 4.15 Personalization and Customer Satisfaction in Mobile Commerce /
HY Sonya Hsu, Southern Illinois University, USA; and
Songpol Kulviwat, Hofstra University, USA...................................................................................... 1649
Chapter 4.16 Integrating Human Computer Interaction in Veterinary Medicine Curricula /
Gale Parchoma, University of Saskatchewan, Canada;
Susan M. Taylor, University of Saskatchewan, Canada;
Jonathan M. Naylor, Ross University School of Veterinary Medicine, West Indies;
Sameeh M. Abutarbush, University of Prince Edward Island, Canada;
Katharina L. Lohmann, University of Saskatchewan, Canada;
Kathy Schwarz, University of Saskatchewan, Canada;
Cheryl Waldner, University of Saskatchewan, Canada;
Sharon Porterfield, University of Saskatchewan, Canada;
Cindy L. Shmon, University of Saskatchewan, Canada;
Lydden Polley, University of Saskatchewan, Canada;
Chris Clark, University of Saskatchewan Canada............................................................................ 1656
Chapter 4.17 Portals in Consumer Search Behavior and Product Customization /
Ian Michael, Zayed University, UAE................................................................................................. 1673
Chapter 4.18 Human Factors in the Development of Trend Detection and Tracking Techniques /
Chaomei Chen, Drexel University, USA; Kaushal Toprani, Drexel University, USA;
and Natasha Lobo, Drexel University, USA...................................................................................... 1678
Chapter 4.19 The Internet, Health Information, and Managing Health: An Examination
of Boomers and Seniors / Christopher G. Reddick, The University of Texas at San Antonio, USA.. 1687
Chapter 4.20 The Online Effect: Transitioning from the Legacy Help Desk to the Online
Task Management System / Kym Mawson-Lee, University of New South Wales, Australia............. 1707
Section V. Organizational and Social Implications
This section includes a wide range of research pertaining to the social and organizational impact of
human-computer interaction around the world. Chapters introducing this section analyze gender dif-

ferences in the adoption and reaction to IT, while later contributions offer an extensive analysis of differing social and cultural dimensions of technology adoption. The inquiries and methods presented in
this section offer insight into the implications of human-computer interaction at both a personal and
organizational level, while also emphasizing potential areas of study within the discipline.
Chapter 5.1 Gender, Race, Social Class and Information Technology /
Myungsook Klassen, California Lutheran University, USA;
and Russell Stockard, California Lutheran University, USA............................................................. 1729
Chapter 5.2 Gender and the Culture of Computing in Applied IT Education /
Susan C. Herring, Indiana University, USA; Christine Ogan, Indiana University, USA;
Manju Ahuja, Indiana University, USA; and Jean C. Robinson, Indiana University, USA.............. 1736
Chapter 5.3 Gender Equalization in Computer-Mediated Communication /
Rosalie J. Ocker, The Pennsylvania State University, USA............................................................... 1745
Chapter 5.4 The Cross-Cultural Dimension of Gender and Information Technology /
Haiyan Huang, The Pennsylvania State University, USA.................................................................. 1753
Chapter 5.5 Enhancing E-Collaboration Through Culturally Appropriate User Interfaces /
Dianne Cyr, Simon Fraser University, Canada................................................................................. 1761
Chapter 5.6 Cultural Barriers of Human-Computer Interaction /
Deborah Sater Carstens, Florida Institute of Technology, USA........................................................ 1769
Chapter 5.7 Technology and Culture: Indian Experiences /
Ramesh C. Sharma, Indira Gandhi National Open University, India;
and Sanjaya Mishra, Indira Gandhi National Open University, India.............................................. 1777
Chapter 5.8 Intercultural Computer-Mediated Communication Between
Chinese and U.S. College Students / Yun Xia, Rider University, USA.............................................. 1786
Chapter 5.9 Global Culture and Computer Mediated Communication /
Susan R. Fussell, Carnegie Mellon University, USA; Qiping Zhang, Long Island University, USA;
and Leslie D. Setlock, Carnegie Mellon University, USA.................................................................. 1801
Chapter 5.10 Linguistics of Computer-Mediated Communication:
Approaching the Metaphor / Rosanna Tarsiero, Gionnethics, Italy.................................................. 1817
Chapter 5.11 Impression Formation in Computer-Mediated Communication
and Making a Good (Virtual) Impression / Jamie S. Switzer, Colorado State University, USA........ 1837
Chapter 5.12 Group Decision Making in Computer-Mediated Communication
as Networked Communication: Understanding the Technology and Implications /
Bolanle A. Olaniran, Texas Tech University, USA............................................................................. 1849
Chapter 5.13 Effects of Computer-Mediated Communication /
Stuart S. Gold, DeVry University, USA.............................................................................................. 1864

Chapter 5.14 Exploring Events as an Information Systems Research Methodology /

Anita Greenhill, The University of Manchester, UK;
and Gordon Fletcher, University of Salford, UK............................................................................... 1871
Chapter 5.15 Cultures Impact on Technology Mediated Learning:
The Role of Horizontal and Vertical Individualism and Collectivism /
Steven Hornik, University of Central Florida, USA;
and Anna Tupchiy, University of Central Florida, USA.................................................................... 1885
Chapter 5.16 Using Social Networking Analysis to Facilitate Knowledge Sharing in
the British Council / Bonnie Wai-yi Cheuk, Improvement Service for
the Scottish Local Government, Scotland.......................................................................................... 1905
Chapter 5.17 Collaboration Challenges in Community Telecommunication Networks /
Sylvie Albert, Laurentian University, Canada; Rolland LeBrasseur, Laurentian University,
Canada......................................................................................................................................................1915
Chapter 5.18 Planning and Managing the Human Factors for the Adoption and Diffusion
of Object-Oriented Software Development Processes /
Magdy K. Serour, University of Technology, Sydney, Australia......................................................... 1937
Chapter 5.19 Product Customization on the Web: An Empirical Study of Factors
Impacting Choiceboard User Satisfaction / Pratyush Bharati, University of Massachusetts,
Boston, USA; and Abhijit Chaudhury, Bryant University, USA........................................................ 1956
Chapter 5.20 Distributed Deception: An Investigation of the Effectiveness
of Deceptive Communication in a Computer-Mediated Environment /
Randall J. Boyle, University of Utah, USA; Charles J. Kacmar, The University of Alabama, USA;
and Joey F. George, Florida State University, USA.......................................................................... 1970
Chapter 5.21 Extrinsic Plus Intrinsic Human Factors Influencing the Web Usage /
Manuel Jess Snchez-Franco, University of Seville, Spain............................................................. 1995
Section VI. Managerial Impact
This section presents contemporary coverage of the managerial implications of human-computer interaction. Particular contributions address virtual networking, knowledge blogs in company environments,
and ambient business. The managerial research provided in this section allows executives, practitioners, and researchers to gain a better sense of the relationship between of business, technology, and
individuals.
Chapter 6.1 Social Impact of Virtual Networking /
Hakikur Rahman, SDNP Bangladesh, Bangladesh........................................................................... 2014
Chapter 6.2 Gender and Telework in Information Technology / Paula F. Saddler, Association of
Certified Fraud Examiners, USA; Donald D. Davis, Old Dominion University, USA;

Katherine A. Selgrade, Old Dominion University, USA;

Debra A. Major, Old Dominion University, USA............................................................................... 2023
Chapter 6.3 Managerial Careers, Gender, and Information Technology Field /
Iiris Aaltio, Lappeenranta University of Technology, Finland.......................................................... 2030

Volume IV.
Chapter 6.4 Social Networking and Knowledge Transfer in Collaborative Product Development /
Katariina Ala-Rmi, University of Oulu, Finland............................................................................. 2037
Chapter 6.5 Reframing Information System Design as Learning Across Communities of Practice /
Kevin Gallagher, Northern Kentucky University, USA;
and Robert M. Mason, University of Washington, USA..................................................................... 2052
Chapter 6.6 Grounding Business Interaction Models: Socio-Instrumental Pragmatism
as a Theoretical Foundation / Gran Goldkuhl, Linkping University
& Jnkping International Business School, Sweden; and Mikael Lind, University
College of Bors, Sweden, & Linkping University, Sweden............................................................ 2071
Chapter 6.7 Managing Socio-Technical Integration in Iterative Information System
Development Projects / Bendik Bygstad, Norwegian School of Information Technology, Norway.. 2090
Chapter 6.8 Human Factors for Networked and Virtual Organizations /
Vincent E. Lasnik, Independent Knowledge Architect, USA.............................................................. 2106
Chapter 6.9 Trust in Computer-Mediated Communications: Implications
for Individuals and Organizations / Susan K. Lippert, Drexel University, USA................................ 2118
Chapter 6.10 Understanding an ERP System Implementation in a Higher
Education Institution: A Grounded Theory Approach / Jose Esteves, Universitat Politecnica
de Catalunga, Barcelona, Spain; and Joan Pastor, Universitat Internacional de Catalunga,
Barcelona, Spain................................................................................................................................ 2132
Chapter 6.11 Ontologies for Scalable Services-Based Ubiquitous Computing /
Daniel Oberle, SAP Research, CEC Karlsruhe, Germany;
Christof Bornhvd, SAP Research, Research Center Palo Alto, USA;
and Michael Altenhofen, SAP Research, CEC Karlsruhe, Germany................................................. 2144
Chapter 6.12 Web Personalization for E-Marketing Intelligence /
Penelope Markellou, University of Patras, Greece & Research Academic Computer Technology
Institute, Greece; Maria Rigou, University of Patras, Greece & Research Academic Computer
Technology Institute, Greece; and Spiros Sirmakessis, Technological Educational Institution of
Messolongi, Greece& Research Academic Computer Technology Institute, Greece........................ 2164
Chapter 6.13 Knowledge Blogs in Firm Internal Use /
Miia Kosonen, Lappeenranta University of Technology, Finland;

Kaisa Henttonen, Lappeenranta University of Technology, Finland;

and Kirsimarja Blomqvist, Lappeenranta University of Technology, Finland.................................. 2181
Chapter 6.14 Towards Ambient Business: Enabling Open Innovation
in a World of Ubiquitous Computing / Christian Schmitt, University of Cologne, Germany;
Detlef Schoder, University of Cologne, Germany; Kai Fischbach, University of Cologne,
Germany; and Steffen Muhle, University of Cologne, Germany....................................................... 2191
Chapter 6.15 Succeses and Failures of SAP Inplementation: A Learning Perspective /
Tanya Bondarouk, University of Twente, The Netherlands; and Maarten van Riemsdijk,
University of Twente, The Netherlands.............................................................................................. 2214
Chapter 6.16 The Human Face of E-Business: Engendering Consumer Initial
Trust Through the Use of Images of Sales Personnel on E-Commerce Web Sites /
Khalid Aldiri, University of Bradford, UK; Dave Hobbs, University of Bradford, UK;
Rami Qahwaji, University of Bradford, UK...................................................................................... 2235
Section VII. Critical Issues
This section addresses conceptual and theoretical issues related to the field of human-computer interaction.
Within these chapters, the reader is presented with analysis of the most current and relevant conceptual
inquires within this growing field of study. Particular chapters address particular algorithms for web
personalization, sociotechnical change, and security for ubiquitous computing systems. Overall, contributions within this section ask unique, often theoretical questions related to the study of human-computer
interaction and, more often than not, conclude that solutions are both numerous and contradictory.
Chapter 7.1 Gender, Place, and Information Technology / Anita Greenhill,
The University of Manchester, UK..................................................................................................... 2259
Chapter 7.2 The Intersection of Gender, Information Technology, and Art /
Linda Doyle, University of Dublin, Ireland;
and Maryann Valiulis, University of Dublin, Ireland........................................................................ 2266
Chapter 7.3 Gender-Based Attitudes Toward Technology /
Konrad Morgan, University of Bergen, Norway;
and Madeleine Morgan, University of Bergen, Norway.................................................................... 2274
Chapter 7.4 Gender in Distance Education Technology /
Colette Wanless-Sobel, University of Minnesota & Inver Hills Community College, USA............... 2278
Chapter 7.5 Gender Differences in Information Technology Acceptance /
Will Wai-kit Ma, Hong Kong University of Science and Technology, China and
Hong Kong Shue Yan College, China; and Allan Hio-kau Yuen, The University of
Kong Kong, China............................................................................................................................. 2287
Chapter 7.6 Discourses in Gender and Technology: Taking a Feminist Gaze /
Sheila French, Manchester Metropolitan University, UK................................................................. 2296

Chapter 7.7 Theorizing Gender and Information Technology Research /

Eileen M. Trauth, The Pennsylvania State University, USA.............................................................. 2309
Chapter 7.8 Socio-Cognitive Model of Trust /
Rino Falcone, Institute for Cognitive Sciences and Technology, National Research Council
of Italy, Italy; and Cristiano Castelfranchi, Institute for Cognitive Sciences and Technology
National Research Council of Italy, Italy........................................................................................... 2316
Chapter 7.9 Information-Communications Systems Convergence Paradigm:
Invisible E-Culture and E-Technologies / Fjodor Ruzic, Institute for Informatics, Croatia............. 2324
Chapter 7.10 IT Artefacts as Socio-Pragmatic Instruments: Reconciling the Pragmatic,
Semiotic, and Technical / Gran Goldkuhl, Linkping University &
Jnkping International Business School, Sweden;
and Pr J. gerfalk, University of Limerick, Ireland & rebro University, Sweden........................ 2341
Chapter 7.11 Ignorance is Bliss: The Effect of Increased Knowledge
on Privacy Concerns and Internet Shopping Site Personalization Preferences /
Thomas P. Van Dyke, The University of North Carolina Greensboro, USA...................................... 2355
Chapter 7.12 The Function of Representation in a Smart Home Context /
Mats Edenius, Stockholm School of Economics, Sweden.................................................................. 2376
Chapter 7.13 Towards a Meta-Model for Socio-Instrumental Pragmatism /
Peter Rittgen, University College of Bors, Sweden......................................................................... 2390
Chapter 7.14 The Effect of Choice and Announcement Duration on
the Estimation of Telephone Hold Time / Philip Kortum, Rice University, USA;
Randolph G. Bias, The University of Texas at Austin, USA;
Benjamin A. Knott, U.S. Air Force Research Laboratory, USA;
and Robert G. Bushey, AT&T Laboratories, USA............................................................................. 2404
Chapter 7.15 Towards Adaptive Security for Ubiquitous Computing Systems:
MOSQUITO and Serenity / Volkmar Lotz, SAP Research, France;
Luca Compagna, SAP Research, France; and Konrad Wrona, SAP Research, France.................... 2424
Chapter 7.16 Trusting Computers Through Trusting Humans: Software Verification
in a Saftey-Critical Information System / Alison Adam, University of Salford, UK;
and Paul Spedding, University of Salford, UK.................................................................................. 2435
Chapter 7.17 Global Information Ethics: The Importance of Being Environmentally Earnest /
Luciano Floridi, Universita degli Studi di Bari, Italy and Oxford University, UK........................... 2450

Chapter 7.18 Emotional Digitalization as Technology of the Postmodern:

A Reflexive Examination from the View of the Industry /
Claus Hohmann, Bergische Universitaet Wuppertal, Germany........................................................ 2462
Chapter 7.19 The Paleolithic Stone Age Effect? Gender Differences Performing
Specific Computer-Generated Spatial Tasks /
Geoffrey S. Hubona, Georgia State University, USA;
and Gregory W. Shirah, National Aeronautics and Space Administration, USA............................... 2473
Chapter 7.20 Optimality-Theoretic Lexical Mapping Theory: A Case Study of
Locative Inversion / One-Soon Her, National Chengchi University, Taiwan.................................... 2499
Section VIII. Emerging Trends
This section highlights research potential within the field of human-computer interaction while exploring
uncharted areas of study for the advancement of the discipline. Chapters within this section highlight
the socio-semantic web, podcasts, and the emergence of increasingly intelligent user interfaces. These
contributions, which conclude this exhaustive, multi-volume set, provide emerging trends and suggestions for future research within this rapidly expanding discipline.
Chapter 8.1 Podcastia: Imagining Communities of Pod-People / Jonathan Cohn, UCLA, USA...... 2530
Chapter 8.2 Maximizing Web Accessibility Through User-Centered Interface Design /
Soonhwa Seok, The University of Kansas, USA................................................................................ 2542
Chapter 8.3 Voice-Enabled User Interfaces for Mobile Devices /
Louise E. Moser, University of California, Santa Barbara, USA;
and P. M. Melliar-Smith, University of California, Santa Barbara, USA......................................... 2556
Chapter 8.4 Multilayered Approach to Evaluate Mobile User Interfaces /
Maria de Ftima Queiroz Vieira Turnell, Universidade Federal de Campina Grande (UFCG), Brazil;
Jos Eustquio Rangel de Queiroz, Universidade Federal de Campina Grande (UFCG), Brazil;
Danilo de Sousa Ferreira, Universidade Federal de Campina Grande (UFCG), Brazil................. 2573
Chapter 8.5 Digital Imaging Trek: A Practical Model for Managing
the Demand of the Digitally Enabled Traveller / Stephen C. Andrade, Johnson
& Wales University, USA; and Hilary Mason, Johnson & Wales University, USA........................... 2590
Chapter 8.6 Intelligent User Interfaces for Mobile Computing /
Michael J. OGrady, University College Dublin, Ireland;
and Gregory M. P. OHare, University College Dublin, Ireland....................................................... 2614
Chapter 8.7 Intelligent User Interfaces for Ubiquitous Computing /
Rainer Malaka, Bremen University, Bremen..................................................................................... 2627

Chapter 8.8 Vive la Diffrence: The Cross-Culture Differences Within Us /

David Gefen, Drexel University, USA; Nitza Geri, The Open University of Israel, Israel;
and Narasimha Paravastu, Central State University, USA............................................................... 2644
Chapter 8.9 Growing Up Wireless: Being a Parent and Being a Child in
the Age of Mobile Communication / Letizia Caronia, University of Bologna, Italy......................... 2655
Chapter 8.10 Politeness as a Social Computing Requirement /
Brian Whitworth, Massey University, New Zealand;
and Tong Liu, Massey University, New Zealand................................................................................ 2675
Chapter 8.11 Neo-Symbiosis: The Next Stage in the Evolution
of Human Information Interaction /
Douglas Griffith, General Dynamics Advanced Information Systems, USA;
and Frank L. Greitzer, Pacific Northwest National Laboratory, USA............................................... 2693
Chapter 8.12 Enabling Programmable Ubiquitous Computing Environments:
A Middleware Perspective / Christine Julien, The University of Texas at Austin, USA;
and Sanem Kabadayi, The University of Texas at Austin, USA......................................................... 2708

xxxi

Preface

The use of, engagement with, and study of technology has become the basis of modern life. As such, the
field of human-computer interaction has emerged as an essential, multidisciplinary field that seeks to
determine how humans interact with different interfaces and how these interfaces can be better designed,
evaluated, and implemented to minimize barriers between human thought and computer behavior. As
we continue to develop new and innovative technologies, we must also strive to more fully understand
how these technologies impact humanity, both on a societal and individual level, and to remain aware
of the latest in human-computer interaction research and exploration.
In recent years, the areas of study related to the interaction between people and technology have
become innumerable. As a result, researchers, practitioners, and educators have devised a variety of
techniques and methodologies to develop, deliver, and, at the same time, evaluate the effectiveness of
the interfaces implemented and used in modern society. The explosion of methodologies in the field has
created an abundance of new, state-of-the-art literature related to all aspects of this expanding discipline.
This body of work allows researchers to learn about the fundamental theories, latest discoveries, and
forthcoming trends in the field of human-computer interaction.
Constant technological and theoretical innovation challenges researchers to remain informed of and
continue to develop and deliver methodologies and techniques utilizing the disciplines latest advancements. In order to provide the most comprehensive, in-depth, and current coverage of all related topics
and their applications, as well as to offer a single reference source on all conceptual, methodological,
technical, and managerial issues in human-computer interaction, Information Science Reference is pleased
to offer a four-volume reference collection on this rapidly growing discipline. This collection aims to
empower researchers, practitioners, and students by facilitating their comprehensive understanding of
the most critical areas within this field of study.
This collection, entitled Human Computer Interaction: Concepts, Methodologies, Tools, and
Applications, is organized into eight distinct sections which are as follows: 1) Fundamental Concepts
and Theories, 2) Development and Design Methodologies, 3) Tools and Technologies, 4) Utilization and
Application, 5) Organizational and Social Implications, 6) Managerial Impact, 7) Critical Issues, and 8)
Emerging Trends. The following paragraphs provide a summary of what is covered in each section of
this multi-volume reference collection.
Section One, Fundamental Concepts and Theories, serves as a foundation for this exhaustive
reference tool by addressing crucial theories essential to understanding human-computer interaction.
Specific issues in human-computer interaction, such as ubiquitous computing, communities of practice,
and online social networking are discussed in selections such as Ubiquitous Computing and the Concept
of Context by Antti Oulasvirta and Antti Salovaara, Sociotechnical Theory and Communities of Practice by Andrew Wenn, and Online Social Networking for New Researched Opportunities by Lionel
Mew. Within the selection Personalization Techniques and Their Application, authors Juergen Anke

xxxii

and David Sundaram provide an overview of the personalization of information systems and the impact
such an approach has on the usability of everything from e-learning environments to mobile devices. The
selections within this comprehensive, foundational section allow readers to learn from expert research
on the elemental theories underscoring human-computer interaction.
Section Two, Development and Design Methodologies, contains in-depth coverage of conceptual
architectures and frameworks, providing the reader with a comprehensive understanding of emerging
theoretical and conceptual developments within the development and utilization of tools and environments that promote interaction between people and technology. Beginning this section is the contribution
Measuring the Human Element in Complex Technologies by Niamh McNamara and Jurek Kirakowski,
which analyzes the role that user satisfaction has in shaping the design of products and the impact this
has on developers. Similarly, Design Methods for Experience Design by Marie Jefsioutine and John
Knight describes the authors framework for web site designthe Experience Design Framework.
Other selections, such as Content Personalization for Mobile Interfaces by Spiridoula Koukia, Maria
Rigou, and Spiros Sirmakessis and Kinetic User Interfaces: Physical Embodied Interaction with Mobile
Ubiquitous Computing Systems by Vincenzo Pallotta, Pascal Bruegger, and Bat Hirsbrunner offer
insight into particular methodologies for the design, development, and personalization of mobile user
interfaces. From basic designs to abstract development, chapters such as Designing and Evaluating InCar User-Interfaces by Gary Burnett and Integrating Usability, Semiotic, and Software Engineering
into a Method for Evaluating User Interfaces by Kenia Sousa, Albert Schilling, and Elizabeth Furtado
serve to expand the reaches of development and design techniques within the field of human-computer
interaction.
Section Three, Tools and Technologies, presents extensive coverage of various tools and technologies
that individuals interact, collaborate, and engage with every day. The emergence of social networking
tools such as blogs and wikis is explored at length in selections such as PDA Plagiarism, Instruction,
and Blogs by Michael Hanrahan, Wikis as Tools for Collaboration by Jane Klobas, and Assessing
Weblogs as Education Portals by Ian Weber. Further discussions of the role of technology in learning
are explored in chapters such as How Technology Can Support Culture and Learning by David Luigi
Fuschi, Bee Ong, and David Crombie and Facilitating E-Learning with Social Software: Attitudes and
Usage from the Students Point of View by Reinhard Bernsteiner, Herwig Ostermann, and Roland
Staudinger. The latter of these two chapters discusses both the theoretical basis for the implementation
of technology in an educational setting and the impact this implementation has on students. Ultimately,
the authors conclude that online social networking has the potential to emerge as a useful tool for both
students and teachers. The rigorously researched chapters contained within this section offer readers
countless examples of the dynamic interaction between humans and technology.
Section Four, Utilization and Application, provides an in-depth analysis of the practical side of
human computer interaction, focusing specifically on environments in which the relationship between
humans and technology has been significant. Mobile device usage is highlighted in the selections A
De-Construction of Wireless Device Usage by Mary R. Lind, Localized User Interface for Improving
Cell phone Users Device Competency by Lucia D. Krisnawati and Restyandito, and Mobile Phone
Use Across Cultures: A Comparison Between the United Kingdom and Sudan by Ishraga Khattab and
Steve Love. These selections offer conceptualization and analysis of the factors impacting wireless device
usage, ultimately determining that many factors, such as culture, familiarity with the device itself, and
ease of use influence an individuals wireless device usage. Further contributions explore how human
factors impact the successful use of technology in areas such as trend detection, veterinary medicine,
mobile commerce, and web browsing. From established applications to forthcoming innovations,
contributions in this section provide excellent coverage of todays global community and demonstrate

xxxiii

how the interaction between humans and technology impacts the social, economic, and political fabric
of our present-day global village.
Section Five, Organizational and Social Implications, includes a wide range of research pertaining
to the organizational and cultural implications of humans reaction to technology. This section begins
with a thorough analysis of the intersection of gender and technology in contributions such as Gender,
Race, Social Class, and Information Technology by Myungsook Klassen and Russell Stockard, Jr.,
Gender and the Culture of Computing in Applied IT Education by Susan C. Herring, Christine Ogan,
Manju Ahuja, and Jean C. Robinson, and Gender Equalization in Computer-Mediated Communication
by Rosalie J. Ocker. Other issues that are surveyed within this section include the implication of cultural
differences within Deborah Sater Carstens Cultural Barriers of Human-Computer Interaction, computer
mediated communication in Bolanle A. Olanirans Group Decision Making in Computer-Mediated
Communication as Networked Communication: Understanding the Technology and Implications, and
community telecommunication networks in Sylvie Albert and Rolland LeBrasseurs Collaboration
Challenges in Community Telecommunication Networks. Overall, the discussions presented in this
section offer insight into the implications of human computer interaction in both organizational and
social settings, as well as provide solutions for existing problems and shortcomings.
Section Six, Managerial Impact, presents contemporary coverage of the managerial applications and
implications of human computer interaction. This collection of research opens with Social Impact of
Virtual Networking by Hakikur Rahman, which documents the emergence of the virtual enterprise and
the impact such a structure has on social communication. Similarly, within the article Human Factors
for Networked and Virtual Organizations, Vincent E. Lasnik emphasizes the role that human factors
engineering must play in the future design of virtual and networked environments. Later contributions,
such as Knowledge Blogs in Firm Internal Use, investigate knowledge transfer within and among
organizations. Within this selection, authors Miia Kosonen, Kaisa Henttonen, and Kirsimarja Blomqvist
identify factors for the implementation of knowledge blogs within organizations and explain why these
tools can be used to encourage institutional memory, create identities, and to inform the organization
itself. The comprehensive research in this section offers an overview of the major issues that practitioners, managers, end users and even consumers must address in order to remain informed about the latest
managerial changes in the field of human computer interaction.
Section Seven, Critical Issues, presents readers with an in-depth analysis of the more theoretical and
conceptual issues within this growing field of study by addressing topics such as security, ethics, and gender
differences in technology adoption and use. Specifically, these topics are discussed in selections such as
Trusting Computers Through Trusting Humans: Software Verification in a Saftey-Critical Information
System by Alison Adam and Paul Spedding and Global Information Ethics: The Importance of Being
Environmentally Earnest by Luciano Floridi. Later selections, which include Emotional Digitalization
as Technology of the Postmodern: A Reflexive Examination from the View of the Industry, review more
novel issues, such as how the digitalization of emotions helps to bridge the gap between technology and
humanity. Specifically, within this chapter, author Claus Hohmann identifies the emotion inherent in the
design of new technologies and investigates how this has and will continue to impact human reaction. In
all, the theoretical and abstract issues presented and analyzed within this collection form the backbone
of revolutionary research in and evaluation of human-computer interaction.
The concluding section of this authoritative reference tool, Emerging Trends, highlights research
potential within the field of human-computer interaction while exploring uncharted areas of study for the
advancement of the discipline. Communicating in the Information Society: New Tools for New Practices by Lorenzo Cantoni and Stefano Tardini presents a framework for the latest digital communication
tools and their implementation, while the evolution of the semantic web is analyzed in Cristian Peraboni

xxxiv

and Laura A. Ripamontis Socio-Semantic Web for Sharing Knowledge. The nature of podcasts and
their role in encouraging collaboration and acting as an educational tool is explored in Stay Tuned for
Podcast U and the Data on M-Learning by Deborah Vess and Michael Gass and Podcastia: Imagining Communities of Pod-People by Jonathan Cohn. Other new trends, such as voice enabled interfaces
for mobile devices, programmable ubiquitous computing environments, and intelligent user interfaces
for mobile and ubiquitous computing are discussed in this collection. This final section demonstrates
that humanitys interaction with technology will continue to grow and evolve, shaping every facet of
modern life.
Although the contents of this multi-volume book are organized within the preceding eight sections
which offer a progression of coverage of important concepts, methodologies, technologies, applications,
social issues, and emerging trends, the reader can also identify specific contents by utilizing the extensive
indexing system listed at the end of each volume. Furthermore, to ensure that the scholar, researcher,
and educator have access to the entire contents of this multi-volume set, as well as additional coverage
that could not be included in the print version of this publication, the publisher will provide unlimited,
multi-user electronic access to the online aggregated database of this collection for the life of the edition free of charge when a library purchases a print copy. In addition to providing content not included
within the print version, this aggregated database is also continually updated to ensure that the most
current research is available to those interested in human-computer interaction.
As technology continues its rapid advancement, the study of how to successfully design, implement
and, ultimately, evaluate human reactions to and interactions with the modern world becomes increasingly critical. Innovations in the design of mobile devices, educational environments, and web sites have
all been made possible through a more thorough understanding of how humans react to and engage with
technological interfaces. Continued evolution in our understanding of the human-computer dynamic will
encourage the development of more usable interfaces and models that aim to more thoroughly understand
the theories of successful human-computer interaction.
The diverse and comprehensive coverage of human-computer interaction in this four-volume, authoritative publication will contribute to a better understanding of all topics, research, and discoveries in
this developing, significant field of study. Furthermore, the contributions included in this multi-volume
collection series will be instrumental in the expansion of the body of knowledge in this enormous field,
resulting in a greater understanding of the fundamentals while also fueling the research initiatives in
emerging fields. We at Information Science Reference, along with the editor of this collection, hope
that this multi-volume collection will become instrumental in the expansion of the discipline and will
promote the continued growth of human-computer interaction.

xxxv

Human Computer Interaction:

Concepts, Methodologies, Tools, and Applications

Panayiotis Zaphiris
City University of London, UK
Chee Siang Ang
City University of London, UK

FUNDAMENTAL CONCEPTS AND THEORIES IN HUMAN COMPUTER

INTERACTION
User-centered design begins with a thorough understanding of the needs and requirements of the users
(Shneiderman, 1992). User needs analysis (or Knowledge Elicitation) and evaluation methods in HCI
are a critical process to the success of requirements and design gathering (Maiden et al., 1995), usability
testing, and user evaluation stages of software development (Zaphiris & Kurniawan, 2001).
Examples of knowledge elicitation methods often start with initial questionnaire feedback, requirements
task walkthroughs, interviews techniques, and focus group sessions. They can rapidly scale upwards to
more complex psychometric design and evaluation processes. For example, various fidelities of prototype
construction, direct and indirect observation practices for monitoring user actions and response time
comparisons, and methods for eliciting mental categorization models, that is, in distinguishing expert
and non-expert technology usage patterns.
The measure of a good experience can vary from person to person; however the appropriate understanding of a usable design comes from gaining the knowledge that it is functional, efficient, and desirable to its intended audience (Kuniavksy, 2003).
John and Marks (1997) identify three key factors to assess the usability of an interface:
Usability is measured by the extent to which the intended goals of use of the overall system are achieved
(effectiveness), the resources that have to be expended to achieve the intended goals (efficiency), and
the extent to which the user finds the overall system acceptable (satisfaction) (John and Marks, 1997).
The usability of a system is also related to issues surrounding its accessibility. There is a broad range
of users to whom Web-based services are directed, and the services provided ought to be accessible to
them (e.g. visually, hearing, physically or cognitively impaired or even people with different experience
of and attitudes towards technology).

xxxvi

In the United Kingdom, The Disability Discrimination Act (DDA) began to come into effect in
December 1996 and brought in measures to prevent discrimination against people on the basis of disability. Part III of the Act aims to ensure that disabled people have equal access to products and services.
Under Part III of the Act, businesses that provide goods, facilities, and services to the general public
(whether paid for or free) need to make reasonable adjustments for disabled people to ensure they do
not discriminate by:

Refusing to provide a service;

providing a service of a lower standard or in a worse manner; and
providing a service on less favorable terms than they would to users without the disability.

There is a legal obligation on service providers to ensure that disabled people have equal access to
Web-based products and services. Section 19(1) (c) of the Act makes it unlawful for a service provider
to discriminate against a disabled person in the standard of service which it provides to the disabled
person or the manner in which it provides it.
An important provision here is that education is not covered by the DDA, but by separate legislation,
titled the Special Educational Needs and Disability Act 2001 (SENDA). This Act introduces the right for
disabled students not to be discriminated against in education, training, and any services provided wholly
or mainly for students, and for those enrolled in courses provided by responsible bodies, including
further and higher education institutions and sixth form colleges. Student services covered by the Act
can include a wide range of educational and non-educational services, such as field trips, examinations
and assessments, short courses, arrangements for work placements and libraries, and learning resources.
In a similar wording to the DDA, SENDA requires responsible bodies to make reasonable adjustments
so that people with disabilities are not at a substantial disadvantage.

DEVELOPMENT AND DESIGN METHODOLOGIES

There are many different elicitation and usability/accessibility evaluation techniques (Cooke, 1994) and
selecting the right technique in a particular situation is not trivial.
Burges Table of Knowledge Elicitation Methods (Burge, 2001) provides an extensive comparative
view of almost all the common KE techniques found in HCI.
In addition, usability and accessibility evaluation techniques are often grouped into two broad areas:
user-based (that often include user testing) and expert based (that often include heuristic evaluation and
cognitive walkthrough) techniques.
King et al. (2004) presented what they called An incremental usability and accessibility evaluation
framework for digital libraries. Their framework is broken down into seven key activities and addresses
all stages of a design of a system.

Activity 1: Conduct Query - Requirement Gathering

Identify satisfaction levels of current users of the system and establish key positive and negative
aspects of the interface, what features they would like to see, and so forth.

Activity 2: Analysis
Evaluate current findings and identify issues not yet addressed

xxxvii

Activity 3: Perform Empirical (user) Evaluations

We regard user testing as the strongest evaluation technique, allowing us to identify real user
problems by observing users interacting with the system. Retrospective focus groups or interviews
conducted after the evaluations also provide a volume of qualitative data.

Activity 4: Analysis
Establish key problems and assess if any areas of the service have not been covered by user evaluations

Activity 5: Expert Evaluations

Appropriate modification of expert evaluation techniques maybe required so that they supplement
previous evaluation findings, and address any areas or issues that have not as yet been covered

Activity 6: Analysis
Analyze all data identifying key issues that need to be addressed in the redesign of the service.
Establish new usability and accessibility goals for the design

Activity 7: Iterative Process

Re-conduct all stages in the iterative framework to evaluate redesign

We now describe some of the key methods that are associated with the above mentioned framework.

Interviewing
This query-based process elicits from users knowledge on a set information topic based on their expertise
in the domain in question. It is useful for obtaining behavioral reasoning and background knowledge.
Interviews can be categorized as structured or unstructured. Structured interviews elicit limited responses
from users, by using a series of closed questions that have to be answered based on given solutions.
This enables the user data to be analysed quicker but is not necessarily as informative as unstructured
(open ended) interviews.
Preece et al. (1994) suggests that interview processes are most effective as semi-structured based on
a series of fixed questions that gradually lead into more in-depth user needs and requirements understanding, then allowing for open ended responses to possibly create new dynamic questions based on
prior structured responses (Macaulay, 1996). On-site stakeholder interviews allows researchers to bring
about a vivid mental model of how users work with existing systems and how new systems can support
them (Mander and Smith, 2002).
Interviews are useful when combined with surveys or questionnaires, as they can be used to improve
the validity of data by clarifying specific issues that were raised in the survey or questionnaire.

Surveys
In conducting surveys, three things are necessary: a) the set of questions, b) a way to collect responses,
and c) access to the demographics group you wish to test. There are several widely reported templates
for acquiring different types of user data, such as the Quality of User Interface Satisfaction (QUIS) by
Chin et al. (1988) and the Computer System Usability Questionnaire (CSUQ) by IBM with Lewis et
al. (1995).

xxxviii

Surveys can be similarly open and closed question based, but also allow us to enquire scalar results
giving indicators of quality in positive and negative statements. Self-filling surveys can be time efficient
to deploy, and results from closed questions can be fast to analyze.
Open questions tend to elicit unanticipated information which can be very useful for early design.
Existing survey sampling techniques include face-to-face, paper- and pencil-based, telephone surveys
where the researcher will fill in the results (which becomes more of an interview style) but there is
modern interest in computer assisted and Web-based surveying techniques.

Focus Groups
This activity is useful for eliciting cross-representative domains of knowledge from several stakeholders/users, in an open discussion format. Sessions are often moderated and tend to be informal by nature,
centering on the creation of new topics from open questions.
Evidence shows that the optimal number needed for a mixed experience focus group is between five
to eight participants, with group size being inversely related to the degree of participation (Millward,
1995).

Observation
Observation methods elicit user knowledge from the way users interact with a prototype or a final product.
It can be direct, whereby a researcher is present and can steer users to particular points in an interaction.
This tends to utilize video camera equipment and note taking to successfully enquire the timeline of user
actions that is getting from point A to point D may require steps B or C.
The other model of observation is indirect, whereby all user actions are captured electronically.
The researcher has to maintain co-operation between users and should only pose questions if clarification is needed.

Paper Prototyping
There are several approaches to paper prototypes, enabling users to create quick and partial designs of
their concepts. It is often used in early stages of the design processes. Though the methodology lacks
standardization, Rettig (1994) distinguishes between high-tech and low-tech views, and the more commonly modeled categories are of low, medium and high fidelity prototypes (Greenberg, 1998). Rudd
et al. (1996) also distinguishes prototypes according to horizontal and vertical prototypes, with vertical representing deep functionality of a limited view to the final output, and horizontal giving a wide
overview of the full functionality of the system but with a weaker depth of understanding. Hall (2001)
discusses the benefits of using various levels of fidelities of prototypes.

Cognitive Walkthrough
Cognitive Walkthrough is an expert based evaluation technique that steps through a scenario/task by
focusing on the users knowledge and goals. The expert evaluator first starts with descriptions of: the
prototype interface, the task(s) from the users perspective, the correct sequence of actions needed to
complete the task using the prototype and any assumptions about the characteristics of the user.
Then the evaluator walks through the tasks using the system, reviewing the actions that are necessary
and attempting to predict how the users will behave

xxxix

A series of key questions are used throughout each sub-task evaluation:

Will the user be trying to achieve the correct effect?

Will the user know that the correct action is available?
Will user know that the correct action will achieve the desired effect?
If the correct action is taken, will the user see that things are going okay?

Heuristic Evaluation
Heuristic Evaluation is an expert review technique where experts inspect the interface to judge compliance with established usability principles (the heuristics)
Heuristic Evaluation is usually conducted in a series of four steps:

Prepare: create a prototype to evaluate; select evaluators; prepare coding sheets to record problems
Determine approach: either set typical user tasks (probably the most useful approach) or allow
evaluators to establish their own tasks or conduct an exhaustive inspection of entire interface
Conduct the evaluation: evaluators inspect interface individually to identify all violations of
heuristics (the usability problems); record the problem (feature and location), severity (based on
frequency, impact, criticality/cost) and heuristic violated
Aggregate and analyze results: group similar problems; reassess severity; determine possible
fixes

Hierarchical Task Analysis (HTA)

Hierarchical task analysis (HTA) is an analytic tool commonly used to describe peoples actions when
interacting with the artifacts. We can analyze peoples work or leisure activities by focusing on what
people do, their goals, mediating artifacts and their background knowledge. The primary use of HTA
is to understand, clarify and organize our knowledge about existing system and work. It is applied in
design and evaluation of the interactive system.
Essentially, HTA is concerned with the structure of tasks people are engaged in and how the tasks
can be broken down into steps of sub-task hierarchies and how the order of sub-tasks can be described
meaningfully.
In HTA, we break down a task from top to bottom, showing a hierarchical relationship amongst the
tasks. This allows us to view the task at different levels of details
A task is the set of activities in which a user engages to achieve a goal. In other words, a task consists of
a sequence of actions performed to achieve a goal. A goal is defined as the desired state of a system.
Usually HTA involves collecting information using data collection techniques (e.g. interviews, questionnaires, observations) and then analyzing it.
HTA can be presented as text or diagrams to show the hierarchy.
There are generally eight steps for task analysis:

Define the purpose of analysis and the boundary of the system.

Identify user groups, select representatives, and identify main tasks of concern.
Design and conduct data collection to elicit information about these tasks.
Identify the goals (and sub-goals) that users are trying to achieve, the steps they engage in, their
rationale, and the information resources they use.

xl

Obtain information from documentation, interviews, questionnaires, focus groups, observation,

ethnography, experiments, and so forth.
Analyze the data to model specific instances of tasks initially, and then consider decomposition of
tasks, balance of the model and stopping rules.
Generalize across the specific task models to create a generic (or composite) task model: From
each task model for the same goal produce a generic model that includes all the different ways of
achieving the goal.
Check the models with users, other stakeholders, analysts, and iterate the analysis process.

TOOLS AND TECHNOLOGIES

HCI community approaches design problems using different tools to help identify the users, their needs
and behaviors, user interface problems. Various kinds of tools have been developed to tackle different
research objectives. We can categorize these tools into three major categories: hardware, software and
analytic tools. In this section we will describe an example of tool for each category

Eye Tracking
Eye tracking has been used for a very long time in psychology, focusing on recording eye movements
while reading. However, in the 1980s researchers began to incorporate eye tracking into issues of human
computer interaction. As technological tools such as the Internet, e-mail, and videoconferencing evolved
into viable means of communication and information sharing during the 1990s and beyond, researchers
started using eye tracking in order to answer question about usability. Eye tracking technologies have
been widely used as a proxy for users attention and the eye movement data gathered helps to understand
where people focus attention, and in what order before they make a selection in their interactions with
computer interfaces.
Goldberg and Kotval (1999) made a convincing argument for the use of eye-movement data in interface evaluation and usability testing. They claim that Interface evaluation and usability testing are
expensive, time-intensive exercises, often done with poorly documented standards and objectives. They
are frequently qualitative, with poor reliability and sensitivity. The motivating goal for their research
work assessing eye movements as an indicator of interface usability was the provision of an improved
tool for rapid and effective evaluation of graphical user interfaces.
They performed an analysis of eye movements (using interface designers and typical users) in order
to assess the usability of an interface for a simple drawing tool. Comparing a good interface with wellorganized tool buttons to a poor interface with a randomly organized set of tool buttons, the authors
could show that the good interface resulted in shorter scan paths that cover smaller areas. The chief merit
of this study was the introduction of a systematic classification of different measures for evaluating the
usability of user interfaces based on eye movement data. These measures were grouped as follows: Measures of Processing (Number of fixations, Fixation Duration, Fixation / Saccade ratio, Scanpath Length,
Scanpath Duration, Convex Hull Area, Spatial Density, Transition Matrix, Number of Saccades, Saccadic
Amplitude), Other Measures (Backtrack, On-target / all-target fixations, Post-target fixations).

Usability Testing Software

Due to the escalating importance of HCI in software and Web design, some tools have been developed to
assist the process of usability testing. Traditionally, usability testing involves the use of video cameras,

xli

screen capture tools, microphones, and so forth. One of the main challenges lies in the integration of the
media in various formats in a coherent way. Together with cheap computer hardware, usability testing
software overcomes this problem by providing a coherent solution, in which Webcam, microphone and
screen capture software operate concurrently under one system managed by the usability software. The
output is seamless presentation of all media file which can be edited and annotated.
Typically, usability testing tools (hardware and software) consists of the following functionalities:

User experience recording:

The most common activity in usability testing is probably recording user performing tasks with the
software or Web site under testing. This includes video recording (of the user), screen capture and
voice recording (verbal protocol). With the usability testing tools, we can set the tasks needed to be
performed, record the performance and administer questionnaires in an integrated environment.

Observation and logging:

Some usability tools support the connected of network computers thus allowing multiple observers
to monitor the users and annotate the videos collaboratively. Furthermore, most tools can also log
the mouse click and keyboard pressing.

Analysis
Although still limited to basic visualization, usability tools can support data analysis by calculating aggregate usability metrics for users or for tasks. For instance, we are able to quickly identify tasks with
low usability metrics and thus focus on improving the design relevant to the tasks.
This reduces the amount of work and time significantly and it makes usability testing less costly.

UTILIZATION AND APPLICATION OF HUMAN COMPUTER INTERACTION

We now describe a number of key areas where HCI can have a practical application.

ComputerAugmented Environments
One application area in which HCI plays an important role is the computer-augmented environments,
or commonly known as augmented reality or mixed reality. It refers to the combination of real world
and computer-generated data visualization. In other words it is an environment which consists of both
real world and virtual reality. For instance, a surgeon might be wearing goggles with computer generated medical data projected on it. The goggles are said to augment the information the surgeon can see
in the real world through computer visualization. Therefore, it is not difficult to see the connection of
augmented reality with ubiquitous computing and wearable computers.
Since its inception, augmented reality has had an impact on various application domains. The most
common use is probably the support of complex tasks in which users need to perform a series of complicated actions while having access to large amount of information at the same time, such as surgery,
assembly and navigation. Apart from these, augmented reality is also used for learning and training,
such as flight and driving simulations.
Augmented reality implementation usually requires additional devices for input and output in order
to integrate computer generated data into real world:

xlii

A Cave Automatic Virtual Environment multi-user, room-sized, high-resolution, 3D video and audio
immersive environment in which the virtual reality environment is projected onto the walls. The user
wearing a location sensor can move within the display boundaries, and the image will move with and
surrounds the user.
A head-up display (HUD) is transparent display that presents data without obstructing the users
view. It is usually implemented on vehicles in which important information is projected directly in the
drivers viewing field. Thus the user does not need to shift attention between what is going on in the real
world and the instrumental panel.
A head-mounted display is a display device, worn on the head or as part of a helmet that has a small
display optic in front of one or both eyes.
Some of these devices have become commercially available and increasingly affordable. The challenge of HCI lies in the design of information visualisation which is not obtrusive to the users tasks.

ComputerBased Learning
A lot of effort has been put in coupling learning and technology to design effective and enjoyable learning. Various areas, namely e-learning, computer-based learning, serious games, etc have emerged, hoping
to utilize the interactive power of computers to enhance teaching and learning experience. A myriad of
design strategies have been proposed, implemented and evaluated, these include the early use of computer
in presentation, drill and practice (the behaviourist paradigm), tutorials (cognitivist paradigm), games,
story telling, simulations (constructivist paradigm), and so forth. As we progress from behaviorist to
constructivist, we notice an explosion of user interface complexity. For instance, drill and practice programs usually consist on a couple of buttons (next, previous buttons, buttons for multiple choice, etc)
while simulations could involve sophisticated visualization (outputs) and various user interface elements
for manipulating parameters (input). Recently computer-based learning has moved from single user offline environments to online network spaces in which a massive number of users can interact with each
other and form a virtual learner community. This social constructivist learning paradigm requires not
only traditional usability treatment, but also sociability design in which the system includes not only the
learning tools, but other sociability elements such as rules and division of labors.

Information Visualization
Information visualization is an area in HCI which can be related to many other areas such as augmented
reality just described before. Most modern computer applications deal with visual outputs. Graphical user
interface has almost entirely replaced command-based interaction in many domains. Information visualization can be defined as the use of computer supported, interaction, visual representations of abstract
data to amplify cognition (Shneiderman, 1992). To amplify cognition means that visualization shifts
cognitive loads to the perceptual system, thus expanding working memory and information storage.
Visualization provides a more perceptually intuitive way of viewing raw data, thus allowing users
to identify relevant patterns which would not have been identified in raw data.
Therefore, it has a huge impact on many applications domains, ranging from engineering, education,
various fields in science, and so forth.
In HCI, the most obvious application is the use of visualization is in the design of graphical user
interface that allows more intuitive interaction between human and computers. Various innovative
interaction styles have been developed such as WIMP (window, icon, menu, pointing device) which is
so familiar in todays software. Three-dimensional graphics are also emerging although currently they

xliii

are mostly used in computer games and computer-aided design. One recent example of 3D graphical
interface is the new windows navigation and management known as Windows Flip 3D in Windows Vista
which allows the user to easily identify and switch to another open window by displaying 3D snapshot
thumbnail preview of all windows in stack.
Today, Information visualization is not only about creating graphical displays of complex information
structures. It contributes to a broader range of social and collaborative activities. Recently, visualization
techniques have been applied on social data to support social interaction, particularly in CMC. This area
is known as social visualization by (Donath, Karahalios, & Vigas, 1999). Other technique such as social
network analysis has also become increasingly important in visualization social data.
Other areas where HCI plays an important role include: Intelligent and agent systems; Interaction
design; Interaction through wireless communication networks; Interfaces for distributed environments;
Multimedia design; Nonverbal interfaces; Speech and natural language interfaces; Support for creativity; Tangible user interfaces; User interface development environments and User support systems.

ORGANISATIONAL AND SOCIAL IMPLICATIONS OF HUMAN COMPUTER

INTERACTION
The application of HCI can have an effect of organisational and social dimensions. For example, the area
of computer supported collaborative work (CSCW) explores the effect the introduction of technology
can have an effect on the organisational structure and the way of work of companies and organisations.
Similarly the study of how we use technology to communicate with each other is gaining strong interest
in the HCI research community.
The expansion of the Internet has resulted in an increase in the usefulness of Computer Mediated
Communication (CMC) and the popularity of online communities. It is estimated that 25% of Internet
users have participated in chat rooms or online discussions (Madden & Rainie, 2003).
It is by now no secret how vital the Internet was, is, and will continue to be in our lives. One of the
most important characteristics of this medium is the opportunities it offers for human-human communication through computer networks. As Metcalfe (1992) points out, communication is the Internets most
important asset. E-mail is just one of the many modes of communication that can occur through the use
of computers. Jones (1995) points out that through communication services, like the Internet, Usenet
and bulletin boards, online communication has for many people supplanted the postal service, telephone
and even the fax machine. All these applications where the computer is used to mediate communication
are called Computer-Mediated Communication (CMC).
December (1997) defines CMC as the process by which people create, exchange, and perceive information using networked telecommunications systems (or non-networked computers) that facilitate
encoding, transmitting, and decoding messages. He emphasizes that studies of CMC view this process
from different interdisciplinary theoretical perspectives (social, cognitive/psychological, linguistic,
cultural, technical, political) and often draw from fields such diverse as human communication, rhetoric
and composition, media studies, human-computer interaction, journalism, telecommunications, computer
science, technical communication and information studies.
Online communities emerge through the use of CMC applications. The term online community is
multidisciplinary in nature, means different things to different people, and is slippery to define (Preece,
2000). For purposes of a general understanding of what online communities are, Rheingolds definition
of online communities is presented:

xliv

[online] communities are social aggregations that emerge from the Net when enough people carry on
those public discussions long enough, with sufficient human feeling, to form Webs of personal relationships in cyberspace. (Rheingold, 1993, 5)
Online communities are also often referred to as cyber societies, cyber communities, Web groups, virtual
communities, Web communities, virtual social networks and e-communities among several others.
The cyberspace is the new frontier in social relationships, and people are using the Internet to make
friends, colleagues, lovers, as well as enemies (Suler, 2004). As Korzeny pointed out, even as early as
1978, online communities are formed around interests and not physical proximity (Korzeny, 1978).
In general, what brings people together in an online community is common interests such as hobbies,
ethnicity, education, beliefs. As Wallace (1999) points out, meeting in online communities eliminates
prejudging based on someones appearance, and thus people with similar attitudes and ideas are attracted
to each other.
It is estimated that as of September 2002 there are over 600 million people online (Nua Internet
Surveys, 2004). The emergence of the so-called global village was predicted years ago (McLuhan,
1964) as a result of television and satellite technologies. However, it is argued by Fortner (1993) that
global metropolis is a more representative term (Choi & Danowski, 2002). If one takes into account
that the estimated world population of 2002 was 6.2 billion (U.S. Census Bureau, 2004), then the online
population is nearly 10% of the world population a significant percentage which must be taken into
account when analyzing online communities. In most online communities, time, distance and availability
are no longer disseminating factors. Given that the same individual may be part of several different and
numerous online communities, it is obvious why online communities keep increasing in numbers, size
and popularity.
Preece et al. (2002) states that an online community consists of people, a shared purpose, policies
and computer systems. She identifies the following member roles:

Moderators and mediators: who guide discussions/serve as arbiters

Professional commentators: who give opinions/guide discussions
Provocateurs: who provoke
General Participants: who contribute to discussions
Lurkers: who silently observe

CMC has its benefits as well as its limitations. For instance, CMC discussions are often potentially
richer than face-to-face discussions. However, users with poor writing skills may be at a disadvantage
when using text-based CMC (Scotcit, 2003).

Examples of CMC and Online Communities

Examples of CMC include asynchronous communication like e-mail and bulletin boards; synchronous
communication like chatting; and information manipulation, retrieval and storage through computers
and electronic databases (Ferris, 1997).
When it comes to Web site designers, choosing which CMC to employ (for instance, forum or chatroom) is not a matter of luck or randomness. Selecting the right CMC tool depends on a lot of factors. For
example, in the case of e-learning, the choice of the appropriate mode of CMC will be made by asking
and answering questions such as (Bates, 1995; CAP, 2004; Heeren, 1996; Resier and Gagne, 1983):

xlv

Are the users spread across time zones? Can all participants meet at the same time?
Do the users have access to the necessary equipment?
What is the role of CMC in the course?
Are the users good readers/writers?
Is the activities time independent?
How much control is allowed to the students?

Audio-conferencing is a real time communication mechanism, since the communication happens

synchronously. Depending on the application, text chat and graphics may also be supported. Videoconferencing, like audio-conferencing, offers a useful mode of communication, but has the added benefit
of being able to also see the participants, instead of just hearing them.
IRC and chats also support synchronous communication, since they enable the users to carry out
conversations through the use of text messaging. MUDs build on chats by providing avatars and graphical environments where the users can engage in interactive fantasy games (Preece, 2000).
World Wide Web web sites are usually asynchronous, providing community information and links to
other sites, but sometimes also have synchronous software, like chats, embedded in them (Preece, 2000).
E-mail is an asynchronous mode of communication usually in the form of text. However, the ability
to add attachments to e-mail messages makes it possible for audio, video and graphics to be used also.
Voice mail is an expansion of e-mail whereby users may record themselves speaking out a message
and then send that voice file to their contact, instead of typing it. Newsgroups, like e-mail, provide an
asynchronous mode of communication, but unlike e-mail where the messages come to the users, it is a
pull technology meaning the users must go to the UseNet groups themselves (Preece, 2000). Finally,
discussion boards, also referred to as forums or bulletin boards provide an asynchronous mode of communication where the users post messages for others to see and respond at their own time.
Preece (2000) describes in detail some of these types of CMC and their different characteristics. In
this chapter, we focus on Wiki and Online Virtual Game Environments which in our view provide a set
of new, novel modes of communication and online community building systems.

Analyzing Online Communities: Frameworks and Methodologies

There are various aspects and attributes of CMC that can be studied to help us better understand
online communities. For instance, the analysis of the frequency of exchanged messages and the formation of social networks, or, the analysis of the content of the exchanged messages and the formation of
virtual communities. To achieve such an analysis a number of theoretical frameworks have been developed and proposed. For example Henri (1992) provides an analytical model for cognitive skills that
can be used to analyze the process of learning within messages exchanged between students of various
online e-learning communities. Masons work (1991) provides descriptive methodologies using both
quantitative and qualitative analysis. Furthermore, five phases of interaction analysis are identified in
Gunawardena et al.s model (1997):

Sharing/Comparing of Information
The Discovery and Exploration of Dissonance or Inconsistency among Ideas, Concepts or Statements
Negotiation of Meaning/Co-Construction of Knowledge
Testing and Modification of Proposed Synthesis or Co-Construction
Agreement Statement(s)/Applications of Newly Constructed Meaning

xlvi

In this section we provide a description of some of the most commonly used online community
evaluation techniques as well as their weaknesses and strengths.

Studying CMC
All of the already mentioned methods can be used to evaluate the usability and accessibility of the online
community. But, apart from usability and accessibility we often also want to evaluate the user experience
and the sociability of the interface. We describe below some of the methods that can assist us in this:

Personas
Findings from interviews and questionnaires can be further used as a basis for developing user profiles
using personas. A persona is a precise description of the user of a system, and of what he/she wishes
to accomplish. (Cooper, 1999). The specific purpose of a persona is to serve as a tool for software and
product design. Although personas are not real people, they represent them throughout the design stage
nd are best based on real data collected through query based techniques.
Personas are rich in details, include name, social history and goals, and are synthesized from findings
through the use of query based techniques with real people (Cooper, 1999). The technique takes user
characteristics into account and creates a concrete profile of the typical user (Cooper, 1999).
For online communities, personas can be used to better understand the participants of the community
and their background. Personas can also be used as a supplement to Social Network Analysis (described
later in this chapter) to get a greater overview of the characteristics of key participants of a community.
Using personas, Web developers gain a more complete picture of their prospective and/or current users
and are able to design the interfaces and functionality of their systems, to be more personalized and
suited for the communication of the members of their online communities.
Advantages of personas include: can be used to create user scenarios; can be anonymous protecting
use privacy; represent the user stereotypes and characteristics.
Disadvantages of personas include: if not enough personas are used, users are forced to fall into a
certain persona type which might now accurately represent them; time-consuming.

Log Analysis
A log, also referred to as Web-log, server log or log-file is in the form of a text file and is used to track
the users interactions with the computer system they are using. The types of interaction recorded include
key presses, device movements and other information about the user activities. The data is collected
and analyzed using specialized software tools and the range of data collected depends on the log settings. Logs are also time stamped and can be used to calculate how long a user spends on a particular
task or how long a user is lingered in a certain part of the Web site (Preece, Rogers & Sharp, 2002). In
addition, an analysis of the server logs can help us find out: when people visited the site, the areas they
navigated, the length of their visit, the frequency of their visits, their navigation patterns, from where
they are connected and details about the computer they are using.
Log analysis is a useful and easy to use tool when analyzing online communities. For example,
someone can use log analysis to answer more accurately questions like student attendance of an online
learning community. Furthermore, logs can identify the Web pages users spend more time viewing, and
also the paths that they used. This helps identify the navigation problems of the Web site, but also gives

xlvii

a visualization of the users activities in the virtual communities. For instance, in the case of e-learning
communities, the log files will show which students are active in the CMC postings even if they are not
active participants (few postings themselves), but just observing the conversations. Preece (2003) notes
that data logging does not interrupt the community, while at the same time can be used to examine mass
interaction.
Advantages of Logs (Preece et al., 2002): helps evaluators analyze users behavior; helps evaluators
understand how users worked on specific tasks; it is unobtrusive; large volumes of data can be logged
automatically.
Disadvantages of Logs (Preece et al., 2002): powerful tools are needed to explore and analyze the
data quantitatively and qualitatively; user privacy issues.

Content and Textual Analysis

Content analysis is an approach to understanding the processes that participants engage in as they exchange messages. There have been several frameworks created for studying the content of messages
exchanged in online communities. Examples include work from Archer, Garrison, Anderson & Rourke
(2001), Gunawardena, Lowe, and Andersons (1997) model for examining the social construction of
knowledge in computer conferencing, Henris (1992) content analysis model and Fahy et al.s (2001)
Transcript Analysis Tool (TAT) which is described in more detail below.
The TAT focuses on the content and interaction patterns at the component level of the transcript (Fahy
et al., 2001). After a lengthy experience with other transcript tools and reviews of previous studies Fahy
et al. (2001), chose to adapt Zhus (1996) analytical model for the TAT. Zhus model (1996) examines
the forms of electronic interaction and discourse, the forms of participation and the direction of participant interaction in computer conferences. The TAT also contains echoes of Vygotskian theory, primarily
those dealing with collaborative sense making, social negotiation and proximal development (Cook &
Ralston, 2003). The TAT developers have come up with the following strategic decisions (Fahy, 2003):
The sentence is the unit of analysis; the TAT is the method of analysis; interaction is the criterion for
judging conference success and topical progression (types and patterns).
The TAT was designed to permit transcript content to be coded reliably and efficiently (Fahy et al.,
2001), while the advantages of TAT are (Fahy, 2003; Cook & Ralston, 2003; Fahy et al., 2001): It reveals
interaction patterns that are useful in assessing different communication styles and online behavioral
preferences among participants; It recognizes the complexity of e-conferences and measures the intensity
of interaction; It enables the processes occurring within the conferences to be noted and recorded; It
probes beyond superficial systems data, which mask the actual patterns of discussion; It relates usefully
to other work in the area; It discriminates among the types of sentences within the transcript; It reflects
the importance of both social and task-related content and outcomes in transcript analysis research.
The unit of analysis of the TAT is the sentence. In the case of highly elaborated sentences, the units
of analysis can be independent clauses which, punctuated differently, could be sentences (Fahy, 2003).
Fahy et al. (2001), have concluded that the selection of message-level units of analysis might partially
explain problematic results that numerous researchers have had with previous transcript analysis work.
They also believe that the finer granularity of sentence-level analysis results in several advantages (Fahy,
2003; Ridley & Avery, 1979): Reliability; Ability to detect and describe the nature of the widely varying
social interaction, and differences in networking pattern, in the interactive behavior of an online community, including measures of social network density and intensity; Confirmation of gender associations
in epistolary/expository interaction patterns, and in the use of linguistic qualifiers and intensifiers.

xlviii

Social Network Analysis (SNA)

Social Network Analysis (SNA) is the mapping and measuring of relationships and flows between
people, groups, organisations, computers or other information/knowledge processing entities. The
nodes in the network are the people and groups while the links show relationships or flows between
the nodes. SNA provides both a visual and a mathematical analysis of human relationships (Krebs,
2004, pp.1). Preece (2000) adds that it provides a philosophy and a set of techniques for understanding
how people and groups relate to each other, and has been used extensively by sociologists (Wellman,
1982; Wellman 1992), communication researchers (Rice, 1994; Rice et al., 1990) and others. Analysts
use SNA to determine if a network is tightly bounded, diversified or constricted; to find its density and
clustering; and to study how the behavior of network members is affected by their positions and connections (Garton, Haythornhwaite & Wellman, 1997; Wellman, 1997; Henneman, 1998; Scott, 2000).
Network researchers have developed a set of theoretical perspectives of network analysis. Some of these
are (Bargotti, 2002):

Focus on relationships between actors than the attributes of actors.

Sense of interdependence: a molecular rather atomistic view.
Structure affects substantive outcomes.
Emergent effects.

The aim of social network analysis is to describe why people communicate individually or in groups
(Preece, 2000, pp. 183), while the goals of SNA are (Dekker, 2002):

To visualize relationships/communication between people and/or groups using diagrams.

To study the factors which influence relationships and the correlations between them?
To draw out implications of the relational data, including bottlenecks.
To make recommendations to improve communication and workflow in an organisation.
Preece (2002) and Beidernikl & Paier (2003) list the following as the limitations of SNA:

More theory that speaks directly to developers of online communities is needed

The data collected may be personal or private.
The analysis of the data is quantitative and specific to the particular network, while common survey
data are qualitative and generalize answers on the parent population.

It is also worth pointing out that network analysis is concerned about dyadic attributes between pairs
of actors (like kinship, roles, and actions), while social science is concerned with monadic attributes of
the actor (like age, sex, and income).
There are two approaches to SNA:

Ego-centered analysis: Focuses on the individual as opposed to the whole network, and only a
random sample of network population is normally involved (Zaphiris, Zacharia, & Rajasekaran,
2003). The data collected can be analyzed using standard computer packages for statistical analysis
like SAS and SPSS (Garton, Haythornthwaite, & Wellman, 1997).

xlix

Whole network analysis: The whole population of the network is surveyed and this facilitates
conceptualization of the complete network (Zaphiris et al., 2003). The data collected can be analyzed using microcomputer programs like UCINET and Krackplot (Garton et al., 1997).

The following are important units of analysis and concepts of SNA (Garton et al., 1997; Wellman,
1982; Hanneman, 2001; Zaphiris et al, 2003; Wellman, 1992):

Nodes: The actors or subjects of study.

Relations: The strands between actors. They are characterized by content, direction, and
strength.
Ties: Connect a pair of actors by one or more relations.
Multiplexity: The more relations in a tie, the more multiplex the tie is.
Composition: This is derived from the social attributes of both participants.
Range: The size and heterogeneity of the social networks.
Centrality: Measures who is central (powerful) or isolated in networks.
Roles: Network roles are suggested by similarities in network members behavior.
Density: The number of actual ties in a network compare to the total amount of ties that the network
can theoretically support.
Reachability: In order to be reachable, connections that can be traced from the source to the required
actor must exit.
Distance: The number of actors that information has to pass through to connect the one actor with
another in the network.
Cliques: Sub-sets of actors in a network, who are more closely tied to each other than to the other
actor who are not part of the subset.

Social Network Analysis is a very valuable technique when it comes to analyzing online communities
as it can provide a visual presentation of the community and more importantly it can provide us with
qualitative and quantitative measures of the dynamics of the community.

EMERGING TRENDS IN HUMAN COMPUTER INTERACTION

Computer Game, Play and Immersion
Researchers from various areas have begun exploring the role of emotion in human activities. This research has proven to yield important results that not only deepen our understanding on human emotion,
but potentially improves the design and development of artifacts. This goes without saying that the topic
of emotion is becoming increasingly important in HCI. Probably one of the most studied domains in
emotional design is the computer game.
Computer games have been widely recognized as the software which is potentially addictive due to
their capability to engage and immerse the players for hours of game play. Therefore, it can be claimed
that emotion is vital to maintain players focus and to create enjoyable experiences.
Unlike work-oriented activity (and thus work-oriented software that mediate them), game play traditionally does not focus on productive outcome. In this case, usability of game play cannot be understood
in a traditional sense. Already new metrics of usability (or playability) have been proposed for designing
playful artifacts/software.

Some effort has also been taken to incorporate game elements into productive activities. For instance,
a project has been carried out to use games to label the contents of images meaningfully. Others have
used games for education and training. An area of research, known as serious games, is expanding
quickly to study productive games.
Recently, playability design has undergone a major transformation as games are becoming increasingly
collaborative with the emergence of massively multiplayer online role-playing games (MMORPGs).
These games are becoming one of the most interesting interactive media of computer-mediated communication and networked activity environments (Taylor, 2002). Understanding the pattern of participation in these game communities is crucial, as these virtual communities function as a major mechanism
of socialization of the players.
Some usability studies have shown that MMORPG design should incorporate what is known as
sociability. For instance research has found that game locations can be designed to encourage different
styles of social interactions

CMC, Online Community and Social Computing

Secondly, with the development of high performance computer networks, the use of computers is becoming increasingly collaborative. Therefore, topics such as CSCW, CMC, or social computing in general
will continue to thrive.
One of the studied issues in this research topic is the use of non-verbal cues in the CMC technology.
Some newer CMC media have attempted to address this issue through the implementation of video
streaming, although a more common way is the use of emoticons that allows the users to express their
emotions through graphical icons that show symbolic facial expressions.
Identity and anonymity is another popular topic regarding CMC since it is possible for users to hide
their real identity simply by choosing a different name. In most cases, even the administrators do not
know their real identity. Furthermore, users may treat CMC as an experimental environment and thus
play with their identity and try out different personalities (Turkle, 1995).
Some studies have been carried out to compare the nature of friendship in conventional CMC and
offline communication. Wellman and Gulia (1999) for instance found that people have a higher number
of online friends because it is easier to make friends online than offline. However, the quality of the
relationship is weaker in an online setting. Although Walther (1995) contended that it is possible to
develop strong online friendship, it takes longer than in offline communication.

3D CMC and Virtual World

Conventional CMC technologies are mainly text-based (Pfeil, 2007). Although the fundamental technology of CMC has not changed in the last decade (Preece & Maloney-Krichmar, 2003), the way of using
it for human-human communication has evolved considerably, from e-mail and synchronous chatting
to online forum, blogs and Wikis.
New technologies such as 3D Virtual spaces, such as MMORPGs and Second Life (Linden Lab, 2003)
offer a much richer form of social interactions where users can not only communicate through text-based
chat, but also interact with each other through virtual artefacts. Furthermore, these technologies not only
mediate social interaction, but also support group formation, which leads to community building. As
such, user interaction is considerably more complicated than the conventional CMC tool.
These 3D worlds often feature a large number of virtual locations users can visit and a variety of
virtual artefacts users can use to interact with, with each other. In some cases, they can even construct

li

new artefacts and locations utilising simple modelling tools and scripting languages. Whilst some 3D
virtual worlds are designed with game-like goal structures that impose obstacles or challenges, some
are completely open, meaning that the users are free to do as they please.
Although sociability issues of conventional CMC are well studied and documented, we have very
little understanding on social interactions in 3D CMC. Therefore, it is worth investigating user activities
in such environments in order to cast some light on the group formation process and other sociability
issues.
It might be potentially more challenging in researching this 3D CMC, in which communication takes
place both through texts and other virtual actions users can perform with their 3D avatars. Unlike the
avatar in conventional CMC which is often a static graphical or animated representation of the user, in
3D CMC, the avatar can interact with other avatars directly in the virtual space. A 3D avatar can perform a wide range of actions on the 3D world and other avatars. For instance, it is not uncommon that
avatars can hug, kiss or wave to each other. There is also research on the facial expression of 3D avatars
(Clarkson et al., 2001), implemented in the 3D CMC context with the intention to enhance non-verbal
communication. Moreover, given the fantasy theme of some 3D CMC environments, new sets of rules
for virtual communication which are completely different from physical communication might arise.
This is worth investigating as well as groups open operate within the boundary of norms and rules that
emerge through user interaction.

Ubiquitous Computing
Another exciting future trend in HCI is the emergence of ubiquitous computing, in which information
processing is thoroughly diffused into objects and experiences of everyday life. In another word, computers are disappearing. Users are no longer consciously engaged in using the computers. Instead they
are operating these devices which are so well integrated into artifacts of everyday activities without
being aware of using the computers.
Perhaps the most obvious example is the mobile phone, and indeed mobile computing has witnessed
an explosion of research interest within and beyond HCI community.
Other less obvious examples could include computerized refrigerators which are able to detect their
contents, plan and recommend a variety of recipes, automatically shop according to the users needs.
The focus of ubiquitous computing from the point of view of HCI suggests a shift from tool-focused
design to activity-focused design. The primary objective is thus to design tools which can be seamlessly
mediate everyday activities without interfering with users tasks. One such area is wearable computing
which has the potential to support human cognitions, facilitate creativity and communication. Unlike
mobile devices, wearable computers are attached to human, thus reducing the possibility of interruption
or displacement.

Novel Interaction Techniques

With the integration of the computers into everyday life, there is a call for new interaction techniques
as keyboards and mice are no longer sufficient as input devices.
Techniques such as natural language processing (NLP), a sub area of artificial intelligent, have been
incorporated into the design of computer devices to recognize speeches, thus making interaction more
intuitive. This automatic speech recognition can not only convert spoken words into machine readable
strings, but also recognize and identify the person who speaks. Already, it has some impacts on our
everyday life particularly on disable people who suffer from motor control impairment.

lii

Perhaps a more ambitious application of NLP is to go beyond text-speech or speech-text conversion

to actually understand human natural language. The research goal in this area is to develop a machine
that can engage in meaningful conversation with human. Although still in its infancy, it is certainly a
promising area which will benefit HCI greatly.
Another interesting interaction mode is what is known as haptic interaction which draws largely from
the infamous interaction style in HCI, direct manipulation. The idea is to provide the users with greater
freedom of manipulation that includes touch and feel. This technique aims to effectively bridge the gap
between tangible interfaces and virtual/intangible representations. One example is the augmentation
of every day objects in which digital data is superimposed on physical objects which can be used to
manipulate the digital data.
Recent development has also begun to utilize other parts of our body for interaction. Eye tracking
for instance, which has been used extensively in research in psychology, cognitive linguistic and other
areas, has been adopted in HCI research to study usability of Web sites and other software. Recently, its
use in HCI has expanded to be applied as an interaction technique. Eye tracking system is now being
used especially for the disabled to surf the Web, e-mail, and so forth with eye movements.

Accessibility and Universal Design

Finally we believe that Endeavour in designing assessable technology will continue to be a main focus
in HCI. One major accessibility research is the Web design since the Web has rapidly become more and
more pervasive in almost everybodys lives.
The challenge lies in the creation of resources that can be used by the widest spectrum of potential
visitors rather than an idealized average, there is a need to consider universality of Web delivery.
The idea accessibility has gone beyond the basic concept of designing for disability to universal
design, allowing access by everyone.
This requires consideration of the needs and requirements of individuals with disabilities, older persons,
people for whom English is a second language, people whose cultures and backgrounds are dissimilar
to those of Web developers, and those using outdated hardware and software, to name a few.
For that reason, the key objective of this research direction is to look at various aspects of universal
Web design and evaluation in a new perspective by focusing on the user aspect of universal Web design
and interaction and to present the wide range of advanced technology that can help disadvantaged users
get access to Web information.
The same idea also applies to software design. For instance, child-computer interaction has emerged
as a sub-field of HCI that deals with designing particularly multimedia learning software for children.
Other active areas of research include wed design for elderly people, gender differences, cultural issues,
and so forth.
All these efforts in accessibility design no longer contribute to the development of ease-of-use applications for specific user groups, but also improve interaction experiences for main stream users in
general.

REFERENCES
Archer, W., Garrison, R. D., Anderson, T., & Rourke, L. (2001). A framework for analyzing critical
thinking in computer conferences. European Conference on Computer-Supported Collaborative Learning, Maastricht, Nerthelands.

liii

Bates, A. W. (1995). Technology, open learning and distance education. London: Routledge.
Beidernikl, G., & Paier, D. (2003, July). Network analysis as a tool for assessing employment policy. In
Proceedings of the Evidence-Based Policies and Indicator Systems Conference 03. London.
Borgatti, S. (2000). What is Social Network Analysis. Retrieved on November 9, 2004 from http://www.
analytictech.com/networks/whatis.htm
Burge, E. L., & Roberts, J. M. (1993). Classrooms with a difference: A practical guide to the use of
conferencing technologies. Ontario: University of Toronto Press.
Burge, J. E. (2001). Knowledge Elicitation Tool Classification. PhD Thesis, Worcester Polytechnic
Institute.
CAP, University of Warwick. (2004). E-Guide: Using computer Mediated Communication in Learning
and Teaching. Retrieved November 8, 2004 from http://www2.warwick.ac.uk/services/cap/resources/
eguides/cmc/cmclearning/
Chin, J. P., Diehl, V. A., Norman, K. L. (1988). Development

of an Instrument Measuring User Satisfaction of the Human-Computer Interface. ACM CHI88 Proceedings, (pp. 213-218).
Choi, J. H., & Danowski, J. (2002). Cultural communities on the net - Global village or global metropolis?: A network analysis of Usenet newsgroups. Journal of Computer-Mediated Communication, 7(3).
Clarkson, M. J., Rueckert, D., Hill, D. L. G., & Hawkes, D. J. (2001). Using photo-consistency to register
2d optical images of the human face to a 3d surface model. IEEE Transactions on Pattern Analysis and
Machine Intelligence, 23(11), 1266-1281.
Cook, D., & Ralston, J. (2003). Sharpening the Focus: Methodological issues in analyzing on-line conferences. Technology, Pedagogy and Education, 12(3), 361-376.
Cooke, N. J. (1994). Varieties of knowledge elicitation techniques. International Journal of HumanComputer Studies, 41, 801-849.
Cooper, A. (1999). The Inmates are Running the Asylum. Indianapolis IN: SAMS, a division of Macmillan Computer Publishing.
December, J. (1997). Notes on defining of computer-mediated communication. Computer-Mediated
Communication Magazine, 3(1).
Dekker, A. H. (2002). A Category-Theoretic Approach to Social Network Analysis. Proceedings of Computing: The Australasian Theory Symposium (CATS), Melbourne, Australia, 28 Jan to 1 Feb 2002.
Donath, J., Karahalios, K., & Vigas, F. (1999, 8 May 2008). Visualizing conversation. Paper presented
at the Proceedings of the 32nd Annual Hawaii International Conference.
Fahy, P. J. (2003). Indicators of support in online interaction. International Review of Research in Open
and Distance Learning, 4(1).
Fahy, P. J., Crawford, G., & Ally, M., (2001). Patterns of interaction in a computer conference transcript.
International Review of Research in Open and Distance Learning, 2(1).
Ferris, P. (1997) What is CMC? An Overview of Scholarly Definitions. Computer-Mediated Communication Magazine, 4(1).

liv

Fortner, R. S. (1993). International communication: History, conflict, and control of the global metropolis. Belmont, CA: Wadsworth.
Garton, L., Haythorthwaite, C., & Wellman, B. (1997). Studying On-line Social Networks. In Jones, S.
(Eds.), Doing Internet Research. Thousand Oaks CA: Sage.
Greenberg, S. (1998). Prototyping for design and evaluation. Retrieved on November 30, 2004 at http://
pages.cpsc.ucalgary.ca/~saul/681/1998/prototyping/survey.html
Gunawardena, C., Lowe, C., & Anderson, T. (1997). Analysis of a Global Online Debate and the Development of an Interaction Analysis Model for Examining Social Construction of Knowledge in Computer
Conferencing. Journal of Educational Computing Research, 17(4), 397-431.
Hall, R. R. (2001). Prototyping for usability of new technology. International Journal.
Hanneman, R. A. (2001). Introduction to Social Netwok Methods. Retrieved on November 9, 2004 from
http://faculty.ucr.edu/~hanneman/SOC157/TEXT/TextIndex.html
Heeren, E. (1996). Technology support for collaborative distance learning. Doctoral dissertation, University of Twente, Enschede.
Henri, F. (1992). Computer Conferencing and Content Analysis, In A. R. Kaye (Ed), Collaborative learning through computer conferencing: The Najaden Papers, (pp. 117-136). Berlin: Springer-Verlag.
John, B. E., & Marks, S. J. (1997) Tracking the Effectiveness of Usability Evaluation Methods. Behaviour and Information Technology, 16(4/5), 188-202.
Jones, S. (1995). Computer-Mediated Communication and Community: Introduction. Computer-Mediated Communication Magazine, 2(3).
King, N., Ma, T.H.Y., Zaphiris, P., Petrie, H., Fraser, H. (2004). An incremental usability and accessibility
evaluation framework for digital libraries. In Brophy, P., Fisher S., Craven, J. (2004) Libraries without
Walls 5: The distributed delivery of library and information services. London, UK: Facet Publishing.
Korzenny, F. (1978). A theory of electronic propinquity: Mediated communication in organizations.
Communication Research, 5,3-23
Krebs, V. (2004). An Introduction to Social Network Analysis. Retrieved November 9, 2004 from http://
www.orgnet.com/sna.html
Kuniavksy, M. (2003). Observing the User Experience. Morgan Kaufmann Publishers.
Lewis, J. R. (1995). IBM Computer Usability Satisfaction Questionnaires: Psychometric Evaluation and
Instructions for Use. International Journal of Human-Computer Interaction, 7(1), 57-78.
Linden Lab. (2003). Second life. Last retrieved 21 June 2007.
Macaulay, L. A. (1996). Requirements engineering. Springer Verlag Series on Applied Computing.
Madden, M., & Rainie, L. (2003). Pew Internet & American Life Project Surveys. Pew Internet &
American Life Project, Washington, DC.
Maiden, N. A. M. Mistry, P., & Sutcliffe, A. G. (1995). How people categorise requirements for reuse:
a natural approach. Proc. of the 2nd IEEE Symposium on Requirements Engineering, (pp. 148-155).

lv

Mander, R., & Smith, B. (2002). Web usability for dummies. New York: Hungry Minds.
Mason, R. (1991). Analyzing Computer Conferencing Interactions. Computers in Adult Education and
Training, 2(3), 161-173.
McLuhan, M. (1964). Understanding media: The extension of man. New York: McGraw Hill
Metcalfe, B. (1992). Internet fogies to reminisce and argue at Interop Conference. InfoWorld.
Millward, L. J. (1995). In G. M. Breakwell, S. Hammond and C. Fife-Shaw (Eds.), Research Methods
in Psychology. Sage.
NUA Internet Surveys (2004). Retrieved October 20, 2004, from http://www.nua.ie/surveys/how_many_
online/index.html
of Human-Computer Studies, 55, 4, 485-502.
Pfeil, U. (2007). Empathy in online communities for older people an ethnographic study. City University London.
Preece, J. (2000) Online Communities: Designing Usability, Supporting Sociability. Chichester, UK:
John Wiley and Sons.
Preece, J., & Maloney-Krichmar, D. (2003). Online communities: Focusing on sociability and usability.
London. In J. A. Jacko & A. Sears (Eds.), Handbook of human-computer interaction. Lawrence Erlbaum
Associates Inc.
Preece, J., Rogers, Y., & Sharp, H. (2002). Interaction Design: Beyond Human-Computer Interaction.
New York, NY: John Wiley & Sons.
Preece, J., Rogers, Y., Sharp, H., Benyon, D., Holland, S., & Carey, T. (1994). Human-Computer Interaction. Addison Wesley.
Reiser, R. A., & Gagn, R. M. (1983). Selecting media for instruction. Englewood Cliffs: Educational
Technology Publications.
Rettig, G. (1994). Prototyping for tiny fingers. Communications of the ACM, 37, (4), 21-27.
Rheingold, H. (1993). The Virtual Community: Homesteading on the Electonic Frontier. Reading: Addison-Wesley.
Rice, R. (1994). Network analysis and computer mediated communication systems. In S. W. J. Galaskiewkz
(Ed.), Advances in Social Network Analysis. Newbury Park, CA: Sage.
Rice, R. E., Grant, A. E., Schmitz, J., & Torobin, J. (1990). Individual and network influences on the
adoption and perceived outcomes of electronic messaging. Social Networks, 12, 17-55.
Rudd, J., Stern, K., & Isensee, S. (1996). Low vs. high fidelity prototyping debate. Interactions, 3(1),
76-85. ACM Press.
SCOTCIT. (2003). Enabling large-scale institutional implementation of communications and information technology (ELICIT). Using Computer Mediated Conferencing. Retrieved November 2, 2004 from
http://www.elicit.scotcit.ac.uk/modules/cmc1/welcome.htm
Scott, J. P. (2000). Social network analysis: A handbook. Sage Publications Ltd.

lvi

Shneiderman, B. (1992). Designing the user interface: Strategies for effective human-computer interaction. 2nd edition.Reading: Addison-Wesley.
Suler, J. (2004). The Final Showdown Between In-Person and Cyberspace Relationships. Retrieved
November 3, 2004 from http://www1.rider.edu/~suler/psycyber/showdown.html
Taylor, T. L. (2002). Whose game is this anyway? Negotiating corporate ownership in a virtual world.
Paper presented at the Computer Games and Digital Cultures Conference Proceedings., Tampere: Tampere University Press.
Turkle, S. (1995). Life on the screen: Identity in the age of the Internet. New York: Simon & Schuster.
U.S Census Bureau (2004). Global Population Profile 2002. Retrieved October 20, 2004 from http://
www.census.gov/ipc/www/wp02.html
Usability Net. (2003). UsabilityNet. Retrieved on December 3, 2004 from http://www.usabilitynet.
org/
Wallace, P. (1999). The Psychology of the Internet. Cambridge: Cambridge University Press.
Walther, J. B. (1995). Relational aspects of computer-mediated communication: Experimental observations over time. Organization Science, 6(2), 186-203.
Wellman, B. (1992). Which types of ties and networks give what kinds of social support? Advances in
Group Processes, 9, 207-235.
Wellman, B. (1982). Studying personal communities. In P. M. N Lin (Ed.), Social Structure and Network
Analysis. Beverly Hills, CA: Sage.
Wellman, B., & Gulia, M. (1999). Net surfers dont ride alone: Virtual communities as communities. In
B. Wellman (Ed.), Networks in the global village (pp. 331-366.). Boulder, CO: Westview Press.
Zaphiris, P., Kurniawan, S. H. (2001). Using Card Sorting Technique to Define the Best Web Hierarchy
for Seniors. Proc. of CHI 2001 Conference on Human Factors in Computing Systems. ACM Press.
Zaphiris, P., Zacharia, G., & Rajasekaran, M. (2003). Distributed

Constructionism through Participatory

Design. In Ghaoui, C. (Eds.) E-education Applications: Human Factors and Innovative Approaches.
London, UK: Idea Group Publishing.
Zhu, E. (1996). Meaning negotiation, knowledge construction, and mentoring in a distance learning
course. In Proceedings of Selected Research and Development Presentations at the 1996 National Convention of the Association for Educational Communications and Technology. Indianapolis, IN. (ERIC
Document Reproduction Service No. ED 397 849).

lvii

About the Editor

Panayiotis Zaphiris is a reader in human-computer interaction at the Centre for HCI Design. He
got his B.Sc and M.Sc. from the University of Maryland College Park (USA). Panayiotis got his PhD
(in human-computer interaction) from Wayne State University where he was also a research assistant
at the Institute of Gerontology. Panayiotis research interests lie in HCI with an emphasis on inclusive
design and social aspects of computing. He is also interested in Internet related research (Web usability, mathematical modelling of browsing behaviour in hierarchical online information systems, online
communities, e-learning, Web based digital libraries and finally social network analysis of online human-to-human interactions). He has authored over 110 papers and four books in this area including a
recently released book titled User-Centered Design of Online Learning Communities (Lambropoulos
and Zaphiris, 2006). He is principal investigator for the AW-Model EPSRC/BBSRC project funded
under the SPARC scheme. Panayiotis was the principal investigator on the JISC-funded Usability
Studies for JISC Services and Information Environment and Information Visualisation Foundation
Study projects which looked at usability and information visualisation in the context of online learning
environments. Finally, he was the co-principal investigator on the largest ever conducted Web accessibility study (the DRC funded An in-depth study of the current state of Web accessibility project) that
assessed at the accessibility of 1000 UK Web sites. More information about Panayiotis research can be
found at: http://www.zaphiris.org
Chee Siang Ang is a research fellow in the Centre for Human-Computer Interaction (HCI) Design.
He is interested in human interactions and social tendencies in the virtual world, particularly Second
Life, from a sociological, psychological and HCI perspective. His main research interests include the
psychology and sociology of computer games, virtual worlds or 3D Computer-Mediated Communication (CMC), learning theories particularly in gaming, digital media such as interactive narrative and
simulation

Section I

Fundamental Concepts
and Theories
This section serves as the foundation for this exhaustive reference tool by addressing crucial theories essential to the understanding of human-computer interaction. Chapters found within these pages provide
an excellent framework in which to position human-computer interaction within the field of information
science and technology. Individual contributions provide overviews of ubiquitous computing, cognitive
informatics, and sociotechnical theory, while also exploring critical stumbling blocks of this field. Within
this introductory section, the reader can learn and choose from a compendium of expert research on the
elemental theories underscoring the research and application of human-computer interaction.

Chapter 1.1

Introduction to Ubiquitous
Computing
Max Mhlhuser
Technische Universitt Darmstadt, Germany
Iryna Gurevych
Technische Universitt Darmstadt, Germany

A Brief History of Ubiquitous

Computing
Mark Weiser
The term ubiquitous computing was coined and
introduced by the late Mark Weiser (1952-1999).
He worked at the Xerox Palo Alto Research Center (PARC, now an independent organization).
PARC was more or less the birthplace of many
developments that marked the PC era, such as
the mouse, windows-based user interfaces, and
the desktop metaphor (note that Xerox STAR
preceded the Apple Lisa, which again preceded
Microsoft Windows), laser printers, many concepts of computer supported cooperative work
(CSCW) and media spaces, and much more. This
success is contributed (among other reasons) to
the fact that PARC managed to integrate technology research and humanities research (computer
science and human factors in particular) in a

truly interdisciplinary way. This is important

to bear in mind since a considerable number of
publications argue that the difference between UC
and Ambient Intelligence was the more technology/networks-centered focus of the former and
the more interdisciplinary nature of the latter
that considered human and societal factors. We
do not agree with this argument, in particular
due to the nature of the original UC research at
PARCand the fact that quite a number of UC
research labs worldwide try to follow the PARC
mindset. Indeed, Mark Weiser concentrated so
much on user aspects that quite a number of his
first prototypes were mere mockups: during corresponding user studies, users had to imagine the
technology side of the devices investigated and
focus on use cases, ideal form factors and desired
features, integration into a pretend intelligent
environment, and so forth.

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Introduction to Ubiquitous Computing

Weisers Vision of UC
Mark Weisers ideas were first exposed to a large
worldwide audience by way of his famous article
The Computer of the 21st Century, published in
Scientific American in 1991. A preprint version
of this article is publicly available at: http://www.
ubiq.com/hypertext/weiser/SciAmDraft3.html.
Maybe the most frequently cited quotation
from this article reads as follows: The most profound technologies are those that disappear. They
weave themselves into the fabric of everyday life
until they are indistinguishable from it. This was
Marks vision for the final step in a development
away from standard PCs, towards a proliferation
and diversification of interconnected computerbased devices. A deeper understanding of Mark
Weisers visions can be drawn from his position
towards three dominant, maybe overhyped trends
in computer science at his time: virtual reality,
artificial intelligence, and user agents. With a
good sense for how to raise public attention,
Mark criticized these three trends as leading in
the wrong direction and positioned UC as a kind
of opposite trend. We will follow Marks arguments for a short while and take a less dramatic
view afterwards.

UC vs. Virtual Reality (VR)

According to Mark, VR brings the world into the
computer, whereas UC brings the computer into
the world. What he meant was that VR technology is generally based on elaborate models of an
existing or imagined (excerpt of the) world. This
model contains not only 3D (geometric) aspects
but many more static and dynamic descriptions of
what is modeled. For instance, digital mockups of
cars have been pushed to the point of simulating
crash tests based on the car /obstacle geometry,
static, and dynamic material characteristics,
laws of physics, and so forth. As the sophistication of models grows, more and more aspects of
the world are entered into the computer, finally

almost everything happens in the virtual space

and even the human becomes a peripheral device
for the computer, attached via data gloves and
head-mounted displays. Mark Weiser criticized
mainly the central and peripheral roles of computers and humans, respectively. He proposed
to follow the UC vision in order to invert these
roles: by abandoning the central role of computers
and by embedding them in the environment (in
physical objects, in particular), room is made for
the human in the center. In this context, he used
the term embodied virtuality as a synonym for
UC. The cartoons in Figure 1 were made by Mark
Weiser and provided by courtesy of PARC, the
Palo Alto Research Center, Inc.

UC vs. Artificial Intelligence (AI)

In essence, Mark Weiser criticized the overly high
expectations associated with AI in the 1980s. In
the late 1980s and early 1990s, that is, at the time
when he developed his UC vision, AI research
had to undergo a serious confidence crisis. The
term AI had not been associated with a commonly
accepted, reasonably realistic definition, so that
the association with human intelligence (or the
human brain) was destined to lead to disappointments. The AI hype had provided researchers with
considerable fundsbut only for a while. Mark
Weiser proposed to take a different approach
towards a higher level of sophistication of computer-based solutions (which had been the goal of
AI at large). He considered it a more reasonable
objective to concentrate on small subsets of intelligent behavior and to dedicate each computer
to such a subset. Higher sophistication would be
fostered by interconnecting the special-purpose
computers and by making them cooperate. This
reasoning lead to the term smart, considered
more modest than the term intelligent. Sensor
technology plays an important role in dedicating
computers to a small subset of understanding
the world around us (a key element of intelligent
behavior). By widely deploying and interconnect-

Introduction to Ubiquitous Computing

Figure 1. Mark Weisers cartoons about UC vs. virtual reality

ing sensor-based tiny computers, one would be

able to integrate environmental data (location,
temperature, lighting, movement, etc.) and use
this information to produce smart behavior of
computers and computerized physical objects.

2.

UC vs. User Agents (UA)

In contrast to virtual reality and artificial intelligence, the term user agent is not very prominent
in the general public. At the time referred to,
UAs were thought as intelligent intermediaries
between the user and the computer world, that is,
as an approach towards increased ease-of-use or
better human-computer interaction. User agents
were often compared to the common perception
of British butlers who are very discreet and unobtrusive, but always at disposal and extremely
knowledgeable about the wishes and habits of
their employers. Following this analogy, UAs
were installed as autonomous software components between applications and users, inspecting
and learning from the user-software application.
Mark Weiser challenged five requirements usually derived from this analogy for user agents and
proposed UA as a better alternative for the first
three; as to the last two, he judged the necessary
base technology as immature:
1.

UAs were supposed to give advice to their

users based on what they had learned. Mark

3.

4.

5.

Weiser asked, in essence, why they would

not do the job themselvesa promise that
UC should fulfill;
UAs were supposed to obey the user, for
example, by applying planning algorithms
to basic operations with the aim to fulfill
the goals set by a user. In contrast to this
approach, UC was intended to behave rather
proactively, that is, to propose and even act
in advance as opposed to reacting on command;
A third widespread requirement suggested
that UAs should intercept the user-application interface. UC in contrast should be
more radical and take over the interaction
or carry out functions on its ownan approach presumed by Mark Weiser to be the
only viable one if humans were to be surrounded by hundreds of computers;
A basic assumption about UAs was that
they would listen to the (interactions of)
the user. Mark Weiser considered natural
language processing technology and speech
recognition technology at his time to be far
too immature to promise satisfying results
in this respect;
UAs should learn the users preferences,
wishes, and so forth by observation. Again,
the necessary (machine learning) technology was judged to be too immature to live
up to this promise.

Introduction to Ubiquitous Computing

We will resume the VR / AI / UA discussion

in the next large section.

Mark Weisers Three Key Devices

We want to finish this lengthy but still extremely
compressed, much simplifying and abstracting
treatment of Mark Weisers contributions by
looking at three devices. These complementary
UC devices were prototyped at his lab; investigated in the context of PARCs typical creative,
team-oriented setting, all three were thought as
electronic replacements for the common analog
information appliances.
The Xerox Pad can be considered to be the
prototype and father of present PDAs, introduced
even before Apples Newton appeared in 1993.
The initial concept was that of an electronic
equivalent to inch-size information bearers,
namely PostIt Notes: easy to create and to stick
almost everywhere, available in large quantities.
As the PDA analogy suggests, the prototypes had
a lot more functionality than PostIt Notesbut
were also a lot more expensive and cumbersome
to handle by design (not only due to short and
mid-term technology limitations).
The Xerox Tab can be considered to be the
prototype and father of present Tablet PCs. The
analogy from the traditional world was that of a
foot-size information bearer, namely a notebook
or notepad. One may infer from the rather stalling
market penetration of Tablet PCs that technology
is still not ready for mass market Tabs today,
but one may also expect to find a pen centric,
foot size, handheld computer to become very successful any time soon. An interesting facet of the
original Tab concept was the idea that Tabs would
in the future lay around for free use pretty much
as one finds paper notebooks today, for example,
as part of the complementary stationery offered
to meeting participants.
The Xerox Liveboard was the prototype of
present electronic whiteboards. A PARC spinoff
company designed and marketed such boards,

and today many companies like Calgary-based

SmartTechnologies Inc. still sell such devices.
Liveboards represented the yard-size information bearers in the family of cooperating devices
for cooperating people. In contrast to many devices
sold today, Liveboards supported multi-user input
pretty early on.
The developments and studies conducted at
Mark Weisers lab emphasized the combination
of the three device types for computer supported
cooperation, and cooperative knowledge work in
particular.
While Mark Weiser was a truly outstanding
visionary person with respect to predicting the
future of hardware, that is, UC nodes (proliferation of worn and embedded networked devices,
specialized instead of personal general-purpose
computers, numbers by far exceeding the number
of human users), two other people were more
instrumental in generating awareness for the two
remaining big challenges mentioned in the preface
of this book, namely integrative cooperation and
humane computing; the former of these challenges
was emphasized by Kevin Kelly, the latter by Don
Norman. A deeper analysis reveals that for the
second aspect, humane computing, it is very difficult to argue about the true protagonists. Readers
remember that Mark Weiser was actually placing
a lot of emphasis on usability, by virtue of his
education and mindset and in the context of the
human focus of PARC. He also coined the exaggerated term invisible for mature technology. On
the other hand, Don Norman was not advocating
the humane computing challenge in all its facets
yet. Nevertheless, we want to highlight him next
as maybe the single most important advocate of
this challenge.

The Book Out of Control by Kevin

Kelly
In 1994, K. Kelly published a book entitled Out
of Control. The thoughts expressed by Kelly
were an excellent complement to Mark Weisers

Introduction to Ubiquitous Computing

publications. While the latter emphasized the

emergence of networked small neuron like (i.e.,
smart) UC nodes, Kelly emphasized the integrated
whole that these neurons should form. His starting
argument was the substantiated observation that
the complexity of the made, that is, of humanmade systems or technology, approached the
complexity of the born, that is, of nature-made
systems, such as human or biological organisms,
human or biological societies (cf. ant colonies),
and so forth.
This observation led to the obvious requirement to investigate the intrinsic principles and
mechanisms of how the born organized, evolved,
and so forth. By properly adopting these principles
to the made, this complexity might be coped
with. Research about the organization and evolution of the born should be particularly concerned
with questions such as: how do they cope with
errors, with change, with control, with goals, and
so forth. For instance, beehives were found not to
follow a controlling head (the queen bee does not
fulfill this function), and it is often very difficult
to discern primary from subordinate goals and
to find out how goals of the whole are realized
as goals of the individuals in a totally decentralized setting.
Kevin Kelly summarizes central findings and
laws of nature several times with different foci.
Therefore, it is not possible to list and discuss
these partly conflicting findings here in detail.
An incomplete list of perceived central laws of
God reads as follows: (1) give away control:
make individuals autonomous, endow them with
responsible behavior as parts of the whole, (2)
accept errors, even build it in as an essential
means for selection and constant adaptation and
optimization, (3) distribute control truly, that is, try
to live with no central instance at all, (4) promote
chunks of different kinds (e.g., hierarchies) for
taming complexity, and (5) accept heterogeneity
and disequilibrium as sound bases for survival.

The Book The Invisible Computer by

Donald Norman
Don Norman emphasized the humane computing grand challenge described in the preface
of this book. World renowned as an expert on
usability and user-centered design, he published
The Invisible Computer in 1999. He considered
the usability problems of PCs to be intrinsically
related to their general-purpose nature and thus
perceived the dawning UC era more as a chance
than a risk for humane computing. The intrinsic
usability problems that he attributed to PCs were
rooted in two main anomalies, according to Don
Norman: (1) PCs try to be all-purpose and all-user
devicesa fact that makes them overly complex,
and (2) PCs are isolated and separated from daily
work and life; truly intuitive usein the context
of known daily tasksis therefore hardly possible. From this analysis, Norman derived various
design guidelines, patterns, and methodological
implications, which we will summarize again at
an extremely coarse level:
1.

2.

3.

4.

He advocated UC nodes using the term

information appliances: dedicated to a
specific task or problem, they can be far
simpler and more optimized;
He further advocated user-centered development: especially with a specific user
group in mind, information appliances as
described previously can be further tailored
to optimally support their users;
Norman stated three key axioms, that is,
basic goals to be pursued during design and
development: simplicity (a drastic contrast
to the epidemic featurism of PC software),
versatility, and pleasurability as an often
forgotten yet success critical factor;
As a cross-reference to the second big UC
challenge (integrative cooperation), he advocated families of appliances that can
be easily and very flexibly composed into
systems.

Introduction to Ubiquitous Computing

History Revised
The preceding paragraphs are important to know
for a deeper understanding of the mindset and roots
of UC. However, about 15 years after the time when
the corresponding arguments were exchanged, it
is important to review them critically in the light
of what has happened since. We will first revise
the three religious disputes that Mark Weiser
conducted against AI, VR, and UAs. To put the
bottom line first, the word versus should rather
be replaced by and today, meaning that the
scientific disciplines mentioned should be (and
have, mostly) reconciled:
As to UC and VR, specialized nodes in a global
UC network can only contribute to a meaningful
holistic purpose if models exist that help to cooperatively process the many specialist purposes
of the UC nodes. In other words, we need the
computer embedded into the world and the world
embedded in the computer. Real Time Enterprises
are a good example for very complex modelsin
this case, of enterprisesfor which the large-scale
deployment of UC technology provides online
connectivity to the computers embedded into the
world, that is, specialized nodes (appliances, smart
labels, etc.). In this case, the complex models are
usually not considered VR models, but they play
the same role as VR models in Mark Weisers
arguments. The progress made in the area of
augmented reality is another excellent example
of the benefit of reconciliation between UC and
VR: in corresponding applications, real-world
vision and virtual (graphical) worlds are tightly
synchronized and overlaid.
As to UC and AI, Mark Weiser had not addressed the issue of how interconnected, smart,
that is, modest, specialized nodes would be
integrated into a sophisticated holistic solution. If
the difference between AI and the functionality of
a single smart UC node (e.g., temperature sensor)
was comparable to the difference between a brain
and a few neurons, then how can the equivalent
of the transition (evolution) from five pounds of

neurons to a well-functioning brain be achieved?

Mark Weiser did not have a good answer to that
questionsuch an answer would have sounded
like AI anyway.
Today, there is still not a simple answer yet. The
most sophisticated computer science technology
is needed in order to meet the integration challenge of how to make a meaningful whole out of
the interconnected UC nodes. However, the state
of the art has advanced a lot and our understanding for what can be achieved and what not (in
short term) has improved. For instance, socionic
and bionic approaches have become recognized
research areas. A mature set of methods and
algorithms is taught in typical Introduction to
AI classes today and has replaced the ill-defined,
fuzzy former understanding of the area. Thus the
boundaries between AI and computer science are
more blurred than ever and their discussion is left
to the public and press.
As to UC and UAs, remember that Mark Weiser
considered UAs as too little in terms of what
they attempted (at least too little for the UC world
envisioned by him), yet too much in terms of
what the underlying technology was able to provide. This left doubts about how the even more
ambitious goals of UC could be met, namely active
(proactive, autonomous, even responsible) rather
than reactive (obeying) behavior. In other words,
Mark Weiser was right when he advocated active
as opposed to reactive behavior, but he had little
to offer for getting there. Luckily, the technologies that he had then considered immature (e.g.,
speech processing, NLP, machine learning) have
advanced a lot since.
All in all, Mark Weisers arguments from 15
years ago (1) provide a deep understanding of the
field, (2) should be modified towards a more conciliatory attitude (in particular with respect to AI
and VR / complex world models), and (3) have
become more substantiated in certain respects
since technology advancements make some of his
more audacious assumptions more realistic (but
most visions of his opponents, too). In other

Introduction to Ubiquitous Computing

words, Mark Weisers visions were and still are

marking the research and developments made by
the UC community. His concepts and predictions
were accurate to a degree that was hardly paralleled by any other visionary person. Restrictions
apply as to his overly drastic opposition to VR,
AI, and UAs: some of the exaggerated promises
of these were repeated by him in the UC context
- right when he denounced the over-expectations
raised by AI and UAs! VR and AI in particular
should be reconciled with UC. Maybe Weiser
underestimated the two grand challenges of the
UC era, namely integrative cooperation and
humane computing.
Kevin Kelly and Donald Norman emphasized
these two challenges, respectively. Looking at the
advancements in totally decentralized systems,
Kellys promises can be evaluated as too extreme
today: bionics social science inspired, and autonomic (or autonomous) computing have advanced
a lot. However, two restrictions still apply: (1) less
decentralized systems still prove to be extremely
viable in daily operationit will be hard for
fully decentralized systems to really prove their
superiority in practice; (2) system-wide goals must
still be planned by some centralized authority
andto a certain extent manuallytranslated
into methods for fully decentralized goal pursuit;
evolution-like approaches that would generate
optimization rules and their pursuit automatically
in a fully decentralized systems are still hardly
viable. As a consequence, the present book will
not only describe the above-mentioned computing
approaches in part Scalability, but also other
aspects of scalability.
As to Don Norman, he was right to advocate simplicity as a primary and key challenge.
However, he maybe underestimated the humane
computing problems associated with the nomadic
characteristics and integrative cooperation
challenge of the UC era. The usability of the
integrated whole that we advocate to build out
of UC nodes is by far not automatically endowed
with easy-to-use user interaction just because the

participating appliances exhibit a high degree of

usability. On the other hand, only the integration
that is, federation of miniature appliances with
large interaction devices (wall displays, room
surround sound, etc.) may be able to provide the
usability desired for an individual device.
As we conclude this section, we should not
forget to mention that the UC era was of course
not only marked by just three visionary people.

Terms and Selected

Standards
While there is a lot of agreement among researchers and practitioners worldwide that the third era
of computing is dawning as the era of networked,
worn/portable and embedded computers, there
is not so much agreement about what to call that
era. This fact is something of an obstacle, for
instance for wider recognition in politics (the
crowd does not scream the same name as one
may put it). This situation is aggravated by the
fact that partial issues and aspects of Ubiquitous
Computing are also suffering from buzzword
inflation. With this background in mind, one may
understand why we list a considerable number
of these buzzwords below and provide a short
explanation, rather than swapping this issue out
into a glossary alone. Knowledge of the following
terms is indeed necessary for attaining a decent
level of UC literacy.

Synonyms for Ubiquitous

Computing
First, we want to look at the terms that describemore or lessthe third era of computing
as introduced:

Post-PC era: The root of this term is obvious, it describes the era that comes after
the second, that is, the PC era. We suggest
avoiding this term since it points at what it

Introduction to Ubiquitous Computing

is not (PCs) rather than at what it actually

is.
Pervasive computing: A distinction between the word ubiquitous and pervasive
is difficult if not artificial. One could argue
that the term pervasive eludes more to the
process of penetration (i.e., to the verb pervade) whereas ubiquitous eludes more to
the final state of this process. We suggest
that pervasive computing and ubiquitous
computing are synonyms, one (pervasive)
being slightly more common in industry (its
origin has been attributed to IBM), the other
one (UC) being slightly more common in
academia.
Ubiquitous computing: The term may be
interpreted as computers everywhere. We
are using it as the notion for the third era of
computing throughout the book and prefer
it, among others, because we try to fight
buzzword mania and dislike the invention
of additional terms for a named concept.
We therefore propose to stick to the first
(reasonable) term invented and somewhat
broadly accepted; since Mark Weiser is the
first visionary person who sketched essential
characteristics of the dawning era and since
he invented the term UC, the question of
what is the oldest well-known term should
not be questionable.
Ambient intelligence: This term was invented in particular in the context of the European
Unions research framework programs (5, 6,
7). As a positive argument, one may say that
the two words reflect the grand challenges
of UC as stated in this book: ambient may
be associated with the challenge of humane
computing, making UC systems an integral
part of our daily life. Intelligence may be
interpreted as the challenge of integrative
cooperation of the whole that consists of
myriads of interconnected UC nodes. On the
downside, one should remember that Mark
Weiser had intentionally avoided the term

intelligence due to the over-expectations

that AI had raised. We suggest avoiding this
term, too, because it is still burdened with
these over-expectations and because it is
still ill defined.
Disappearing / invisible / calm computing: All three terms are less common than
UC and pervasive computing. Their roots
have been discussed in the historical context
above. Obviously, disappearing describes
again a process while invisible describes
a final state. Calm emphasizes hearing as
opposed to vision like the other two. In any
case, the terms invisible and disappearing are not very well chosen (despite our
tribute to Don Norman) since computers
and interfaces that have totally disappeared
cannot be commanded or controlled by humans any more. Since we doubt that 100%
satisfactory service to the user can be paid
at all without leaving the customer, that is
the user, the option to explicitly influence
the service behavior, we consider the term
misleading. We favor again Mark Weisers
notion of computers that are so well interwoven with the fabric of our lives that we
hardly notice them.
Mixed-mode systems: This is a term used
to describe the heterogeneity of UC nodes, in
contrast to the rather resource rich, general
purpose PCs of the last era. This term is
even less common, but pops up every now
and then like those previously discussed,
and should not be used to describe UC as
a whole since it emphasizes a particular
aspect.
Tangible bits: This term has found some
currency in the Netherlands and Japan, but
remained rather uncommon in general. It
refers mainly to the fact that networked
computers are becoming part of the physical
world.
Real time enterprise: This term has been
explained in the preface of the book and is not

Introduction to Ubiquitous Computing

thought as a synonym for UC, but rather as a

very important and cutting-edge application
domain that may drive down the learning
curve, that is, prices of UC hardware and
solutions.
It was mentioned in the preface that some authors argued in favor of one or the other of the UC
synonyms, saying that their choice was more farreaching in time (the other ones being intermediate
steps) or space (the other ones only comprising a
subset of the relevant issues). However, we cannot
follow these arguments, mainly because research
labs and projects around the world work on the
same subjects, some more advanced or holistic,
some less ambitious or more specialized, carrying
the names UC, pervasive computing, and ambient
intelligence rather randomly.

Towards a Taxonomy of UC Nodes

Throughout this book, UC nodes will be categorized according to different aspects. In the
context of reference architectures further below,
we will emphasize the role of UC nodes in a holistic picture. In the present paragraph, we want
to try categorizing them as devices. It should be
noted that in the preface of the book, we already
provided a preliminary, light weight introduction.
The difference between carried (worn, portable)
and encountered nodes was emphasized and four
preliminary categories (wearables, sensors, appliances, and smart labels) were briefly described.
It soon became clear that smart labels attached
to goods must be distinguished again from those
attached to humans, although the base technology
may be the same.
In a second, more serious attempt to categorize
UC nodes as device categories, we propose the
following distinction (see Figure 2):
1.

Devices attached to humans

a. Devices carried: Here we further distinguish three subcategories: (1) mobile

2.

devices, synonymous with portable

devices, contain rather general purpose
computers and range from laptops via
PDAs to mobile phones and the like,
(2) smart badges, that is, smart labels
serve for identification, authentication
and authorization of humans and possibly further purposes, and (3) body
sensors of all kinds play an increasingly important role in particular in
the fitness and health context;
b. Devices worn: These wearables range
from truly sophisticated, computeraugmented cloths and accessories to
prototypes that are built from standard
components (PDA in a holster with
headset, etc.). A further categorization
is not attempted since the spectrum is
rather blurred;
c. Devices implanted: while there is
a lot of hype about implanted RFID
tags and networked health implants,
the many issues (e.g., health, privacy,
or dependability) around the necessary
device-environment communication
have not permitted this category to
become widespread.
Devices encountered
a. Smart items denote computer-augmented physical objects. The terms
smart object and smart product
are used with subtle differences depending on the context to denote more
sophisticated variants of smart items,
such as smart items that proactively
communicate with the users. We suggest treating smart items as the most
general term and to distinguish the
following subcategories: (1) smart
tags as the least sophisticated variant:
they can be considered to be mimicry
for embedded computers: by attaching a smart tag to a physical object,
a physically remote computer (often

Introduction to Ubiquitous Computing

Figure 2. Taxonomy of UC devices

b.

in proximity, though) can take over

some of the functionality that would
be embedded otherwise. This approach
opens the door for turning even the
cheapest products into UC nodes. The
term smart label is sometimes used
synonymously; sometimes it is used
as the comprehensive term for smart
tags and smart badges (attached to
humans, see earlier discussion). We
suggest sticking to the term smart
tag for the smart item sub-category
described here; (2) networked sensor
nodes, and (3) networked appliances
denote the other subcategories of smart
items. They were already introduced
in the preface of this book.
Smart environments denote the
surroundings of smart items, that is,
the additional communication and
compute power installed in order to
turn an assembly of smart items into
a local, meaningful whole.

The reader must be aware that all terms arranged in the taxonomy are not settled yet for a
common understanding. For instance, one might
argue whether a sensor network that computes
context information for networked appliances and
users should be considered a set of smart items

10

(as we defined it) or a part of the smart environment. Nevertheless, we find it useful to associate
a well-defined meaning with these terms and to
apply it throughout the book (see Figure 2).
In addition, it should be noted that smart
environments (with integrated smart items)
constitute a particularly important research
areamaybe because they permit researchers
and project leaders to implement self-contained
little UC worlds without a need for multiparty
agreements about interoperability standards. In
particular, smart homes were among the first
subjects of investigation in the young history of
UC. Prestigious projects in the smart home area
were and are conducted by industry (Microsoft
eHome, Philips AmbientIntelligence initiative,
etc.) and academia (GeorgiaTech AwareHome,
MIT House, etc.). HP made an early attempt
to overcome the isolation of such incompatible
islands by emphasizing standard middleware in
the Cooltown project). Quite a number of projects
about smart homes terminated without exciting
results, not to the least due to insufficient business
impact (note our argument in favor of Real Time
Enterprises as a more promising subject). More
recently, smart homes projects have focused on
issues considered to be particularly promising,
as was discussed in the preface to this book.
Important areas comprise home security, energy
conservation, home entertainment, and particu-

Introduction to Ubiquitous Computing

larly assisted living for the aging societya topic

considered particularly interesting in Europe (1
year prolongation of independepnt living saving
about half a billion Euros in Germany alone). Renowned large-scale projects were carried out, for
example, in Zwijndrecht (Belgium) and Tnsberg
(Norway) in this respect.

A Few More Relevant Terms

A few more UC termsand sometimes, corresponding conceptsare worth mentioning.

Smart dust is a term used for sensor networks

if the emphasis is on miniaturization and the
concept is based on one-time deployment
and zero maintenance. Environment data
sensors are often cited as an example, the
vision then is to deploy them, for instance,
from an aircraft, and let them monitor the
environment until they fail. Environmentfriendly degradation is a major issue here,
of course.
Things that think was the name of an early
UC project led by Nicholas Negroponte at
the MIT media lab. Other authors have since
hijacked the term.
Smart paper denotes the vision of a display
device that would exhibit characteristics
comparable to traditional paper in terms of
weight, robustness, readability, and so forth,
and loadable with the content of newspapers,
journals, books and so forth,. it would help
to save paper and revolutionize the press
distribution channels and more. Many projects that were not even close to this vision
had, and continue to have, the name smart
paper.
Smart wallpaper is a similar term to smart
paper in that it extrapolates the above mentioned characteristics to wall-size devices.
Smart <you-name-it>: virtually every
noun has been associated with the attribute
smart recently, not always alluding to the

characteristics of UC nodes. For instance,

smart materials are supposed to adapt to the
context of use, with no IT involved. Most of
the time though, smart <something> alludes
to a physical object that has been augmented
with an embedded computer.
The Internet of things is a term favored by
the press. It is not considered appropriate
as a term for UC as a whole by the authors
since it emphasizes the hardware side of UC
as opposed to the human side, which was
already described as crucial and as a major
challenge (cf. humane computing). Most
publications that favor this term concentrate
on the two standards discussed in the following section.

The EPCglobal Standard

As mentioned at the beginning, we will only
sketch two important standards in the UC context. Other standards are too unimportant, too
immature, too specific (they might be treated in
one of the focused parts of this book), or part of
the background knowledge about well-established
technology that this book cannot cover. The first
standard to mention is EPCglobal and was mentioned in the preface of this book. As mentioned,
it is meant to succeed the barcodes that encode
the European article number or universal product
code on current consumer products. The 96-bit
Electronic Product Code EPC is usually stored
on RFIDs (a subcategory of smart tags, as we can
now say) and can be read:

From a greater distance (e.g., 10m)

With better reading accuracy
With much less effort (e.g., en-passant by a
RFID reader gate as opposed to carefully
with line-of-sight connection by a barcode
scanner)
In bulk (RFID readers can read, for example,
a hundred tags at once)

11

Introduction to Ubiquitous Computing

Since the EPC contains a 36-bit serial number,

individual items can be tracked and traced. For
instance, theft can be much more easily attributed
to criminals, product life cycles can be recorded
more accurately, product lots with manufacturing
errors can be called back more specifically, etc. On
the other hand, the serial number may in principle
be used to trace an individual, too, if she carries
around an RFID tagged product. This privacy
issue has raised many concerns in recent years
and amplified the decision of the whole sales and
retail industry to focus on tagging their containers, palettes, cases, etc., for a start. So-called item
level tagging is only envisioned for highly valuable
goods initially; it may enter the mass market when
tag prices and system costs have come down and
after settling the privacy issues.
Figure 3 depicts the functioning of EPC smart
tags in an overall IT infrastructure. In step 1, an
EPC code is read from a product. In the example,
each carton on the palette could contain a number of tagged products. The residual example
would then explain the action for just one such
tag. Usually prior to reading the tag, the system
has already searched and discovered servers capable of resolving certain ranges of EPC code.
Based on the results of this discovery process,
the appropriate resolution node, called an ONS
server, is asked to resolve the EPC code, that
Figure 3. RFID / EPC scheme

12

is to translate it into a global Internet address

where the relevant product information is actually
stored. The product information is encoded in a
standardized way, using the so-called product
markup language PML, an XML derivate.
The second generation of RFID tags introduced
in 2006 features improved bulk reading (hundreds
of tags simultaneously), size and cost improvements. Printable tags have become common:
these paper labels with embedded RFID chips can
be custom imprinted with custom human-readable
information. The chips themselves are not altered
in the printer and they come with pre-assigned
EPC codes from the manufacturer.

The OSGi Standard

The Open Services Gateway Initiative (OSGi) is
an industry driven nonprofit consortium. OSGi
standardized a Java virtual machine (JVM). This
JVM can be considered a standardized virtual
computer that runs on any real computer and is
capable of executing programs that are transmitted
to it, so-called bundles. OSGi standardizes not
only the format for bundles, but also the necessary protocols and procedures for authenticating
and authorizing senders of bundles, for replacing
and updating bundles (remote maintenance), for
discovering other bundles, and so forth. OSGi

Introduction to Ubiquitous Computing

bundles are particularly useful for controlling

the functionality of networked appliances. Possible use cases include SetTopBoxes, Vehicles
(note that car electronics today requires much
shorter maintenance cycles than the mechanical
parts, especially for software updates!), consumer
electronics, and so forth. As to smart homes, the
favored concept is that of a residential gateway
that is connected to the global Internet and receives updates for smart home appliances via
OSGi. The residential gateway may then forward
bundle updates and so forth to the relevant appliances if needed.
OSGi has a number of deficiencies. For instance, it is not considered to be very resource
effective. Nevertheless, it has tremendous impact
as a de facto standard for dealing with some of
the elementary aspects of coping with global UC
systems in a platform and vendor independent
way.

Reference Architectures for

Ubiquitous Computing
The Importance and Role of a
Reference Architecture
A sophisticated distributed infrastructure is
needed in order to make a myriad of networked
UC nodes communicate and cooperate. If interoperability is to take on a worldwide scale, means
for agreement among arbitrary participants must
be provided. Ideally, the move from isolated proprietary UC solutions to a world of cooperating
UC components is driven by so-called reference
architectures which establish several levels of
agreement: on level one, a common terminology
and conceptualization of UC systems is established in order for researchers and practitioners to
speak the same language and to work on the same
global UC vision. On the second level, a common
understanding of the ensemble and components of
a typical UC system is established, including the

potential roles of the components. On level three,

basic functional principles can then be agreed
upon. A fourth level is desirable but beyond the
scope of reference architectures, that is concrete
standards for intercomponent cooperation. This
level is discussed in the introduction to the part
Scalability.

Reference Architectures in a More

Realistic World
In reality, a worldwide common understanding and
corresponding standards have to be developed in a
struggle for the best solution. Real life has a large
impact on what becomes widespread. By real
life we mean breaking research results, industry
practice, experiences gained with proprietary
prototypes and realizations, user acceptance, and
not least business interests defended by global
industrial players. Nevertheless, the exercise
of proposing and refining reference architecturesin communication with the stakeholders
mentionedplays a key role in a struggle for
globally interoperable solutions. Here reference
architectures must be invented and published and
then consolidated and reiterated based on feedback
by the stakeholders.

Prominent Examples from the Past

The ISO reference architecture for open systems
interconnection (OSI) was developed in the 1970s
as an important step towards global networks. OSI
was very successful in that it led to a common
terminology and a common understanding of the
components of computer networks including their
roles. The fourth level aforementioned above: ISO
standards for communication protocol, were not
nearly as successful as the reference architecture
itself. Rather, the Internet protocols TCP and IP
took over almost the entire market. Nevertheless,
the OSI reference architecture was extremely
influential on the computer networking community as a whole and on the Internet in particular.

13

Introduction to Ubiquitous Computing

Component reference architectures take

a birds-eye view on the world addressed.
They define a number of cooperating components or rather component types, and
specify inter-component cooperation at a
certain level of detail. Again, a kind of art
of right-sizing exists: too few component
types do not really help to understand and
discern relevant roles and specializations
common to the world addressed, too many
component types lead to overly complex
architectures and problems in matching
reference and reality.

Another ISO reference architecture is ODP (open

distributed processing). It emphasizes complex
distributed systems and applications. An influential contribution of ODP is its support for different
viewpoints of various stakeholders. In particular,
ODP emphasized the importance of enterprise
modeling for application development. All too
often, applications are modeled and built with a
technology focus and thus neglect the (dynamically changing) organization they should support.
ODP addresses important issues, but came at a
time when distributed applications were usually
rather simple: ODP was considered overkill.

Layered Architectures vs.

Component Architectures

Although we focus on the Computer Networks

/ Distributed Systems aspects of UC in the remainder of this chapter, readers should note that the
entire book represents a holistic approach.

Before we introduce concrete reference architectures, it is worth recalling the two complementary
flavors:

14

Layered reference architectures serve as a

blueprint for layered software architectures.
Both arrange sets of functions into layers
that act as virtual machines: only the what
(provided functionality and how to access
it) must be known to users in higher layers,
whereas the internal how (realization)
is hidden and can be independently modified. The layer stack represents the range
from higher to lower function sets, where
higher means closer to what users and applications need and lower means closer
to what hardware provides. Strict variants
preclude higher layer components to access
lower layers except for the one immediately
below. Recent research has concentrated on
approaches for automatic, selective custom
configuration of the entire layer stack, according to the needs of applicationsthis
trend is important in the UC world where
dedicated, resource-poor UC nodes cannot
host fat all-purpose layers.

Why Component Reference

Architectures are Important for UC
The OSI reference architecture assumes a network
consisting of rather homogeneous nodes, namely
general-purpose computers with sufficient CPU
and memory capacity. Accordingly, a common
definition of a computer network reads as follows:
A computer network CN is a set of autonomous
nodes AN, each of which disposes of CPU(s) and
memory, plus a Communication Subsystem CSS
capable of exchanging messages between any of
the nodes: CN :== {AN} CSS.
In the definition, all nodes are created equal.
At a closer look, computer networks rely on four
mandatory constituents of nodes (ANs):
1. Communication capability: The capacity
of exchanging messages with other nodes
through the CSS.
2. Address: A unique identifier that can be
used to specify the recipient or sender of
messages.

Introduction to Ubiquitous Computing

3. Processor: A general purpose CPU.

4. Memory: Means for storingat leastincoming messages.
In a UC world, resource scarcity and the
special-purpose nature of many nodes are key
issues.
A holistic UC approach must scale from
servers to sensors and support the consideration
of smart labels etc. The definition of a UC node
must be different from the one abovethe four
constituents now read as follows:
1. Communication is mandatory, but may be
passive (cf. passive RFID tags)
2. Address is not necessarily a unique identifier;
for example, in a sensor network, a random
node out of a redundant set with identical
address may provide a certain functionality
3. Processor becomes an optional constituent
4. Memory becomes an optional constituent,
too
With the above modifications, not all nodes
are autonomous (ANs) any more.

Proposed UC Component Reference

Architectures
The definition introduces a first possibility for
distinguishing nodes as components of an application, that is, from the component architecture point
of view. However, it only discerns between existing versus missing fundamental characteristics.
More interesting is the aspect of different roles
that nodes can play in the networknot application specific roles, but fundamental roles in the set
of cooperating resources. Thus UC systems will
take on more complex node topologies than what
was considered in the eras of simple interprocess
communication and client-server computing.
In addition, a holistic approach needed for UC
systems raises issues such as security, which are

important when trying to find important node

types at different levels of granularity.
One of the first proposals for a UC component
reference architecture was made by the Fraunhofer
research institute FOKUS in Berlin. They did not
distinguish different node types that would assume different roles, but identified important roles
that each UC node may potentially assume. Their
concept is coined I-Centric Services and achieved
a certain level of influence on the industrial Object
Management Group (OMG). In their view, a UC
node (usually a software service) should provide
standard interfaces for four major issues:
1. Discovery of peers in a spontaneous, configuration-free manner
2. Maintainance, i.e., software update and
revision
3. Reservation, that is, pre-allocation of some
of the nodes resources as a basis for service
guarantees
4. Configuration as a means for customizing
the service for a dedicated role
Nodes that conform to these interfaces are
called super distributed objects (SDO) in this
proposal.
We will discuss another component architecture in some more detail since it attempts to
discern between more specific roles of UC nodes.
It was developed in the Telecooperation Group at
the Technische Universitt Darmstadt and is called
Mundo, see Figure 4. Mundo distinguishes five
different node types: Me, Us, It, We, and They.
Me (Minimal Entity): Mundo emphasizes the
importance of a distinct personal UC node, that
is the device tightly associated with its user: the
Me. Every user uses exactly only one Me at any
time. The rationale is rooted in the envisioned
ubiquity of computer support in everyday life: if
every step that one takes is potentially computer
supported and controlled, then humans need a
high level of trust that the computers do the right
thing. For instance, users will want to make sure

15

Introduction to Ubiquitous Computing

Figure 4. Mundo reference architecture

that their actions are only recorded and disclosed

to the degree they consent to or that is legally
imposed. As another example, they want to be
sure that they only trigger actions which they
understand in their legal, financial, and other
consequences and that they agree to. To this end,
the Mundo researchers propose to conceptualize
a single, truly owned UC node type that acts in
the users stead and controls when, how, and to
what extent other UC node types are invited or
chartered to participate in actions. Since computer
use becomes ubiquitous, such a personally-owned
node type must be carried along virtually at all
times. This imposes strong requirements with
respect to miniaturization, robustness, and the
conflicting goals of (a) the impossibility to falsify
or duplicate such a node, and (b) the possibility
to replace it easily in case of theft or failure. An
important research questions is concerned with
the minimum functionality of a Me.
Me nodes are considered as the representation of their users in the digital worlda digital
persona involved in all user activities. It is a small
wearable computer with minimal functionality.
In order to support interaction with UC environments in a sensible way, the term minimal must
be associated with a set of specific requirements
regarding size, identity, security, interaction, context awareness, and networking. The design was
guided by the principle that the minimal feature set
of a system is determined by the worst-case environmental conditions under which the application
must run satisfactorily (Satyanarayanan, 2001).
This leads to a focus on speech based interac-

16

tion and it is described in detail by Aitenbichler,

Kangasharju, and Mhlhuser (2004). Any Me
can augment its capabilities through association
with other entities of the Mundo architecture as
described next.
Us (Ubiquitous aSsociable object): Minimization pressure will not permit feature-rich Mes.
Hence, they must be able to connect to other
mobile devices or devices embedded into the
environment to offer more powerful services
to their users, such as large display space. This
process is called association and such devices
are called ubiquitous associable objects (Us). A
Us is a computing device that extends the users
personal environment by adding storage, processing capacity, displays, interaction devices,
and so forth. During association, the Me sends
authentication information to the Us, sets up a
secure communication link, and personalizes
the Us to suit the users preferences and needs.
For privacy reasons, any personalization of a Us
becomes automatically unavailable if it is out of
range of the users Me.
It (smart ITem): There are also numerous
smart items that do not support association that
would classify them as Us. Vending machines,
goods equipped with radio frequency IDs, and
landmarks with what is functionality are just
a few examples. Such devices are called smart
items (Its). An It is any digital or real entity that
has an identity and can communicate with a Us
or the Me. Communication may be active or passive. Memory and computation capabilities are
optional (cf. the four constituents of a UC node
described previously).
We (Wireless group Environment): Ad-hoc
networking is restricted to an area near to the
user of a Me device, as connections with remote
services will involve a non ad hoc network infrastructure. The functionality of a wireless group
environment is to bring together two or more
personal environments consisting of a Me and
arbitrary Us entities each. It enables cooperation
between the devices and also allows for sharing

Introduction to Ubiquitous Computing

and transferring hardware (e.g., Us devices) and

software or data between We users.
They (Telecooperative Hierarchical ovErlaY)
stands for the backbone infrastructure as part of
the Mundo component architecture. It connects
users to the (nonlocal) world, and delivers services
and information to the user. The They integrates
different physical networks and provides transparent data access to users. Frequently used data
may be cached on Us devices.

UC Layered Reference Architectures

Many actual UC projects are based on a layered
architecture. Most of them are just first approaches
to software architectures, only a few of them are
intended to serve as a crystallization point for the
community and future standards. Nevertheless,
one of them may turn out to be so successful
that a future reference architecture will evolve
from it. We will concentrate on a small selection
of the few projects that have a general reference
model in mind. They concentrate on different
challenges or foci, that is their findings will have
to be merged if a holistic layered architecture is
to be derived.
A first focus is the enterprise modeling that
ODP already addressed. A reference architecture
worth mentioning here is ODSI, the so-called open
distributed services infrastructure (Bond, 2001).
Although already outdated, ODSI was influential
since it fostered the move away from ODPs more
top-down approach to a component-based, that is
service based approach that supports the concept
of applications being compositions of services.
Other reference architectures emphasize
Smart Environments. Two facets are important
and investigatedstillin different camps even
as to the work on reference architectures: smart
information spaces and smart physical spaces. By
smart information spaces, we mean environments
which concentrate on cooperative treatment of ITand data/media centric work (cf. Mark Weisers
three initial UC devices). Smart physical spaces

are often called smart spaces or more specifically

smart houses, labs, offices, homes etc. Work on
these kinds of environments emphasizes the
tangible, physical (computer-augmented) objects
to be handled.
As for smart information spaces, an interesting reference architecture was proposed in the
LifeSpaces project in South Australia (Bright &
Vernik, 2004). Their architecture incorporates
some of the findings from ODSI and distinguishes
four layers:
1.

2.

3.

4.

Enterprise model: This layer supports

rules, processes, and organizational models
of roles and services in the enterprise.
Coordination and control including interaction support: On this layer, a shared and
persistent event space of limited capacity,
and an agent-based workspace infrastructure
are offered.
Enterprise bus: This term refers to a communication layer based on the publish/subscribe paradigm.
The service layer: Here, the core functionality is represented by easily composable
services. An enterprise bus is offered for
services to communicate and cooperate; this
bus connects so-called peers which host the
services.

As for smart physical spaces, a prominent

example is the reference architecture developed
by the Gator Tech Smart House project of the
University of Florida (see Figure 5). The reference
architecture depicted is a more recent version of
what was published by Helal, Mann, El-Zabadani,
King, Kaddoura, and Jansen (2005) and is included
by courtesy of the authors (the commercial version
is called Atlas now). For more information, the
reader may consult the groups Web Site at www.
icta.ufl.edu or the Atlas Web site at www.pervasa.
com. The architecture emphasizes sensors (plus
actuators) and networked embedded devices at
the lowest layer as the hardware foundation of UC

17

Introduction to Ubiquitous Computing

Figure 5. SmartSpace middleware layered reference architecture

applications. The OSGI standard is exploited for

customizing and maintaining these sensors and
embedded devices in a dedicated second layer.
The third layer contains three large parts which
reflect major insights into the nature of the UC
world (note that these insights have a large influence on the present book, too):

18

The context management layer reflects the

importance of context-awareness for UC as
a whole, as discussed in the preface of the
book;
The service layer reflects services (and
service-oriented architectures, SOA) as
the dominating paradigm for building
autonomous software components in a UC
setting;
The knowledge layer reflects the fact that
large-scale service composition cannot rely
on standardized interfaces that are distributed prior to software (service) development; rather, service discovery and service
interaction must rely on machine readable
descriptions of the service semantics available at runtime;

Due to the strictly service-oriented concept

used, application development boils down
to service composition; the top layer offers
corresponding tools.

In conclusion, it should have become clear

that both a component based and a layered reference architecture, if widely accepted, would be
important steps from UC islands towards truly
global UC. The reference architectures presented
could serve as a basis for better communication
among the UC protagonists and for the necessary
standards.

References
Aitenbichler, E., Kangasharju, J., & Mhlhuser,
M. (2004). Talking assistant headset: A smart digital identity for ubiquitous computing. Advances
in pervasive computing (pp. 279-284). Austrian
Computer Society.
Bond, A. (2001). ODSI: Enterprise service co-ordination. In Proceedings of the 3rd International
Symposium on Distributed Objects and Applications DOA01 (pp. 156-164). IEEE Press.

Introduction to Ubiquitous Computing

Bright, D., & Vernik, R. (2004). LiveSpaces: An

interactive ubiquitous workspace architecture
for the enterprise in embedded and ubiquitous
computing. Springer (pp. 982-993).

Helal, A.A., Haskell, B., Carter, J.L., Brice, R.,

Woelk, D., & Rusinkiewicz, M. (1999). Any time,
anywhere computing: Mobile computing concepts
and technology. Springer.

Helal, S., Mann, W., El-Zabadani, H., King, J.,

Kaddoura, Y., & Jansen, E. (2005, March). The
Gator Tech Smart House: A programmable pervasive space. IEEE Computer, 38(3), 64-74.

Huber, A. J. F. & Huber, J.F. (2002): UMTS and

mobile computing. Artech House.

Satyanarayanan, M. (2001). Pervasive computing:

Vision and challenges. IEEE Personal Communications (pp. 10-17). IEEE Press.

Additional Reading
Aarts, E., & Encarnaco J. L. (Eds.). (2006). True
visions. The emergence of ambient intelligenc.
Berlin, Germany: Springer.
Adelstein, F., Gupta, S. K. S. et al. (2004). Fundamentals of mobile and pervasive computing. New
York: McGraw-Hill Professional Publishing.
Antoniou, G., & van Harmelen, F. (2004). A semantic web primer. Massachusetts: MIT Press.
Hansmann, U., Merk, L. et al. (2003). Pervasive
computing handbook. The mobile world. Berlin,
Germany: Springer.
Hedgepeth, W.O. (2006): RFID metrics: Decision
making tools for todays supply chains. University
of Alaska, Anchorage, USA

Jurafsky, D., & Martin, J. H. (2000). Speech und

language processing. Upper Saddle River, NJ:
Prentice Hall.
Lu, Y., Staff, L.Y., Zhang, Y., Yang, L.T., &
Ning, H. (2008). The Internet of things. Taylor
& Francis Group
McTear, M. F. (2004). Spoken dialogue technolog.
London: Springer.
Moreville, P. (2005): Ambient findability.
OReilly.
Riva, G., Vatalaro, F., Davide, F., & Alcaiz, M.
(2005): Ambient intelligence. Amsterdam: IOS
Press.
Sharp, H., Rogers, Y., Preece, J. (2002). Interaction
design: Beyond human-computer interaction. J.
Wiley & Sons.
Stajano, F. (2002). Security for ubiquitous computing. Cambridge: John Wiley & Sons, Ltd.

Weber, W., Rabaey, J. M., & Aarts, E.

(Eds.). (2005). Ambient intelligence, Berlin,
Germany: Springer.

This work was previously published in Handbook of Research on Ubiquitous Computing Technology for Real Time Enterprises,
edited by M. Mhlhuser and I. Gurevych, pp. 1-20, copyright 2008 by Information Science Reference, formerly known as Idea
Group Reference (an imprint of IGI Global).

19

20

Chapter 1.2

Ubiquitous Computing and

the Concept of Context
Antti Oulasvirta
Helsinki Institute for Information Technology, Finland
Antti Salovaara
Helsinki Institute for Information Technology, Finland

INTRODUCTION
Mark Weiser (1991) envisioned in the beginning
of the 1990s that ubiquitous computing, intelligent
small-scale technology embedded in the physical
environment, would provide useful services in the
everyday context of people without disturbing the
natural flow of their activities.
From the technological point of view, this
vision is based on recent advances in hardware
and software technologies. Processors, memories,
wireless networking, sensors, actuators, power,
packing and integration, optoelectronics, and biomaterials have seen rapid increases in efficiency
with simultaneous decreases in size. Moores
law on capacity of microchips doubling every 18
months and growing an order of magnitude every
five years has been more or less accurate for the last
three decades. Similarly, fixed network transfer
capacity grows an order of magnitude every three
years, wireless network transfer capacity every

5 to 10 years, and mass storage every 3 years.

Significant progress in power consumption is less
likely, however. Innovations and breakthroughs
in distributed operating environments, ad hoc
networking, middleware, and platform technologies recently have begun to add to the ubiquitous
computing vision on the software side.
Altogether, these technological advances
have a potential to make technology fade into
the background, into the woodwork and fabric
of everyday life, and incorporate what Weiser
(1991) called natural user interfaces. Awareness
of situational factors (henceforth, the context)
consequently was deemed necessary for this
enterprise. This article looks at the history of
the concept of context in ubiquitous computing
and relates the conceptual advances to advances
in envisioning human-computer interaction with
ubiquitous computing.

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Ubiquitous Computing and the Concept of Context

BACKGROUND
Ubiquitous Computing Transforms
Human-Computer Interaction
Human-computer interaction currently is shifting its focus from desktop-based interaction to
interaction with ubiquitous computing beyond
the desktop. Context-aware services and user
interface adaptation are the two main application classes for context awareness. Many recent
prototypes have demonstrated how context-aware
devices could be used in homes, lecture halls,
gardens, schools, city streets, cars, buses, trams,
shops, malls, and so forth.
With the emergence of so many different ways
of making use of situational data, the question of
what context is and how it should be acted upon
has received a lot of attention from researchers
in HCI and computer science. The answer to this
question, as will be argued later, has wide ramifications for the design of interaction and innovation
of use purposes for ubiquitous computing.

HISTORY
Context as Location
In Weisers (1991) proposal, ubiquitous computing
was realized through small computers distributed
throughout the office. Tabs, pads, and boards
helped office workers to access virtual information
associated to physical places as well as to collaborate over disconnected locations and to share
information using interfaces that take locational
constraints sensitively into account. Although
Weiser (1991) never intended to confine context to
mean merely location, the following five years of
research mostly focused on location-based adaptation. Want et al. (1992) described the ActiveBadge,
a wearable badge for office workers that could
be used to find and notify people in an office.
Weiser (1993) continued by exploring systems for

sharing drawings between disconnected places

(the Tivoli system). Schilit et al. (1994) defined
context to encompass more than locationto
include people and resources as wellbut their
application examples were still mostly related to
location sensing (i.e., proximate selection, location-triggered reconfiguration, location-triggered
information, and location-triggered actions).
Want, et al. (1995) added physical parameters like
time and temperature to the definition. Perhaps the
best-known mobile application developed during
this location paradigm era was the CyberGuide
(Long et al., 1996), an intelligent mobile guide
that could be used to search for nearby services
in a city. This paradigm was also influential in
the research on Smart Spaces, such as intelligent
meeting rooms.

The Representational Approach to

Context
Although the idea that location equals context was
eventually dismissed, many researchers coming
from computer science still believed that contexts
were something that should be recognized, labeled,
and acted upon (Schmidt et al., 1998). Here, context
was supposed to be recognized from sensor data,
labeled, and given to applications that would use
it as a basis for adaptation. Dey et al.s (1999) five
Ws of contextWho, Where, When, What, and
Whyextended this approach and demonstrated
convincing examples of how a labeled context
could be used for presenting, executing, and tagging information. Tennenhouses (2000) proactive
computing paradigm endorsed a similar way of
thinking about context, emphasizing the role of
computers in doing real-time decisions on behalf
of (or pro) the user. A somewhat similar approach
that also attempts to delegate decision-making
responsibility to intelligent systems is taken
by the Ambient Intelligence (AmI) technology
program of the European Union (ISTAG). One
part of the AmI vision entails intelligent agents
that assume some of the control responsibility
from the users.
21

Ubiquitous Computing and the Concept of Context

The latest widely referred to definition was

given by Dey et al. (2001), who defined context
as any information that characterizes a situation
related to the interaction between users, applications, and the surrounding environment (p. 106).
Satyanarayanans (2001) formulation of pervasive
computing also belongs to this line of thinking,
but the author has chosen to avoid defining context
and merely admits that it is rich and varies.
In his review of context definitions over the
years, Dourish (2004) calls this the representational approach to context. Recent work within
this branch has come close to finding the limits to
recognizing and labeling contexts. For example,
simple physical activities of a person in a home
environment can be recognized with about 8085% accuracy (Intille et al., 2004), as can be the
interruptability of a person working in an office
(Fogarty et al., 2004). Some critics have drawn
parallels from this enterprise to problems encountered in strong AI (Erickson, 2002).

FUTURE TRENDS

give special consideration to activities that people

engage in and highlight their dynamic nature,
different from the labeling-oriented representational approach. Activity-centered approaches
emphasize both turn taking in communication
between the user and the applications (Fischer,
2001) and acknowledge the situated and timevarying nature of the needs that a user has in his
or her life (Greenberg, 2001). This line of research
highlights the difficulties that exist in making
correct inferences about a users tasks through
sensor information. Considerations of social
issues in ubiquitous computing design include
questions of how to fit computation intelligence
into peoples routines in an unremarkable manner
(Tolmie et al., 2002) and how peoples patterns of
interaction with humans and computers change
when computationally augmented artifacts are
adopted into use. Yet another emerging idea from
HCI addresses specifically the aim to be free from
distraction; that is, when it is appropriate to interrupt the user at his or her present task. Some call
systems with such inferring capabilities attentive
user interfaces (Vertegaal, 2003).

New Directions Inspired by Human

and Social Sciences

CONCLUSION

By the year 1996, other approaches to context

were beginning to emerge. Wearable computing (Mann, 1996) looked at personal wearable
computers able to help us remember and capture
our everyday experiences through video and
sound recording of context. Tangible bits (Ishii
& Ullmer, 1997), although inspired by ubiquitous
computing, looked at context not as something
that had to be reacted upon but as surroundings
of the user that could be augmented with tangible
(i.e., graspable) computers and ambient media that
display digital information using distraction-free
output channels.
More recently, researchers have started to emphasize the social context and issues in peoples
practices and everyday conduct. These approaches

HCI in ubiquitous computing has been both

inspired and constrained by conceptual developments regarding the concept of context. Weisers
(1991) initial work caused researchers to conceive
context narrowly as encompassing mainly location
and other static, easily measurable features of a
users context. After about five years of research,
the restrictiveness of this definition was realized,
and broader definitions were formulated. Still,
context mainly was pursued by computer scientists and seen as something that must be labeled
and reacted upon to adapt user interfaces. More
recent work by human and social scientists has
emphasized the role of user studies and theoretical
reasoning in understanding what context entails
in a particular application.

22

Ubiquitous Computing and the Concept of Context

REFERENCES
Dey, A. K., & Abowd, G. D. (1999). Towards a better understanding of context and contextawareness [technical report]. Atlanta: Georgia Institute
of Technology.

Long, S., Aust, D., Abowd, G., & Atkeson, C.

(1996). Cyberguide: Prototyping context-aware
mobile applications. Proceedings of the Conference on Companion on Human Factors in Computing Systems (CHI96).

Dey, A. K., Salber, D., & Abowd, G. A. (2001).

A conceptual framework and a toolkit for supporting the rapid prototyping of context-aware
applications. Human-Computer Interaction,
16(2-4), 97-166.

Mann, S. (1996). Smart clothing: Wearable

multimedia computing and personal imaging
to restore the technological balance between
people and their environments. Proceedings of
the Fourth ACM International Conference on
Multimedia.

Dourish, P. (2004). What we talk about when

we talk about context. Personal and Ubiquitous
Computing, 8(1), 19-30.

Satyanarayanan, M. (2001). Pervasive computing:

Vision and challenges. IEEE Personal Communications, 8(4), 10-17.

Erickson, T. (2002). Some problems with the

notion of context-aware computing. Communications of the ACM, 25(2), 102-104.

Schilit, B., Adams, N., & Want, R. (1994). Contextaware computing applications. Proceedings of the
IEEE Workshop on Mobile Computing Systems
and Applications.

Fischer, G. (2001). Articulating the task at hand

and making information relevant to it. HumanComputer Interaction, 16(2-4), 243-256.
Fogarty, J., Hudson, S. E., & Lai, J. (2004). Examining the robustness of sensor-based statistical
models of human interruptibility. Proceedings of
the SIGCHI Conference on Human Factors in
Computing Systems (CHI04).
Greenberg, S. (2001). Context as a dynamic
construct. Human-Computer Interaction, 16(24), 257-268.
Intille, S., Bao, L., Munguia Tapia, E., & Rondoni, J. (2004). Acquiring in situ training data for
context-aware ubiquitous computing applications.
Proceedings of the SIGCHI Conference on Human
Factors in Computing Systems (CHI04).
Ishii, H., & Ullmer, B. (1997). Tangible bits: Towards seamless interfaces between people, bits
and atoms. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems
(CHI97).

Schmidt, A., Beigl, M., & Gellersen, H. -W. (1998).

There is more to context than location. Computers
& Graphics, 23(6), 893-901.
Tennenhouse, D. (2000). Proactive computing.
Communications of the ACM, 43(5), 43-50.
Tolmie, P., Pycock, J., Diggins, T., MacLean, A.,
& Karsenty, A. (2002). Unremarkable computing.
Proceedings of the SIGCHI Conference on Human
Factors in Computing Systems (CHI02).
Vertegaal, R. (2003). Introduction to a special issue on attentive user interfaces. Communications
of the ACM, 46(3), 31-33.
Want, R., Hopper, A., Falco, V., & Gibbons, J.
(1992). Active badge location system. ACM Transactions on Information Systems, 10(1), 91-102.
Want, R., et al. (1995). An overview of the ParcTab
ubiquitous computing experiment. IEEE Personal
Communications, 2(6), 28-43.
Weiser, M. (1991). The computer for the 21s t century. Scientific American, 265(3), 66-75.

23

Ubiquitous Computing and the Concept of Context

Weiser, M. (1993). Some computer science issues

in ubiquitous computing. Communications of the
ACM, 36(7), 75-84.

KEY TERMS
Attentive User Interfaces: AUIs are based
on the idea that modeling the deployment of
user attention and task preferences is the key for
minimizing the disruptive effects of interruptions.
By monitoring the users physical proximity, body
orientation, eye fixations, and the like, AUIs can
determine what device, person, or task the user
is attending to. Knowing the focus of attention
makes it possible in some situations to avoid interrupting the users in tasks that are more important
or time-critical than the interrupting one.

Proactive Computing: A research agenda of

developing interconnected devices and agents,
equipped with faster-than-human-speed computing capabilities and means to affect real-world
phenomena that a user can monitor and steer
without a need to actively intervene in all decision-making situations. By raising the user above
the traditional human-computer interaction loop,
efficiency and freedom from distraction are expected to be enhanced.
Tangible Bits: According to Hiroshi Ishii of
MIT, the smooth transition of users focus of
attention between background and foreground
using ambient media and graspable objects is a
key challenge of Tangible Bits (p. 235).

Context: That which surrounds and gives

meaning to something else. (Source: The Free
On-line Dictionary of Computing, http://foldoc.
doc.ic.ac.uk/foldoc/)

Tangible User Interfaces: Systems that give

a physical form to digital information through
augmenting tools and graspable objects with
computing capabilities, thus allowing for smooth
transitions between the background and foreground of the users focus of attention.

Peripheral Computing: The interface attempts to provide attentionally peripheral awareness of people and events. Ambient channels
provide a steady flow of auditory cues (i.e., a
sound like rain) or gradually changing lighting
conditions.

Unremarkable Computing: An approach

that focuses on designing domestic devices that
are unremarkable to users. Here, unremarkable is
understood as the use of a device being a part of a
routine, because, it is believed, routines are invisible in use for those who are involved in them.

Pervasive Computing: Technology that provides easy access to information and other people
anytime and anywhere through a mobile and
scalable information access infrastructure.

Wearable Computing: Technology that

moves with a user and is able to track the users motions both in time and space, providing real-time
information that can extend the users knowledge
and perception of the environment.

This work was previously published in Encyclopedia of Human Computer Interaction, edited by C. Ghaoui, pp. 630-633, copyright 2006 by Information Science Reference, formerly known as Idea Group Reference (an imprint of IGI Global).

24

25

Chapter 1.3

Gender and End-User

Computing
Laura Beckwith
Oregon State University, USA
Margaret Burnett
Oregon State University, USA
Shraddha Sorte
Oregon State University, USA

INTRODUCTION
Although gender differences in a technological
world are receiving significant research attention,
much of the research and practice has aimed at
how society and education can impact the successes and retention of female computer science
professionals. The possibility of gender issues
within software, however, has received almost no
attention, nor has the population of female end
users. However, there is relevant foundational
research suggesting that gender-related factors
within a software environment that supports
end-user computing may have a strong impact
on how effective male and female end users can
be in that environment. Thus, in this article, we
summarize theory-establishing results from
other domains that point toward the formation of

grounded hypotheses for studying gender differences in end-user computing.

There has been much background research
relevant to human issues of end-user computing, which we define here as problem-solving
using computer software, also termed end-user
programming in some of the literature (e.g.,
Blackwell, 2002; Green & Petre, 1996; Nardi,
1993). (See the glossary for definitions of these and
related terms.) Despite this, few researchers have
considered potential gender HCI issues and gender
differences that may need to be accounted for in
designing end-user computing environments. The
most notable exception is Czerwinskis pioneering
research on the support of both genders in navigating through 3-D environments (Czerwinski, Tan,
& Robertson, 2002; Tan, Czerwinski, & Robertson, 2003). Although individual differences, such

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Gender and End-User Computing

as experience, cognitive style, and spatial ability,

are likely to vary more than differences between
gender groups, evidence from Czerwinskis work
as well as work in other domains, such as psychology and marketing, has found gender differences
relevant to computer usage. In fact, some research
has shown that some software is (unintentionally)
designed for males (Huff, 2002).
One reason gender HCI issues in end-user computing are important is that ignorance of gender
issues has already proven to be dangerous: todays
low percentage of computer science females
(Camp, 1997) has been directly attributed to the
past unawareness of gender issues in computer science education and in the workforce. There is a risk
that if gender HCI issues in end-user computing
environments are ignored, a similar phenomenon
could occur with female end users.

WHAT COULD GO WRONG?

What gender differences might matter in the
design of end-user computing environments?
Consider the following scenario in one particular
end-user computing environment.
Imagine a female teacher engaged in preparing a spreadsheet to track her students scores
and to calculate ways of providing students with
the best grades. Part of her process of preparing
her spreadsheet is to test the spreadsheet. While
she is engaged in testing, the system surprises
her by decorating some of the spreadsheet cells,
as in Figure 1.
The surprises were intentionally placed into
the software by the designers relying on a strategy for end-user computing environments called
Surprise-Explain-Reward (Wilson et al., 2003).
The surprise, which was intended to capture the
teachers attention and arouse her curiosity, reveals
the presence of an information gap (Lowenstein,
1994). In this case the system is using the surprise
to interest her in assertions (Burnett et al., 2003),
which she can use to guard against future errors

26

by specifying, for example, that the value of a

cell calculating a grade average should always
fall between 0 and 100.
What could go wrong in surprising the user?
According to Lowensteins information gap
theory, a user needs to have a certain level of
confidence in order to reach a useful level of
curiosity (Lowenstein, 1994). However, given
documented gender differences in computer confidence (Busch, 1995; Huff, 2002), the teachers
level of computer confidence could interfere with
the surprises ability to capture her interest.
Returning to our scenario, suppose for this
particular user, the surprise is effective at arousing her curiosity, she looks to the object that surprised her (the assertion) for an explanation. The
explanation, viewed through a tooltip, includes the
semantics, possible actions she can take (regarding
the assertion), and the future reward(s) of taking
the action. See Figure 1.
What could go wrong with the explanation?
According to one theory, males and females process information differently (Meyers & Sternthal,
1991; ODonnell & Johnson, 2001), and thus both
the presentation and the content of the explanation may impact its effectiveness for males versus
females. If the information needed by the user is
not effectively communicated, the users ability
to problem solve is likely to be reduced.
Another role of the explanation is to help users make a reasonably accurate assessment of the
risk in taking some actionbut since males and
females differ in their perceptions of risk (Byrnes,
Miller, & Schafer, 1999), the explanation may
need to serve these two populations differently
in this respect as well. (An example of risk may
be the fear that the user will lose their work if
they try a certain feature.) If one gender perceives
an explanation of a feature as communicating
higher levels of risk than another, the users with
higher risk perceptions may avoid supposedly
risky features that may be important to overall
effectiveness.

Gender and End-User Computing

Figure 1. A spreadsheet calculating the average of three homework scores. Assertions about the ranges
and values are shown above each cells value. For example, on HomeWork1 there is a user-entered
assertion (noted by the stick figure) of 0 to 50. The other three cells have assertions guessed by the
Surprise-Explain-Reward strategy. Since the value in HomeWork1 is outside of the range of the assertion, a red circle notifies the user of the violation. A tooltip (lower right) shows the explanation for one
of the guessed assertions.

Perhaps the most important role of explanations

is to make clear the rewards of using particular
features of the software. Providing information
about rewards in the explanation is consistent with
the implications of the Model of Attention Investment (Blackwell, 2002), an analytic model of user
problem-solving behavior that models the costs,
benefits, and risks users weigh in deciding how
to complete a task. An implication of this model
is that if the system provides the user an idea of
future benefits, users can better assess if the cost
of using a feature (here assertions) is worth their
time. The reward aspect of the strategy refers to
rewards such as the automatic detection of errors, which is depicted by the red circle around
HomeWork1s erroneous value in Figure 1.
What could go wrong with rewards? Since
males and females are often motivated by different
factors, there may be gender differences in what
actually is a perceived reward. If the rewards
are only tailored to one genders perceptions of
rewards, the other gender may not be motivated to
use the devices that will help them be effective.
In this end-user computing scenario, potential
problems arose that may be addressable within the
end-user computing software itself. Four issues
that arose here were (1) software features whose
effects depend upon users computer confidence
(discussed in the section on confidence), (2) the
softwares ability to communicate effectively with

users (discussed in support), (3) the possibility of

a users perception of risk interfering with the user
choosing to use appropriate features (discussed in
confidence), and (4) possible differences between
a users actual motivations and the softwares
attempt to reward users for using particular
features (discussed in motivation). These issues
together form a useful organizational framework
for considering gender HCI.

CONFIDENCE
This document uses the term confidence for the
interrelated concepts of self-confidence, self-efficacy, overconfidence, and perceived risk.
From the field of computer science, there is
substantial evidence of low confidence levels as
computer science females compare themselves
to the males (Margolis, Fisher, & Miller, 1999).
Of particular pertinence to end-user computing,
however, is the evidence showing that low confidence relating to technology is not confined to
computer science females (Busch, 1995; Huff,
2002; Torkzadeh & Van, 2002).
As a measure of confidence, researchers often use self-efficacy, as was done in the Busch
study. Self-efficacy is belief in ones capabilities
to perform a certain task (Bandura, 1994). There
is specific evidence that low self-efficacy impacts

27

Gender and End-User Computing

attitudes toward a new software package prior to its

use (Hartzel, 2003). Taken together, this research
suggests that a first experience with end-user
computing software can impact attitudes which
may in turn impact users future choices to (or
not to) use some features in their software.
Overconfidence matters too, because it can
prevent people from suspecting errors, leading
to misplaced trust in erroneous programs. In
particular, overconfidence in spreadsheet correctness is common (Panko, 1998). There is evidence
(Lunderberg, Fox, & Punchochar, 1994) suggesting gender differences in overconfidence just as
in under confidence. Hence, designing methods
to help alleviate overconfidence in end-user computing needs to be carefully targeted specifically
toward overconfident users.
Perception of risk is tied to confidence, and
impacts the decisions people make. According to
the attention investment model (a model of how
users allocate their attention in problem solving)
(Blackwell, 2002), a user may choose not to follow
through a particular action, if they decide that the
costs and/or risks are too high in relation to the
benefits of taking that action. Perception of risk
thus plays an important role in a users decisionmaking about whether to use some features, and
can lead to differences in actual behavior.
Differences in perception of risk have been
tied to gender. For example, a meta-analysis of
150 studies on gender and risk taking found that
females engaged in less risk taking than males
(Byrnes, Miller, & Schafer 1999). The meta-analysis did not address risks of computer use directly.
However, it did find that intellectual risk taking,
defined as activities involving mathematical or
spatial reasoning skills, was greater in males than
in females (Byrnes, Miller, & Schafer 1999).
To obtain evidence about whether confidence
might directly impact gender HCI issues for enduser computing environments, we conducted a
preliminary think-aloud study in which participants attempted to debug two spreadsheets, given
the support of Surprise-Explain-Reward devices

28

(Beckwith & Burnett, 2004). To our surprise,

the females confidence levels dropped over the
course of the study much more than did the males
confidence levels. This result suggests that enduser computing environment itselfwhich, like
other end-user computing environments, was designed without knowledge of gender-related HCI
principlesis not currently serving the females
needs as well as the males. A subsequent follow-up
study (Beckwith et al., 2005) confirmed this: in the
spreadsheet paradigm, ties were found between
females confidence issues and low utilization
of features aimed to support problem solving,
resulting in effectiveness problems.

SUPPORT
We will use the term support to mean built-in
aspects of the software, such as on-line help systems and Figure 1s tooltips that help users learn
or understand the environment.
The systems approach to help users achieve
mastery in remembering the softwares devices
may depend on a users learning style. One survey of university students found that students
with an abstract random learning style were
significantly more likely to be female and, as
a result, could find computer-based instruction
ineffective for learning (Ames, 2003). Other researchers have also found gender differences in
learning styles (Heffler, 2001; Severiens & ten
Dam, 1997). One implication of these findings
is that end-user computing may need to support
several learning styles, especially if some users are
easily dissuaded by support devices not sensitive
to their learning style.
Problem-solving style also shows gender differences, at least for computer games (Kafai, 1998).
Researchers found that, unlike boys, rather than
working in a linear fashion through the game,
girls prefer to explore and move freely about a
game (Gorriz & Medina, 2000). In another difference in problem-solving style, boys games

Gender and End-User Computing

typically depend upon competition, whereas girls

prefer collaboration and working together (Kafai,
1998). For end-user computing environments,
these problem-solving differences suggest differences in the support provided by the system. For
example, supporting both linear and non-linear
problem-solving styles and providing avenues
for both competition and collaboration may be
important for softwares success at supporting
both genders adequately.
Finally, the theory of selectivity suggests
gender differences which may impact how users
process the information support devices provide.
The theory of selectivity, from research in the
area of marketing, states that males and females
differ in their information processing strategies
(Meyers & Sternthal, 1991; ODonnell & Johnson,
2001). According to this theory of selectivity,
females are more likely to employ elaborative
information processing strategies, regardless of
whether the task is simple or complex in nature.
Males, however, are more likely to select heuristic
processing strategies that minimize cognitive effort and reduce information load for simple tasks,
switching to an elaborative strategy only on more
complex tasks.
These gender differences have been shown to
impact diverse software-related activities, ranging
from users perceptions of Web sites used for ecommerce (Simon, 2001) to users performance
on auditing tasks (ODonnell & Johnson, 2001).
For end-user computing environments, this research may have implications for informing end
users of important information via the softwares
support devices.

MOTIVATION
Research has shown that computer science females
are motivated by how technology can help other
people, whereas males tend to enjoy technology
for its own sake (Margolis, Fisher, & Miller,
1999). These differences are also found with other

females who use technology, such as architects,

NASA scientists, and filmmakers. One study
(Brunner, Bennett, & Honey, 1998) found that
females described technological objects as tools
to help integrate personal and professional lives
and to facilitate creativity and communication,
whereas males described them as technological
devices to increase command and control over
nature and one another. The gender differences
found in that study are summarized in Table 1.
The technology acceptance model (TAM)
(Morris & Dillon, 1997; Venkatesh & Morris,
2000) provides a model of users acceptance
and usage behavior of technology. According to
TAM, user acceptance, and ultimately technology
use, is determined by two key beliefs: perceived
usefulness and perceived ease of use (Venkatesh
& Morris, 2000). Perceived usefulness is the
degree to which a user believes that using the
system will enhance their performance, and
perceived ease of use is the degree to which
the user believes that using the system will be
free of effort. According to one study, the relative
importance of each differs by gender (Venkatesh
& Morris, 2000); women were more influenced
by perceived ease of use whereas men were more
influenced by perceived usefulness.

Future Trends and

CONCLUSION
To date, there has been little research on how the
design of software itself may interact with gender
differences. Still, foundational work from several
domains strongly suggests that such differences
may have critical impacts on users success in
end-user computing.
Although research is beginning to emerge
providing some insights into genders importance
in end-user computing environments, it is still
largely an open question. Also open are questions
of what specific types of differences matter in
such environments and what amelioration strategies are possible.
29

Gender and End-User Computing

Table 1. Summary of gender differences in fantasizing about technology (Brunner, Bennett, & Honey,
1998). Reprinted with permission.
Women
fantasize about it as a
MEDIUM
2 see it as a
TOOL
3 want to use it for
COMMUNICATION
4 are im pressed with its potentia l for
CREATION
5 see it as
EXPRESSIVE
6 ask it for
FLEXIBILITY
7 are concerned with its
EFFECTIVENESS
8 like its ability to facil itate
SHARING
9 are concerned with
INTEGRATING it into their
personal lives
10 talk about wanting to
EXPLORE
worlds
11 are
EMPOWERED by it
1

To help provide a foundation upon which

researchers interested in these issues can build,
this article has drawn from five domains to summarize literature relevant to these questions. All
of the literature surveyed identifies one or more
issues that potentially impact end users success
that are also potentially addressable within the
software systems. Together, the open questions
and survey are intended to provide a foundation
for future investigation.

ACKNOWLEDGMENT
This work was supported in part by Microsoft
Research and by the EUSES Consortium via NSF
grants ITR 0325273 and CNS 0420533.

30

Men
fantasize abou t it as aPRODUCT
see it as a
WEAPON
want to use it for
CONTROL
are impressed with its potential for
POWER
see it a sINSTRUMENTAL
ask it for
SPEED
are concerned with its
EFFICIENCY
like its ability to facil itate
AUTONOMY
are in tent on
CONSUMING it
talk about using it to
EXPLOIT
resources and po tentialities
want
TRANSCENDENCE

REFERENCES
Ames, P. (2003). Gender and learning styles
interactions in students computer attitudes.
Journal of Educational Computing Research,
28(3), 231-244.
Bandura, A. (1994). Self-efficacy. In V. S. Ramachaudran (Ed.), Encyclopedia of human behavior
(Vol. 4, pp. 71-81). New York: Academic Press.
Beckwith, L., & Burnett, M. (2004, September
26-29). Gender: An important factor in end-user
programming environments? IEEE Symp. Visual
Languages and Human-Centric Computing (pp.
107-114), Rome, Italy.
Beckwith, L., Burnett, M., Wiedenbeck, S., Cook,
C., Sorte, S., & Hastings, M. (2005, April 2-7).
Effectiveness of end-user debugging software
features: Are there gender issues? ACM Confer-

Gender and End-User Computing

ence on Human Factors in Computing Systems

(pp. 869-878), Portland, Oregon.
Blackwell, A. (2002, September 3-6). First steps
in programming: A rationale for Attention Investment models. IEEE Symp. Human-Centric
Computing Languages and Environments (pp.
2-10), Arlington, VA.
Brunner, C., Bennett, D., & Honey, M. (1998).
Girl games and technological desire. In J. Cassell & H. Jenkins (Eds.), From Barbie to Mortal
Kombat: Gender and computer games (pp. 72-88).
Cambridge, MA: MIT Press.
Burnett, M., Cook, C., Pendse, O., Rothermel,
G., Summet, J., & Wallace, C. (2003, May 3-10).
End-user software engineering with assertions
in the spreadsheet paradigm. International Conference on Software Engineering (pp. 93-103).
Portland, Oregon.
Busch, T. (1995). Gender differences in self-efficacy and attitudes toward computers. Journal of
Educational Computing Research, 12, 147-158.
Byrnes, J. P., Miller, D. C., & Schafer W. D. (1999).
Gender differences in risk taking: A meta-analysis, Psychological Bulletin, 125, 367-383.
Camp, T. (1997). The incredible shrinking
pipeline. Communications of the ACM, 40(10),
103-110.
Czerwinski, M., Tan, D., & Robertson, G. (2002,
April 20-25). Women take a wider view. ACM
Conference on Human Factors in Computing
Systems (pp. 195-202), Minneapolis, MN.
Gorriz, C., & Medina, C. (2000). Engaging girls
with computers through software games. Communications of the ACM, 43(1), 42-49.
Green, T. R. G., & Petre, M. (1996). Usability
analysis of visual programming environments:
A cognitive dimensions framework. Journal
of Visual Languages and Computing, 7(2), 131174.

Hartzel, K. (2003). How self-efficacy and gender

issues affect software adoption and use. Communications of the ACM, 43(9), 167-171.
Heffler, B. (2001). Individual learning style and
the learning style inventory. Educational Studies,
27(3), 307-316.
Huff, C. (2002). Gender, software design, and
occupational equity. SIGCSE Bulletin, 34(2),
112-115.
Kafai, Y. (1998). Video game design by girls
and boys: Variability and consistency of gender
differences. In J. Cassell & H. Jenkins (Eds.),
From Barbie to Mortal Kombat: Gender and
computer games (pp. 90-114). Cambridge, MA:
MIT Press.
Lowenstein, G. (1994). The psychology of curiosity. Psychological Bulletin 116(1), 75-98.
Lunderberg, M., Fox, P., & Punchochar, J. (1994).
Highly confidence but wrong: Gender differences
and similarities in confidence judgments. Journal
of Educational Psychology, 86(1), 114-121.
Margolis, J., Fisher, A., & Miller F. (1999). Caring
about connections: Gender and computing. IEEE
Technology and Society, 18(4), 13-20.
Meyers-Levy, J., & Sternthal, B. (1991). Gender
differences in the use of message cues and judgments. Journal of Marketing Research, 28(1),
84-96.
Morris, M., & Dillon, A. (1997). The influence
of user perceptions on software utilization: Application and evaluation of a theoretical model
of technology acceptance. IEEE Software, 14(4),
58-76.
Nardi, B. (1993). A small matter of programming:
Perspectives on end-user computing. Cambridge,
MA: MIT Press.
ODonnell, E., & Johnson, E. (2001). Gender
effects on processing effort during analytical

31

Gender and End-User Computing

procedures. International Journal of Auditing,

5, 91-105.
Panko, R. (1998). What we know about spreadsheet errors. Journal of End User Computing,
10(2), 15-21.
Severiens, S., & ten Dam, G. (1997). Gender
and gender identity difference in learning styles.
Educational Psychology, 17(1/2), 79-93.
Simon, S. (2001). The impact of culture and
gender on Web sites: An empirical study. The
DATA BASE for Advances in Information Systems,
32(1), 18-37.
Tan, D., Czerwinski, M., & Robertson, G. (2003,
April 5-10). Women go with the (optical) flow.
ACM Conference on Human Factors in Computing
Systems (pp. 209-215), Fort Lauderdale, FL.
Torkzadeh, G., & Van Dyke, T. (2002). Effects
of training on Internet self-efficacy and computer
user attitudes. Computers in Human Behavior,
18, 479-494.
Venkatesh, V., & Morris, M. (2000). Why dont
men ever stop to ask for directions? Gender,
social influence, and their role in technology
acceptance and usage behavior. MIS Quarterly,
24(1), 115-139.

Wilson, A., Burnett, M., Beckwith, L., Granatir,

O., Caasburn, L., Cook, C., Durham, M., & Rothermel, G. (2003, April 5-10). Harnessing curiosity
to increase correctness in end-user programming.
ACM Conference on Human Factors in Computing
System (pp. 305-312). Fort Lauderdale, FL.

KEY Terms
End User: Users who are not trained programmers.
End-User Computing: Computer-supported
problem solving by end users, using systems such
as spreadsheets, multimedia authoring tools,
and graphical languages for demonstrating the
desired behavior.
End-User Programming: A term synonymous with end-user computing.
Gender HCI: Human-computer interaction
(HCI) work that takes gender differences into
account.
Overconfidence: Higher self-efficacy than is
warranted by a users abilities.
Self-Efficacy: Belief in ones capabilities to
perform a certain task.
Under Confidence: Lower self-efficacy than
is warranted by a users abilities.

This work was previously published in Encyclopedia of Gender and Information Technology, edited by E. Trauth, pp. 398-404,
copyright 2006 by Information Science Reference, formerly known as Idea Group Reference (an imprint of IGI Global)

32

33

Chapter 1.4

The Theoretical Framework of

Cognitive Informatics
Yingxu Wang
University of Calgary, Canada

Abstract
Cognitive Informatics (CI)
is a transdisciplinary
enquiry of the internal information processing
mechanisms and processes of the brain and
natural intelligence shared by almost all science
and engineering disciplines. This

article presents

an intensive review of the new field of CI. The

structure of the theoretical framework of CI is
described encompassing the Layered Reference
Model of the Brain (LRMB), the OAR model of
information representation, Natural Intelligence
(NI) vs. Artificial Intelligence (AI), Autonomic
Computing (AC) vs. imperative computing, CI
laws of software, the mechanism of human perception processes, the cognitive processes of formal
inferences, and the formal knowledge system.
Three types of new structures of mathematics,
Concept Algebra (CA), Real-Time Process Algebra (RTPA), and System Algebra (SA), are created
to enable rigorous treatment of cognitive processes
of the brain as well as knowledge representation and manipulation in a formal and coherent
framework. A wide range of applications of CI

in cognitive psychology, computing, knowledge

engineering, and software engineering has been
identified and discussed.

Introduction
The development of classical and contemporary
informatics, the cross fertilization between computer science, systems science, cybernetics, computer/software engineering, cognitive science,
knowledge engineering, and neuropsychology, has
led to an entire range of an extremely interesting
and new research field known as Cognitive Informatics (Wang, 2002a, 2003a, b, 2006b; Wang,
Johnston & Smith 2002; Wang & Kinsner, 2006).
Informatics is the science of information that
studies the nature of information; its processing,
and ways of transformation between information,
matter, and energy.
Definition 1. Cognitive Informatics (CI) is a
transdisciplinary enquiry of cognitive and information sciences that investigates the internal in-

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

The Theoretical Framework of Cognitive Informatics

formation processing mechanisms and processes

of the brain and natural intelligence, and their
engineering applications via an interdisciplinary
approach.
In many disciplines of human knowledge,
almost all of the hard problems yet to be solved
share a common root in the understanding of
the mechanisms of natural intelligence and the
cognitive processes of the brain. Therefore, CI is
a discipline that forges links between a number of
natural science and life science disciplines with
informatics and computing science.
The structure of the theoretical framework
of CI is described in Figure 1, which covers the
Information-Matter-Energy (IME) model (Wang,
2003b), the Layered Reference Model of the Brain

(LRMB) (Wang, Wang, Patel & Patel, 2006), the

Object-Attribute-Relation (OAR) model of information representation in the brain (Wang, 2006h;
Wang & Wang, 2006), the cognitive informatics
model of the brain (Wang, Liu, & Wang, 2003;
Wang & Wang, 2006), Natural Intelligence (NI)
(Wang, 2003b), Autonomic Computing (AC)
(Wang, 2004), Neural Informatics (NeI) (Wang,
2002a, 2003b, 2006b), CI laws of software (Wang,
2006f), the mechanisms of human perception processes (Wang, 2005a), the cognitive processes of
formal inferences (Wang, 2005c), and the formal
knowledge system (Wang, 2006g).
In this article, the theoretical framework of
CI is explained in the fundamental theories of
CI section. Three structures of new descriptive

Figure 1. The theoretical framework of CI

Th e Th eo retical F ram ew o rk o f C o g n itive In fo rm atics (C I)

CI
Theories (T)

CI
Applications (A)
Descriptive
Mathematics for
CI (M)

T1
The IME
model
T2
The LRMB
model

T7
CI laws of
software

M1
Concept
algebra (CA)

A1
Future generation
Computers

T3
The OAR
model

T8
Perception
processes

M2
RTPA

A2
Capacity of human
memory

T4
CI model of
the brain

T9
Inference
processes

M3
System algebra
(SA)

A3
Autonomic
computing

T5
Natural
intelligence

T10
The knowledge
system

T6
Neural
informatics
A9
Cognitive complexity
of software

34

A4
Cognitive properties
of knowledge
A5
Simulation of
cognitive behaviors

A8
Deductive semantics
of software

A7
CI foundations of
software engineering

A6
Agent
systems

The Theoretical Framework of Cognitive Informatics

mathematics such as Concept Algebra (CA),

Real-Time Process Algebra (RTPA), and System
Algebra (SA) are introduced in the denotational
mathematics for CI in order to rigorously deal with
knowledge and cognitive information representation and manipulation in a formal and coherent
framework. Applications of CI are discussed,
which covers cognitive computing, knowledge
engineering, and software engineering. Then,
it draws conclusions on the theories of CI, the
contemporary mathematics for CI, and their applications.

The Fundamental Theories

of CI
The fundamental theories of CI encompass 10
transdisciplinary areas and fundamental models,
T1 through T10, as identified in Figure 1. This
section presents an intensive review of the theories developed in CI, which form a foundation
for exploring the natural intelligence and their
applications in brain science, neural informatics,
computing, knowledge engineering, and software
engineering.

The Information-Matter-Energy
Model
Information is recognized as the third essence of
the natural world supplementing to matter and energy (Wang, 2003b), because the primary function
of the human brain is information processing.
Theorem 1. A generic worldview, the IME
model states that the natural world (NW) that
forms the context of human beings is a dual world:
one aspect of it is the physical or the concrete
world (PW), and the other is the abstract or the
perceptive world (AW), where matter (M) and
energy (E) are used to model the former, and
information (I) to the latter, that is:

NW PW || AW

= p(M, E)|| a(I)

= n(I, M, E)

(1)

where || denotes a parallel relation, and p, a, and
n, are functions that determine a certain PW, AW,
or NW, respectively, as illustrated in Figure 2.
According to the IME model, information
plays a vital role in connecting the physical world
with the abstract world. Models of the natural
world have been well studied in physics and other
natural sciences. However, the modeling of the
abstract world is still a fundamental issue yet to
be explored in cognitive informatics, computing,
software science, cognitive science, brain sciences, and knowledge engineering. Especially
the relationships between I-M-E and their transformations are deemed as one of the fundamental
questions in CI.
Corollary 1. The natural world NW(I, M, E),
particularly part of the abstract world, AW(I), is
cognized and perceived differently by individuals because of the uniqueness of perceptions and
mental contexts among people.
Corollary 1 indicates that although the physical world PW(M, E) is the same to everybody, the
natural world NW(I, M, E) is unique to different
individuals because the abstract world AW(I), as a
part of it, is subjective depending on the information an individual obtains and perceives.
Corollary 2. The principle of transformability
between IME states that, according to the IME
model, the three essences of the world are predicated to be transformable between each other as
described by the following generic functions f1
to f6:

I = f1 (M)

M = f2 (I) f1 -1(I)

I = f3 (E)

(2.1)
(2.2)
(2.3)

35

The Theoretical Framework of Cognitive Informatics

E = f4 (I) f3 -1(I)

E = f5 (M)

M = f6 (E) f5 -1(E)

(2.4)
(2.5)
(2.6)

where a question mark on the equal sign denotes

an uncertainty if there exists such a reverse function (Wang, 2003b).
Albert Einstein revealed Functions f5 and f6,
the relationship between matter (m) and energy
(E), in the form E = mC2, where C is the speed
of light. It is a great curiosity to explore what the
remaining relationships and forms of transformation between I-M-E will be. To a certain extent,
cognitive informatics is the science to seek possible solutions for f1 to f4. A clue to explore the
relations and transformability is believed in the
understanding of the natural intelligence and its
information processing mechanisms in CI.
Definition 2. Information in CI is defined as a
generic abstract model of properties or attributes
of the natural world that can be distinctly elicited,

generally abstracted, quantitatively represented,

and mentally processed.
Definition 3. The measurement of information, Ik,
is defined by the cost of code to abstractly represent
a given size of internal message X in the brain in
a digital system based on k, that is:
Ik

f :X

Sk

logk X

Definition 4. The metalevel representation of

information, Ib, is that when k = b = 2, that is:

Layer 6
Higher cognitive functions

Layer 5
Meta cognitive functions

Layer 4
Action

Subconscious
cognitive
processes

Layer 3
Perception

Layer 2
Memory

Layer 1
Sensation

36

(3)

where Ik is the content of information in a k-based

digital system, and Sk is the measurement scale
based on k. The unit of Ik is the number of k-based
digits (Wang, 2003b).
Equation 3 is a generic measure of information sizes. When a binary digital representation
system is adopted, that is k = b = 2, it becomes
the most practical one as follows.

Figure 3. LRMB model

Conscious
cognitive
processes

The Theoretical Framework of Cognitive Informatics

I b = f : X Sb
= log b X

(4)

where the unit of information, Ib, is a bit.

Note that the bit here is a concrete and deterministic unit, and it is no longer probability-based
as in conventional information theories (
Bell, 1953;
Shannon, 1948
). To a certain extent, computer
science and engineering is a branch of modern
informatics that studies machine representation
and processing of external information; while
CI is a branch of contemporary informatics that
studies internal information representation and
processing in the brain.
Theorem 2. The most fundamental form of information that can be represented and processed
is binary digit where k = b = 2.
Theorem 2 indicates that any form of information in the physical (natural) and abstract (mental) worlds can be unified on the basis of binary
data. This is the CI foundation of modern digital
computers and NI.

The Layered Reference Model

of the Brain
The LRMB (
Wang et al., 2006
) is developed to
explain the fundamental cognitive mechanisms
and processes of natural intelligence. Because
a
variety of life functions and cognitive processes
have been identified in CI, psychology, cognitive
science, brain science, and neurophilosophy, there
is a need to organize all the recurrent cognitive
processes in an integrated and coherent framework. The LRMB model explains the functional
mechanisms and cognitive processes of natural intelligence that encompasses 37 cognitive processes
at six layers known as the sensation, memory,
perception, action, metacognitive, and higher
cognitive layers from the bottom-up as shown
in Figure 3. LRMB elicits the core and highly
repetitive recurrent cognitive processes from a

huge variety of life functions, which may shed

light on the study of the fundamental mechanisms
and interactions of complicated mental processes,
particularly the relationships and interactions
between the inherited and the acquired life functions as well as those of the subconscious and
conscious cognitive processes.

The OAR Model of Information

Representation in the Brain
Investigation into the cognitive models of information and knowledge representation in the
brain is perceived to be one of the fundamental
research areas that help to unveil the mechanisms
of the brain. The Object-Attribute-Relation
(OAR) model (Wang, 2006h; Wang et al., 2003)
describes human memory, particularly the longterm memory, by using the relational metaphor,
rather than the traditional container metaphor that
used to be adopted in psychology, computing, and
information science. The OAR model shows that
human memory and knowledge are represented by
relations, that is, connections of synapses between
neurons, rather than by the neurons themselves as
the traditional container metaphor described. The
OAR model can be used to explain a wide range
of human information processing mechanisms
and cognitive processes.

The Cognitive Informatics Model

of the Brain
The human brain and its information processing
mechanisms are centred in CI. A

cognitive informatics model of the brain is proposed in Wang

and Wang (2006), which explains the natural
intelligence via interactions between the inherent (subconscious) and acquired (conscious) life
functions. The model demonstrates that memory
is the foundation for any natural intelligence.
Formalism in forms of mathematics, logic, and
rigorous treatment is introduced into the study
of cognitive and neural psychology and natural

37

The Theoretical Framework of Cognitive Informatics

informatics. Fundamental
cognitive mechanisms
of the brain, such as the architecture of the thinking engine, internal knowledge representation,
long-term memory establishment, and roles of
sleep in long-term memory development have
been investigated (Wang & Wang, 2006).

Natural Intelligence (NI)

Natural Intelligence (NI) is the domain of CI.
Software and computer systems are recognized as
a subset of intelligent behaviors of human beings
described by programmed instructive information
(Wang, 2003b; Wang & Kinsner, 2006). The relationship between Artificial Intelligence (AI) and
NI can be described by the following theorem.
Theorem 3. The law of compatible intelligent
capability states that artificial intelligence (AI)
is always a subset of the natural intelligence
(NI), that is:
AI NI

(5)

Theorem 3 indicates that AI is dominated by

NI. Therefore, one should not expect a computer
or a software system to solve a problem where
humans cannot. In other words, no AI or computing system may be designed and/or implemented
for a given problem where there is no solution
being known by human beings.

Neural Informatics (NeI)

Definition 5. Neural Informatics (NeI) is a new
interdisciplinary enquiry of the biological and
physiological representation of information and
knowledge in the brain at the neuron level and
their abstract mathematical models (Wang, 2004;
Wang & Wang, 2006).
NeI is a branch of CI, where memory is recognized as the foundation and platform of any
natural or artificial intelligence (Wang & Wang,
2006).

38

Definition 6. The Cognitive Models of Memory

(CMM) states that the architecture of human
memory is parallel configured by the Sensory Buffer Memory (SBM), Short-Term Memory (STM),
Long-Term Memory (LTM), and Action-Buffer
Memory (ABM), that is:


CMM SBM

|| STM
|| LTM
|| ABM
(6)
where the ABM is newly identified in Wang and
Wang (2006).
The major organ that accommodates memories in the brain is the cerebrum or the cerebral
cortex. In particular, the

association and premotor cortex in the frontal lobe, the temporal lobe,
sensory cortex in the frontal lobe, visual cortex
in the occipital lobe, primary motor cortex in the
frontal lobe, supplementary motor area in the
frontal lobe, and procedural memory in cerebellum (Wang & Wang, 2006).
The CMM model and the mapping of the four
types of human memory onto the physiological
organs in the brain reveal a set of fundamental
mechanisms of NeI. The OAR model of information/knowledge representation described in
the OAR model of information representation in
the brain section provides a generic description
of information/knowledge representation in the
brain (Wang, 2006h; Wang et al., 2003).
The theories of CI and NeI explain a number
of important questions in the study of NI. Enlightening conclusions derived in CI and NeI are
such as: (a) LTM establishment is a subconscious
process; (b) The long-term memory is established
during sleeping; (c) The major mechanism for
LTM establishment is by sleeping; (d) The general
acquisition cycle of LTM is equal to or longer than
24 hours; (e) The mechanism of LTM establishment is to update the entire memory of information
represented as an OAR model in the brain; and
(f) Eye movement and dreams play an important

The Theoretical Framework of Cognitive Informatics

role in LTM creation. The latest development in

CI and NeI has led to the determination of the
magnificent and expected capacity of human
memory as described in the Estimation of the
Capacity of Human Memory section.

Cognitive Informatics Laws of

Software
It is commonly conceived that software as an
artifact of human creativity is not constrained by
the laws and principles discovered in the physical
world. However, it is unknown what constrains
software. The new informatics metaphor proposed by the author in CI perceives software is a
type of instructive and behavioral information.
Based on this, it is asserted that software obeys
the laws of informatics. A comprehensive set of
19 CI laws for software have been established in
Wang (2006f), such as:
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.

Abstraction
Generality
Cumulativeness
Dependency on cognition
Three-dimensional behavior space known
as the object (O), space (S), and time (T)
Sharability
Dimensionless
Weightless
Transformability between I-M-E
Multiple representation forms
Multiple carrying media
Multiple transmission forms
Dependency on media
Dependency on energy
Wearless and time dependency
Conservation of entropy
Quality attributes of informatics
Susceptible to distortion
Scarcity

The informatics laws of software extend the

knowledge on the fundamental laws and proper-

ties of software where the conventional product

metaphor could not explain. Therefore, CI forms
one of the foundations of software engineering
and computing science.

Mechanisms of Human
Perception Processes
Definition 7. Perception is a set of interpretive
cognitive processes of the brain at the subconscious cognitive function layers that detects,
relates, interprets, and searches internal cognitive
information in the mind.
Perception may be considered as the sixth
sense of human beings, which almost all cognitive life functions rely on. Perception is also an
important cognitive function at the subconscious
layers that determines personality. In other words,
personality is a faculty of all subconscious life
functions and experience cumulated via conscious
life functions.
According to LRMB, the main cognitive
processes at the perception layer are emotion,
motivation, and attitude (Wang, 2005a). The relationship between the internal emotion, motivation,
attitude, and the embodied external behaviors can
be formally and quantitatively described by the
motivation/attitude-driven behavioral (MADB)
model (Wang & Wang, 2006), which demonstrates
that complicated psychological and cognitive
mental processes may be formally modeled and
rigorously described by mathematical means
(Wang, 2002b, 2003d, 2005c).

The Cognitive Processes of

Formal Inferences
Theoretical research is predominately an inductive process, while applied research is mainly a
deductive one. Both inference processes are based
on the cognitive process and means of abstraction.
Abstraction is a powerful means of philosophy
and mathematics. It is also a preeminent trait of
the human brain identified in CI studies (Wang,

39

The Theoretical Framework of Cognitive Informatics

2005c). All formal logical inferences and reasonings can only be carried out on the basis of
abstract properties shared by a given set of objects
under study.
Definition 8. Abstraction is a process to elicit a
subset of objects that shares a common property
from a given set of objects and to use the property
to identify and distinguish the subset from the
whole in order to facilitate reasoning.
Abstraction is a gifted capability of human
beings. Abstraction is a basic cognitive process
of the brain at the metacognitive layer according
to LRMB (Wang et al., 2006). Only by abstraction can important theorems and laws about the
objects under study be elicited and discovered
from a great variety of phenomena and empirical
observations in an area of inquiry.
Definition 9. Inferences are a formal cognitive
process that reasons a possible causality from
given premises based on known causal relations
between a pair of cause and effect proven true
by empirical arguments, theoretical inferences,
or statistical regulations.
Formal inferences may be classified into the
deductive, inductive, abductive, and analogical
categories (Wang, 2005c). Deduction is a cognitive process by which a specific conclusion
necessarily follows from a set of general premises. Induction is a cognitive process by which a
general conclusion is drawn from a set of specific
premises based on three designated samples in
reasoning or experimental evidences. Abduction
is a cognitive process by which an inference to
the best explanation or most likely reason of
an observation or event. Analogy is a cognitive
process by which an inference about the similarity of the same relations holds between different
domains or systems, and/or examines that if two
things agree in certain respects, then they probably agree in others. A summary of the formal

40

definitions of the five inference techniques is

shown in Table 1.
For seeking generality and universal truth,
either the objects or the relations can only be
abstractly described and rigorously inferred by
abstract models rather than real-world details.

The Formal Knowledge System

Mathematical thoughts (Jordan & Smith, 1997)
provide a successful paradigm to organize and
validate human knowledge, where once a truth
or a theorem is established, it is true until the
axioms or conditions that it stands for are changed
or extended. A proven truth or theorem in mathematics does not need to be argued each time
one uses it. This is the advantage and efficiency
of formal knowledge in science and engineering.
In other words, if any theory or conclusion may
be argued from time-to-time based on a wiser
idea or a trade-off, it is an empirical result rather
than a formal one.
The Framework of Formal Knowledge (FFK)
of mankind (Wang, 2006g) can be described as
shown in Figure 5. An FFK is centered by a set
of theories. A theory is a statement of how and
why certain objects, facts, or truths are related. All
objects in nature and their relations are constrained
by invariable laws, no matter if one observed them
or not at any given time. An empirical truth is a
truth based on or verifiable by observation, experiment, or experience. A theoretical proposition is
an assertion based on formal theories or logical
reasoning. Theoretical knowledge is a formalization of generic truth and proven abstracted
empirical knowledge. Theoretical knowledge
may be easier to acquire when it exists. However,
empirical knowledge is very difficult to be gained
without hands-on practice.
According to the FFK model, an immature
discipline of science and engineering is characterized by its body of knowledge not being
formalized. Instead of a set of proven theories,
the immature disciplines document a large set of

The Theoretical Framework of Cognitive Informatics

observed facts, phenomena, and their possible or

partially working explanations and hypotheses.
In such disciplines, researchers and practitioners
might be able to argue every informal conclusion
documented in natural languages from time-totime probably for hundreds of years, until it is
formally described in mathematical forms and
proved rigorously.
The disciplines of mathematics and physics
are successful paradigms that adopt the FFK
formal knowledge system. The key advantages of
the formal knowledge system are its stability and
efficiency. The former is a property of the formal
knowledge that once it is established and formally
proved, users who refers to it will no longer need
to reexamine or reprove it. The latter is a property
of formal knowledge that is exclusively true or
false that saves everybodys time from arguing
a proven theory.

Denotational Mathematics
for CI
The history of sciences and engineering shows
that new problems require new forms of math-

ematics. CI is a new discipline, and the problems

in it require new mathematical means that are
descriptive and precise in expressing and denoting human and system actions and behaviors.
Conventional analytic mathematics are unable
to solve the fundamental problems inherited in
CI and related disciplines such as neuroscience,
psychology, philosophy, computing, software
engineering, and knowledge engineering. Therefore, denotational mathematical structures and
means (Wang, 2006c) beyond mathematical logic
are yet to be sought.
Although there are various ways to express
facts, objects, notions, relations, actions, and
behaviors in natural languages, it is found in CI
that human and system behaviors may be classified into three basic categories known as to be,
to have, and to do. All mathematical means and
forms, in general, are an abstract and formal description of these three categories of expressibility
and their rules. Taking this view, mathematical
logic may be perceived as the abstract means
for describing to be, set theory describing to
have, and algebras, particularly process algebra,
describing to do.

Table 1. Definitions of formal inferences

.

Inference
technique

Formal description
Primitive form

Composite form

Usage

Abstraction

S, p e E S,
p(e)

To elicit a subset of elements

with a given generic property.

Deduction

x X, p(x) a
X, p(a)

(x X, p(x) q(x)) x X, p(x)

a X, p(a) (a X, p(a) q(a))

To derive a conclusion based

on a known and generic
premises.

Induction

((a X, P(a)) (k, k+1

X, (P(k) P(k+1)))
x X, P(x)

((a X, p(a) q(a)) (k, k+1

X, ((p(k) q(k)) (p(k+1)
q(k+1)))) x X, p(x) q(x)

To determine the generic

behavior of the given list or
sequence of recurring patterns
by three samples.

Abduction

(x X, p(x) q(x))
(a X, q(a) p(a))

(x X, p(x) q(x) r(x) q(x))

(a X, q(a) (p(a) r(a)))

To seek the most likely

cause(s) and reason(s) of an
observed phenomenon.

Analogy

a X, p(a) b
X, p(b)

(a X, p(a) q(a)) (b X, p(b)

q(b))

To predict a similar
phenomenon or consequence
based on a known
observation.

41

The Theoretical Framework of Cognitive Informatics

Figure 4. The framework of formal knowledge (FFK)

The Formal Knowledge System

Discipline

Doctrine
Definitions

Propositions

Hypotheses

Theories

Concepts
Factors

Lemmas

Corollaries

Truths
Phenomena

Theorems

Empirical verifications
Formal proofs

Laws

Arguments

Principles

Instances

Rules
Models

Case studies
Statistical norms

Methodologies
Algorithms

Theorem 4. The utility of mathematics is the

means and rules to express thought rigorously and
generically at a higher level of abstraction.
Three types of new mathematics, Concept
Algebra (CA), Real-Time Process Algebra
(RTPA), and System Algebra (SA), are created
in CI to enable rigorous treatment of knowledge
representation and manipulation in a formal and
coherent framework. The three new structures
of contemporary mathematics have extended
the abstract objects under study in mathematics
from basic mathematical entities of numbers
and sets to a higher level, that is, concepts, behavioral processes, and systems. A wide range
of applications of the denotational mathematics
in the context of CI has been identified (Wang,
2002b, 2006d, e).

42

Concept Algebra
A concept is a cognitive unit (
Ganter & Wille,
1999
; Quillian,

1968;
Wang, 2006e) by which
the meanings and semantics of a real-world or an
abstract entity may be represented and embodied
based on the OAR model.
Definition 10. An abstract concept c is a 5-tuple,
that is:



c
i
o

(7)
c (O, A, R , R , R )
where
O is a nonempty set of object of the concept,
O = {o1, o2, , om} = U, where U denotes
a power set of U.

A is a nonempty set of attributes, A = {a1,
a2, , an} = M.

The Theoretical Framework of Cognitive Informatics

Rc O A is a set of internal relations.

Ri C C is a set of input relations, where
C is a set of external concepts.
Ro C C is a set of output relations.

A structural concept model of c = (O, A, Rc,

R , Ro) can be illustrated in Figure 6, where c, A,
O, and R, R = {Rc, Ri, Ro}, denote the concept, its
attributes, objects, and internal/external relations,
respectively.

tion, specification, and instantiation as shown in

Figure 6 and Table 2 (Wang, 2006e). In Figure
6, R = {Rc, Ri, Ro}, and all nine associations describe composing rules among concepts, except
instantiation that is a relation between a concept
and a specific object.

Definition 11. Concept algebra is a new mathematical structure for the formal treatment of
abstract concepts and their algebraic relations,
operations, and associative rules for composing complex concepts and knowledge (Wang,
2006e).
Concept algebra deals with the algebraic relations and associational rules of abstract concepts.
The associations of concepts form a foundation
to denote complicated relations between concepts
in knowledge representation. The associations
among concepts can be classified into nine categories, such as inheritance, extension, tailoring,
substitute, composition, decomposition, aggrega-

Definition 12. A generic knowledge K is an n-nary

relation Rk among a set of n multiple concepts in
C, that is
n

K = Rk : ( XCi ) C
(8)
i=1

where

UC
i=1

= C , and
+

, , , , , } .
Rk = {, , ,

In Definition 12, the relation Rk is one of the

concept operations in CA as defined in Table
2 (Wang, 2006e) that serves as the knowledge
composing rules.
Definition 13. A concept network CN is a hierarchical network of concepts interlinked by the set
of nine associations defined in CA, that is:

Figure 6. The nine concept association operations as knowledge composing rules

c1

Inheritance
Extension

A1

Tailoring

R1

Substitute

c2

o21

Instantiation

A2

R21

A21

R2

Composition
O1

Decomposition

O2

Aggregation
Specification

43

The Theoretical Framework of Cognitive Informatics

i=1

i= j

CN = R : XCi XC j

(9)

where Rk R.
Because the relations between concepts are
transitive, the generic topology of knowledge is
a hierarchical concept network. The advantages
of the hierarchical knowledge architecture K
in the form of concept networks are as follows:
(a) Dynamic: The knowledge networks may be
updated dynamically along with information
acquisition and learning without destructing the
existing concept nodes and relational links. (b)
Evolvable: The knowledge networks may grow
adaptively without changing the overall and existing structure of the hierarchical network.
A summary of the algebraic relations and operations of concepts defined in CA are provided
in Table 2.

Real-Time Process Algebra (RTPA)

A key metaphor in system modeling, specification, and description is that a software system can
be perceived and described as the composition
of a set of interacting processes. Hoare (1985),
Milner (1989), and others developed various algebraic approaches to represent communicating and
concurrent systems, known as process algebra.
A process algebra is a set of formal notations
and rules for describing algebraic relations of
software processes. Real-Time Process Algebra
(Wang, 2002b, 2005b) extends process algebra to
time/event, architecture, and system dispatching
manipulations in order to formally describe and
specify architectures and behaviors of software
systems. A process in RTPA is a computational
operation that transforms a system from a state
to another by changing its inputs, outputs, and/
or internal variables. A process can be a single
metaprocess or a complex process formed by using
the process combination rules of RTPA known
as process relations.

44

Definition 14. Real-Time Process Algebra is a

set of formal notations and rules for describing
algebraic and real-time relations of software
processes.

RTPA models 17 metaprocesses and 17 process relations. A
metaprocess is an elementary and
primary process that serves as a common and basic
building block for a software system. Complex
processes can be derived from metaprocesses by
a set of process relations that serves as process
combinatory rules. Detailed semantics of RTPA
may be referred to in Wang (2002b).
Program modeling is on coordination of
computational behaviors with given data objects.
Behavioral or instructive knowledge can be modeled by RTPA. A
generic program model can be
described by a formal treatment of statements,
processes, and complex processes from the bottom-up in the program hierarchy.
Definition 15. A process P is a composed listing
and a logical combination of n metastatements pi
and pj, 1 i < n, 1 < j m = n+1, according to
certain composing relations rij, that is:



n 1


P = ( pi rij p j ), j = i + 1
i =1

(10)

= (...((( p1 ) r12 p2 ) r23 p3 ) ... rn 1,n pn )

where the big-R notation (Wang, 2002b, 2006i)

is adopted to describes the nature of processes as
the building blocks of programs.
Definition 16. A program P is a composition of
a finite set of m processes according to the time-,
event-, and interrupt-based process dispatching
rules, that is:
P=

R(@ e P )
k =1

(11)

Equations 9.1 and 10.1 indicate that a program

is an embedded relational algebraic entity. A

The Theoretical Framework of Cognitive Informatics

statement p in a program is an instantiation of a

metainstruction of a programming language that
executes a basic unit of coherent function and
leads to a predictable behavior.
Theorem 5. The embedded relational model
(ERM) states that a software system or a program P is a set of complex embedded relational
processes, in which all previous processes of a
given process form the context of the current
process, that is:
m

P = (@ e P )
k =1

(12)

n 1

=
[@ e
( p (k ) r (k ) p (k ))], j = i + 1
k =1

i =1

ij

Definition 17. A metaprocess is the most basic and

elementary processes in computing that cannot
be broken up further. The set of metaprocesses P
encompasses 17 fundamental primitive operations
in computing as follows:

P ={:=, , , , , , , |, |,, , , ,
!, , ,}

(14)

The definitions, syntaxes, and formal semantics of each of the metaprocesses and process
relations may be referred to RTPA (Wang, 2002b,
2006f). A complex process and a program can
be derived from the metaprocesses by the set of
algebraic process relations. Therefore, a program
is a set of embedded relational processes as described in Theorem 5.
A summary of the metaprocesses and their
algebraic operations in RTPA are provided in
Table 2.

System Algebra (SA)

ERM presented in Theorem 5 provides a unified mathematical model of programs (Wang,

2006a) for the first time, which reveals that a
program is a finite and nonempty set of embedded binary relations between a current statement
and all previous ones that formed the semantic
context or environment of computing.

R = {, , |, ||, ,,, , , ||,

, |||, , , t, e, i}

(13)

Definition 18. A process relation is a composing rule for constructing complex processes by
using the metaprocesses. The process relations
R of RTPA are a set of 17 composing operations
and rules to built larger architectural components
and complex system behaviors using the metaprocesses, that is:

Systems are the most complicated entities and

phenomena in the physical, information, and
social worlds across all science and engineering
disciplines (Klir, 1992; von Bertalanffy, 1952;
Wang, 2006d). Systems are needed because the
physical and/or cognitive power of an individual
component or person is not enough to carry out
a work or solving a problem. An abstract system
is a collection of coherent and interactive entities
that has stable functions and clear boundary with
external environment. An abstract system forms
the generic model of various real world systems
and represents the most common characteristics
and properties of them.
Definition 19. System algebra is a new abstract
mathematical structure that provides an algebraic
treatment of abstract systems as well as their relations and operational rules for forming complex
systems (Wang, 2006d).
Abstract systems can be classified into two
categories known as the closed and open systems.
Most practical and useful systems in nature are
open systems in which there are interactions between the system and its environment. However,
for understanding easily, the closed system is
introduced first.

45

The Theoretical Framework of Cognitive Informatics

Definition 20. A closed systemis a 4-tuple, that

is :
(15)
S = (C, R, B, )

Definition 21. An open system S is a 7-tuple,

that is:
S = (C, R, B, , )
= (C, Rc, Ri, Ro, B, , )

where

where the extensions of entities beyond the closed

system are as follows:

C is a nonempty set of components of the

system, C = {c1, c2, , cn}.
R is a nonempty set of relations between
pairs of the components in the system, R =
{r1, r2, , r m}, R C C.
B is a set of behaviors (or functions), B =
{b1, b2, , bp}.
is a set of constraints on the memberships
of components, the conditions of relations,
and the scopes of behaviors, = {1, 2,
, q}.

is the environment of S with a nonempty

set of components C outside C.
Rc C C is a set of internal relations.
Ri C C is a set of external input relations.
Ro C C is a set of external output relations.

An open system S = (C, Rc, Ri, Ro, B, , ) can

be illustrated in Figure 7 (Wang, 2006d).

Most practical systems in the real world are

not closed. That is, they need to interact with
external world known as the environment in
order to exchange energy, matter, and/or information. Such systems are called open systems.
Typical interactions between an open system and
the environment are inputs and outputs.

Theorem 6. The equivalence between open and

closed systems states that an open system S is
equivalent to a closed system , or vice verse,
when its environment or is conjoined, respectively, that is:

S = S S

(17)

S = S S

Figure 7. The abstract model of an open system

U

Ri1

R1
C1

B1

Ro1

1
S

Rc1

Rc1
R2

Ri2

C2

B2
2

46

(16)

Ro2

The Theoretical Framework of Cognitive Informatics

According to Theorem 6, any subsystem of a

closed system is an open system S. That is, any
supersystem S of a given set of n open systems Sk,
plus their environments k, 1 k n, is a closed
system. The algebraic relations and operations of
systems in SA are summarized in Table 2.

There was a myth on an ideal system in conventional systems theory that supposes the work
down by the ideal system W(S) may be greater
than the sum of all its components W(ei), that is:
n

W ( S ) W (ei ).
i =1

Theorem 7. The Wangs first law of system science, system fusion, states that system conjunction
or composition between two systems S1 and S2
creates new relations R12 and/or new behaviors
(functions) B12 that are solely a property of the
new supersystem S determined by the sizes of
the two intersected component sets #(C1) and
#(C2), that is:

According to Theorems 7 and 8, the ideal

system utility is impossible to achieve.
A summary of the algebraic operations and
their notations in CA, RTPA, and SA is provided
in Table 2. Details may be referred to in Wang
(2006d, g).

(18)
R12 = #(R) - (#(R1) + #(R2))
= (#(C1 + C2))2 - ((#(C1))2 +(#(C2))2)
= 2 (#(C1) #(C2))

The last two sections have reviewed the latest

development of fundamental researches in CI,
particularly its theoretical framework and descriptive mathematics. A wide range of applications
of CI has been identified in multidisciplinary and
transdisciplinary areas, such as: (1) The architecture of future generation computers; (2) Estimation
the capacity of human memory; (3) Autonomic
computing; (4) Cognitive properties of information, data, knowledge, and skills in knowledge
engineering; (5) Simulation of human cognitive
behaviors using descriptive mathematics; (6)
Agent systems; (7) CI foundations of software
engineering; (8) Deductive semantics of software;
and (9) Cognitive complexity of software.

The discovery in Theorem 7 reveals that the

mathematical explanation of system utilities is
the newly gained relations R12 and/or behaviors
(functions) B12 during the conjunction of two
systems or subsystems. The empirical awareness
of this key system property has been intuitively
or qualitatively observed for centuries. However,
Theorem 7 is the first rigorous explanation of
the mechanism of system gains during system
conjunctions and compositions. According to
Theorem 7, the maximum incremental or system
gain equals to the number of by-directly interconnection between all components in both S1 and S2,
that is, 2(#(C1) #(C2)).
Theorem 8. The Wangs 2nd law of system science, the maximum system gain, states that work
done by a system is always larger than any of its
components, but is less than or is equal to the sum
of those of its components, that is:
n

W ( S ) W (ei ),
1

(19)

i =1

W ( S ) > max(W (e )), e E

i
i
S

Applications of CI

The Architecture of Future

Generation Computers
Conventional machines are invented to extend
human physical capability, while modern information processing machines, such as computers, communication networks, and robots, are
developed for extending human intelligence,
memory, and the capacity for information processing (Wang, 2004). Recent advances in CI
provide formal description of an entire set of
cognitive processes of the brain (Wang et al.,

47

The Theoretical Framework of Cognitive Informatics

Table 2. Taxonomy of contemporary mathematics for knowledge representation and manipulation

Operations

Concept
Algebra

System
Algebra

Super/sub relation

Related /
independent

Real-Time Process Algebra

Meta Processes
Assignment

:=

/ /

Evaluation

Equivalent

Addressing

Branch

Consistent

Memory allocation

Switch

||

Memory release

While-loop

Repeat-loop

Overlapped

Read

Sequence
Jump

Conjunction

Elicitation

Comparison

Input

Recursion

Definition

Output

Procedure call

Difference
Inheritance

Write

Timing

For-loop

Parallel

||

Concurrence

Extension

Increase

Interleave

Tailoring

Decrease

Substitute

Exception
detection

Interrupt

Time-driven dispatch

Event-driven
dispatch

Interrupt-driven
dispatch

Skip

Composition

Decomposition

Stop

Aggregation/
generalization

System

Specification

Instantiation

P i p e l i n e

Duration

2006). The fundamental research in CI also creates an enriched set of contemporary denotational
mathematics (Wang, 2006c), for dealing with
the extremely complicated objects and problems
in natural intelligence, neural informatics, and
knowledge manipulation.
The theory and philosophy behind the next
generation computers and computing methodologies are CI (Wang, 2003b, 2004). It is commonly

48

Relational Operations

|||

believed that the future-generation computers,

known as the cognitive computers, will adopt
non-von Neumann (von Neumann, 1946) architectures. The key requirements for implementing
a conventional stored-program controlled computer are the generalization of common computing
architectures and the computer is able to interpret
the data loaded in memory as computing instructions. These are the essences of stored-program

The Theoretical Framework of Cognitive Informatics

Figure 8. The architecture of a cognitive machine

IE

Enquiries

LTM

KMU

LTM

Knoledge

LTM

BMU

ABM

Behaviors

ABM

EMU

LTM

Experience

ABM

SMU

ABM

Skills

CM = IE || PE

Stimuli

Interactions

SBM

BPU

ABM

Behaviors

SBM

EPU

LTM

Experience

PE
The Cognitive Machine (CM)

controlled computers known as the von Neumann

(1946) architecture. Von Neumann elicited five
fundamental and essential components to implement general-purpose programmable digital
computers in order to embody the concept of
stored-program-controlled computers.

an extended Turing machine (TM), where the

power and functionality of all components of TM
including the control unit (with wired instructions), the tape (memory), and the head of I/O,
are greatly enhanced and extended with more
powerful instructions and I/O capacity.

Definition 22. A von Neumann Architecture

(VNA) of computers is a 5-tuple that consists of the
components: (a) the arithmetic-logic unit (ALU),
(b) the control unit (CU) with a program counter
(PC), (c) a memory (M), (d) a set of input/output
(I/O) devices, and (e) a bus (B) that provides the
data path between these components, that is:

Definition 24. A Wang Architecture (WA) of

computers, known as the Cognitive Machine as
shown in Figure 8, is a parallel structure encompassing an Inference Engine (IE) and a Perception
Engine (PE) (Wang, 2006b, g), that is:

VNA (ALU, CU, M, I/O, B)

(20)

Definition 23. Conventional computers with

VNA are aimed at stored-program-controlled
data processing based on mathematical logic and
Boolean algebra.
A VNA computer is centric by the bus and
characterized by the all purpose memory for
both data and instructions. A VNA machine is

WA (IE || PE)
= ( KMU// The knowledge manipulation
unit
|| BMU//The behavior manipulation unit
|| EMU // The experience manipulation
unit
|| SMU// The skill manipulation unit
)
|| ( BPU // The behavior perception unit
|| EPU // The experience perception unit
)

(21)

49

The Theoretical Framework of Cognitive Informatics

As shown in Figure 8 and Equation 21, WA

computers are not centered by a CPU for data
manipulation as the VNA computers do. The WA
computers are centered by the concurrent IE and
PE for cognitive learning and autonomic perception based on abstract concept inferences and
empirical stimuli perception. The IE is designed
for concept/knowledge manipulation according
to concept algebra (Wang, 2006e), particularly
the nine concept operations for knowledge acquisition, creation, and manipulation. The PE is
designed for feeling and perception processing
according to RTPA (Wang, 2002b) and the formally described cognitive process models of the
perception layers as defined in the LRMB model
(Wang et al., 2006).
Definition 25. Cognitive computers with WA
are aimed at cognitive and perceptive concept/
knowledge processing based on contemporary
denotational mathematics, that is, CA, RTPA,
and SA.
As that of mathematical logic and Boolean
algebra are the mathematical foundations of VNA
computers. The mathematical foundations of WA
computers are based on denotational mathematics (Wang, 2006b, c). As described in the LRMB
reference model (Wang et al., 2006), since all the
37 fundamental cognitive processes of human
brains can be formally described in CA and RTPA
(Wang, 2002b, 2006e). In other words, they are
simulatable and executable by the WA-based
cognitive computers.

Estimation of the Capacity of

Human Memory
Despite the fact that the number of neurons in
the brain has been identified in cognitive and
neural sciences, the magnitude of human memory
capacity is still unknown. According to the OAR
model, a recent discovery in CI is that the upper
bound of memory capacity of the human brain is
in the order of 108,432 bits (Wang et al., 2003). The
determination of the magnitude of human memory
capacity is not only theoretically significant in CI,
but also practically useful to unveil the human
potential, as well as the gaps between the natural
and machine intelligence. This result indicates
that the next generation computer memory systems may be built according to the OAR model
rather than the traditional container metaphor,
because the former is more powerful, flexible,
and efficient to generate a tremendous memory
capacity by using limited number of neurons in
the brain or hardware cells in the next generation
computers.

Autonomic Computing
The approaches to implement intelligent systems
can be classified into those of biological organisms,
silicon automata, and computing systems. Based
on CI studies, autonomic computing (Wang, 2004)
is proposed as a new and advanced computing
technique built upon the routine, algorithmic, and
adaptive systems as shown in Table 3.
The approaches to computing can be classified into two categories known as imperative and
autonomic computing. Corresponding to these,

Table 3. Classification of computing systems

Constant
Variable
Type of behavior

Event (I)

50

Behavior (O)
Constant
Variable
Routine
Adaptive
Algorithmic
Autonomic
Deterministic
Nondeterministic

The Theoretical Framework of Cognitive Informatics

computing systems may be implemented as imperative or autonomic computing systems.

Definition 26. An imperative computing system
is a passive system that implements deterministic, context-free, and stored-program controlled
behaviors.
Definition 27. An autonomic computing system
is an intelligent system that autonomously carries out robotic and interactive actions based on
goal- and event-driven mechanisms.
The imperative computing system is a traditional passive system that implements deterministic, context-free, and stored-program controlled
behaviors, where a behavior is defined as a set of
observable actions of a given computing system.
The autonomic computing system is an active
system that implements nondeterministic, context-dependent, and adaptive behaviors, which do
not rely on instructive and procedural information,
but are dependent on internal status and willingness that formed by long-term historical events
and current rational or emotional goals.
The first three categories of computing techniques as shown in Table 3 are imperative. In
contrast, the autonomic computing systems are
an active system that implements nondeterministic, context-sensitive, and adaptive behaviors.
Autonomic computing does not rely on imperative
and procedural instructions, but are dependent
on perceptions and inferences based on internal
goals as revealed in CI.

Cognitive Properties of Knowledge

Almost all modern disciplines of science and engineering deal with information and knowledge.
According to CI theories, cognitive information
may be classified into four categories known as
knowledge, behaviors, experience, and skills as
shown in Table 4.
Definition 28. The taxonomy of cognitive information is determined by its types of inputs and
outputs to and from the brain during learning
and information processing, where both inputs
and outputs can be either abstract information
(concept) or empirical information (actions).
It is noteworthy that the approaches to acquire
knowledge/behaviors and experience/skills are
fundamentally different. The former may be obtained either directly based on hands-on activities
or indirectly by reading, while the latter can never
be acquired indirectly.
According to Table 4, the following important
conclusions on information manipulation and
learning for both human and machine systems
can be derived.
Theorem 9. The principle of information acquisition states that there are four sufficient categories
of learning known as those of knowledge, behaviors, experience, and skills.
Theorem 9 indicates that learning theories
and their implementation in autonomic and intelligent systems should study all four categories of

Table 4. Types of cognitive information

Type of Output
Type
of Input

Ways of
Acquisition

Abstract Concept

Abstract Concept Empirical Action

Knowledge
Behavior

Direct or indirect

Empirical Action

Experience

Direct only

Skill

51

The Theoretical Framework of Cognitive Informatics

cognitive information acquisitions, particularly

behaviors, experience, and skills rather than only
focusing on knowledge.
Corollary 3. All the four categories of information
can be acquired directly by an individual.
Corollary 4. Knowledge and behaviors can be
learnt indirectly by inputting abstract information, while experience and skills must be learned
directly by hands-on or empirical actions.
The above theory of CI lays an important
foundation for learning theories and pedagogy
(Wang, 2004, 2006e). Based on the fundamental work, the IE and PE of cognitive computers
working as a virtual brain can be implemented on
WA-based cognitive computers and be simulated
on VNA-based conventional computers.

Simulation of Human Cognitive

Behaviors using the Contemporary
Mathematics
The contemporary denotational mathematics as
described in The Denotational Mathematics for CI
section, particularly CA and RTPA, may be used
to simulate the cognitive processes of the brain as
modeled in LRMB (Wang et al., 2006). Most of
the 37 cognitive processes identified in LRMB,
such as the learning (Wang, 2006e), reasoning
(Wang, 2006b), decision making (Wang et al.,
2004), and comprehension (Wang & Gafurov,
2003) processes, have been rigorously modeled
and described in RTPA and CA. Based on the
fundamental work, the inference engineering and
perception engine of a virtual brain can be implemented on cognitive computers or be simulated
on conventional computers. In the former case, a
working prototype of a fully autonomic computer
will be realized on the basis of CI theories.

52

Agent Systems
Definition 29. A software agent is an intelligent
software system that autonomously carries out
robotic and interactive applications based on
goal-driven mechanisms (Wang, 2003c).
Because a software agent may be perceived as
an application-specific virtual brain (see Theorem
3), behaviors of an agent are mirrored human
behaviors. The fundamental characteristics of
agent-based systems are autonomic computing,
goal-driven action-generation, knowledge-based
machine learning. In recent CI research, perceptivity is recognized as the sixth sense that serves
the brain as the thinking engine and the kernel of
the natural intelligence. Perceptivity implements
self-consciousness inside the abstract memories of
the brain. Almost all cognitive life functions rely
on perceptivity such as consciousness, memory
searching, motivation, willingness, goal setting,
emotion, sense of spatiality, and sense of motion.
The brain may be stimulated by external and
internal information, which can be classified as
willingness-driven (internal events such as goals,
motivation, and emotions), event-driven (external
events), and time-driven (mainly external events
triggered by an external clock). Unlike a computer,
the brain works in two approaches: the internal
willingness-driven processes, and the external
event- and time-driven processes. The external
information and events are the major sources
that drive the brain, particularly for conscious
life functions.
Recent research in CI reveals that the foundations of agent technologies and autonomic
computing are CI, particularly goal-driven action generation techniques (Wang, 2003c). The
LRMB model (Wang et al., 2006) described in
the Layered Reference Model of the Brain section
may be used as a reference model for agent-based
technologies. This is a fundamental view toward

The Theoretical Framework of Cognitive Informatics

the formal description and modeling of architectures and behaviors of agent systems, which are
created to do something repeatable in context,
to extend human capability, reachability, and/or
memory capacity. It is found that both human
and software behaviors can be described by a
3-dimensional representative model comprising action, time, and space. For agent system
behaviors, the three dimensions are known as
mathematical operations, event/process timing,
and memory manipulation (Wang, 2006g). The
3-D behavioral space of agents can be formally
described by RTPA that serves as an expressive
mathematical means for describing thoughts and
notions of dynamic system behaviors as a series
of actions and cognitive processes.

CI Foundations of
Software Engineering
Software is an intellectual artifact and a kind of
instructive information that provides a solution for
a repeatable computer application, which enables
existing tasks to be done easier, faster, and smarter,
or which provides innovative applications for the
industries and daily life. Large-scale software
systems are highly complicated systems that have
never been handled or experienced precedent by
mankind.
The fundamental cognitive characteristics
of software engineering have been identified as
follows (Wang, 2006g):

The inherent complexity and diversity

The difficulty of establishing and stabilizing
requirements
The changeability or malleability of system
behavior
The abstraction and intangibility of software
products
The requirement of varying problem domain
knowledge
The non-deterministic and polysolvability
in design

The polyglotics and polymorphism in implementation

The dependability of interactions among
software, hardware, and human beings

The above list forms a set of fundamental

constraints for software engineering, identified
as the cognitive constraints of intangibility, complexity, indeterminacy, diversity, polymorphism,
inexpressiveness, inexplicit embodiment, and unquantifiable quality measures (Wang, 2006g).
A set of psychological requirements for
software engineers has been identified, such as:
(a) Abstract-level thinking; (b) Imagination of
dynamic behaviors with static descriptions; (c)
Organization capability; (d) Cooperative attitude
in team work; (e) Long-period focus of attentions;
(f) Preciseness; (g) Reliability; and (h) Expressive
capability in communication.

Deductive Semantics of Software

Deduction is a reasoning process that discovers
new knowledge or derives a specific conclusion
based on generic premises such as abstract rules
or principles. In order to provide an algebraic
treatment of the semantics of program and human
cognitive processes, a new type of formal semantics known as deductive semantics is developed
(Wang, 2006f, g).
Definition 30. Deductive semantics is a formal
semantics that deduces the semantics of a program
from a generic abstract semantic function to the
concrete semantics, which are embodied onto the
changes of status of a finite set of variables constituting the semantic environment of computing
(Wang, 2006g).
Theorem 10. The semantics of a statement p, (p),
on a given semantic environment in deductive
semantics is a double partial differential of the
semantic function, f (p) = , on the sets of variables
S and executing steps T, that is:

53

The Theoretical Framework of Cognitive Informatics

(p) =

=
=

Cognitive Complexity of Software

2
2
fq ( p ) =
v p (t , s )
t s
t s

#T ( p ) # S ( p )

R Rv

(t , s j )

p i
i =0
j =1
1 #{s1 , s 2 , ..., s m }

R
i =0

j =1

v p (ti , s j )

s1 s 2 sm

t
v01 v02 v0 m
= 0

(t , t ] v
v1m
0 1 11 v12

(22)

t denotes the discrete time immediately

where
before and after the execution of p during (t0,
t1], and # is the cardinal calculus that counts the
number of elements in a given set, that is n =
#T(p) and m=#S(p).
The first partial differential in Equation 22
selects all related variable S(p) of the statement
p from . The second partial differential selects
a set of discrete steps of ps execution T(p) from
. According to Theorem 10, the semantics of a
statement can be reduced onto a semantic function
that results in a 2-D matrix with the changes of
values for all variables over time along program
execution.
Deductive semantics perceives that the carriers
of software semantics are a finite set of variables
declared in a given program. Therefore, software
semantics can be reduced onto the changes of values of
these variables. The deductive mathematical models
of semantics and the semantic environment at various
composing levels of systems are formally described.
Properties of software semantics and relationships between the software behavioral space and the semantic
environment are discussed. Deductive semantics is
applied in the formal definitions and explanations of
the semantic rules of a comprehensive set of software
static and dynamic behaviors as modeled in RTPA.
Deductive semantics can be used to define abstract
and concrete semantics of software and cognitive
systems, and facilitate software comprehension and
recognition by semantic analyses.

54

The estimation and measurement of functional

complexity of software are an age-long problem in software engineering. The cognitive
complexity of software (Wang, 2006j) is a new
measurement for cross-platform analysis of
complexities, sizes, and comprehension effort
of software specifications and implementations
in the phases of design, implementation, and
maintenance in software engineering. This work
reveals that the cognitive complexity of software
is a product of its architectural and operational
complexities on the basis of deductive semantics
and the abstract system theory. Ten fundamental
basic control structures (BCSs) are elicited from
software architectural/behavioral specifications
and descriptions. The cognitive weights of those
BCSs are derived and calibrated via a series of
psychological experiments. Based on this work,
the cognitive complexity of software systems
can be rigorously and accurately measured and
analyzed. Comparative case studies demonstrate
that the cognitive complexity is highly distinguishable in software functional complexity and size
measurement in software engineering.
On the basis of the ERM model described
in Theorem 5 and the deductive semantics of
software presented in The deductive semantics
of software section, the finding on the cognitive
complexity of software is obtained as follows.
Theorem 11. The sum of the cognitive weights
of all rij, w(rij), in the ERM model determines
the operational complexity of a software system
Cop, that is:

n 1
(23)
Cop =
w(rij ), j = i + 1

i =1

A set of psychological experiments has been

carried out in undergraduate and graduate classes
in software engineering. Based on 126 experiment
results, the equivalent cognitive weights of the

The Theoretical Framework of Cognitive Informatics

Table 5. Calibrated cognitive weights of BCSs

BCS
1
2
3

RTPA Notation

|
| |

Description
Sequence
Branch
Switch

For-loop

Repeat-loop

While-loop

7
8
9

|| or

Function call
Recursion
Parallel

7
11
15

10

Interrupt

22

10 fundamental BCSs are statistically calibrated

as summarized in Table 5 (Wang, 2006j), where
the relative cognitive weight of the sequential
structures is assumed one, that is, w1 = 1.
According to deductive semantics, the complexity of a software system, or its semantic space,
is determined not only by the number of operations, but also by the number of data objects.
Theorem 12. The cognitive complexity Cc(S) of a
software system S is a product of the operational
complexity Cop(S) and the architectural complexity Ca(S), that is:
Cc ( S ) = Cop ( S ) Ca ( S )
nC #(Cs (Ck ))

= {

k =1
nCLM

i =1

w(k , i )}

{ OBJ(CLM k ) +
k =1

(24)
nC

OBJ(Ck )}
k =1

[FO]

Based on Theorem 12, the following corollary

can be derived.
Corollary 5. The cognitive complexity of a software system is proportional to both its operational
and structural complexities. That is, the more the
architectural data objects and the higher the op-

Calibrated cognitive weight

1
3
4

erational complicity onto these objects, the larger

the cognitive complexity of the system.
Based on Theorem 11, the cognitive complexities of four typical software components (Wang,
2006j) have been comparatively analyzes as
summarized in Table 6. For enabling comparative analyses, data based on existing complexity
measures, such as time, cyclomatic, and symbolic
(LOC) complexities, are also contrasted in Table
6.
Observing Table 6 it can be seen that the first
three traditional measurements cannot actually
reflect the real complexity of software systems
in software design, representation, cognition,
comprehension, and maintenance. It is found
that (a) Although four example systems are with
similar symbolic complexities, their operational
and functional complexities are greatly different. This indicates that the symbolic complexity
cannot be used to represent the operational or
functional complexity of software systems. (b)
The symbolic complexity (LOC) does not represent the throughput or the input size of problems.
(c) The time complexity does not work well for
a system where there are no loops and dominate
operations, because in theory that all statements

55

The Theoretical Framework of Cognitive Informatics

Table 6. Measurement of software system complexities

System

Time
Cyclomatic Symbolic
complexity complexity complexity
(Ct [OP])
(Cm [-])
(Cs [LOC])

IBS (a)
IBS (b)
MaxFinder
SIS_Sort

O(n)
O(n)
O(m+n)

1
2
2
5

7
8
5
8

in linear structures are treated as zero in this

measure no matter how long they are. In addition, time complexity cannot distinguish the real
complexities of systems with the same asymptotic
function, such as in Case 2 (IBS (b)) and Case 3
(Maxfinder). (d) The cognitive complexity is an
ideal measure of software functional complexities
and sizes, because it represents the real semantic
complexity by integrating both the operational
and architectural complexities in a coherent
measure. For example, the difference between
IBS(a) and IBS(b) can be successfully captured by
the cognitive complexity. However, the symbolic
and cyclomatic complexities cannot identify the
functional differences very well.

Conclusions
This article has presented an intensive survey of
the recent advances and ground breaking studies in
Cognitive informatics, particularly its theoretical
framework, denotational mathematics, and main
application areas. CI has been described as a new
discipline that studies the natural intelligence and
internal information processing mechanisms of
the brain, as well as processes involved in perception and cognition. CI is a new frontier across
disciplines of computing, software engineering,
cognitive sciences, neuropsychology, brain sciences, and philosophy in recent years. It has
been recognized that many fundamental issues
in knowledge and software engineering are based
on the deeper understanding of the mechanisms

56

Cognitive complexity
Operational
Architectural
Cognitive
complexity
complexity
complexity
(Cop [F])
(Ca [O])
(Cc [FO])
13
5
65
34
5
170
115
7
805
163
11
1,793

of human information processing and cognitive

processes.
A coherent set of theories for CI has been
described in this article, such as the Information-Matter-Energy model, Layered Reference
Model of the Brain, the OAR model of information representation, Natural Intelligence vs.
Artificial Intelligence, Autonomic Computing
vs. imperative computing, CI laws of software,
mechanisms of human perception processes, the
cognitive processes of formal inferences, and the
formal knowledge system. Three contemporary
mathematical means have been created in CI
known as the denotational mathematics. Within
the new forms of denotational mathematical means
for CI, Concept Algebra has been designed to deal
with the new abstract mathematical structure of
concepts and their representation and manipulation in learning and knowledge engineering.
Real-Time Process Algebra has been developed
as an expressive, easy-to-comprehend, and
language-independent notation system, and a
specification and refinement method for software
system behaviors description and specification.
System Algebra has been created to the rigorous
treatment of abstract systems and their algebraic
relations and operations.
A wide range of applications of CI has been
identified in multidisciplinary and transdisciplinary areas, such as the architecture of future
generation computers, estimation the capacity of
human memory, autonomic computing, cognitive properties of information, data, knowledge,

The Theoretical Framework of Cognitive Informatics

and skills in knowledge engineering, simulation

of human cognitive behaviors using descriptive
mathematics, agent systems, CI foundations of
software engineering, deductive semantics of
software, and cognitive complexity of software
systems.

von Bertalanffy, L. (1952). Problems of life: An

evolution of modern biological and scientific
thought. London: C. A. Watts.

Acknowledgment

Wang, Y. (2002, August). On cognitive informatics

(Keynote Speech). In Proceedings of the 1st IEEE
International Conference on Cognitive Informatics (ICCI02) (pp. 34-42), Calgary, Canada. IEEE
CS Press.

The author would like to acknowledge the Natural Science and Engineering Council of Canada
(NSERC) for its support to this work. The author
would like to thank the anonymous reviewers for
their valuable comments and suggestions.

References
Bell, D. A. (1953). Information theory. London:
Pitman.
Ganter, B., & Wille, R. (1999). Formal concept
analysis (pp. 1-5). Springer.
Hoare, C. A. R. (1985). Communicating sequential
processes. Prentice Hall.
Jordan, D. W., & Smith, P. (1997). Mathematical

techniques: An introduction for the engineering,

physical, and mathematical sciences (2nd ed.).
Oxford, UK: Oxford University Press.

Klir, G. J. (1992). Facets of systems science. New

York: Plenum.
Milner, R. (1989). Communication and concurrency. Englewood Cliffs, NJ: Prentice Hall.
Quillian, M. R. (1968). Semantic memory. In M.
Minsky (Ed.), Semantic information processing.
Cambridge, MA: MIT Press.
Shannon, C. E. (1948). A mathematical theory of
communication. Bell System Technical Journal,
27, 379-423, 623-656.

von Neumann, J. (1946). The principles of largescale computing machines. Reprinted in Annals
of History of Comp., 3(3), 263-273.

Wang, Y. (2002). The real-time process algebra

(RTPA). The International Journal of Annals of
Software Engineering, 14, 235-274.
Wang, Y. (2003). Cognitive

informatics: A new
transdisciplinary research field. Brain and Mind:
A Transdisciplinary Journal of Neuroscience and

Neurophilosophy, 4(2), 115-127.

Wang, Y. (2003).
On cognitive informatics. Brain
and Mind: A Transdisciplinary Journal of Neuroscience and Neurophilosophy, 4(2), 151-167.
Wang, Y. (2003, August). Cognitive informatics
models of software agent systems and autonomic
computing (Keynote Speech). In Proceedings of
the International Conference on Agent-Based
Technologies and Systems (ATS03) (p. 25), Calgary Canada. University of Calgary Press.
Wang, Y. (2003). Using process algebra to describe
human and software system behaviors. Brain and
Mind: A

Transdisciplinary Journal of Neuroscience and Neurophilosophy

, 4(2), 199213.
Wang, Y. (2004, August). On autonomic computing and cognitive processes (Keynote Speech). In
Proceedings of the 3rd IEEE InternationalConference on Cognitive Informatics (ICCI04) (pp.
3-4), Victoria, Canada. IEEE CS Press.
Wang, Y. (2005, August). On the cognitive processes of human perceptions. In Proceedings of
the 4th IEEE International Conference on Cog-

57

The Theoretical Framework of Cognitive Informatics

nitive Informatics (ICCI05) (pp. 203-211), Irvin,

California. IEEE CS Press.
Wang, Y. (2005, May 1-4). On the mathematical laws of software. In Proceedings of the 18th
Canadian Conference on Electrical and Computer
Engineering (CCECE05) (pp. 1086-1089), Saskatoon, Saskatchewan, Canada.
Wang, Y. (2005, August). The cognitive processes
of abstraction and formal inferences. In Proceedings of the 4th IEEE International Conference
on Cognitive Informatics (ICCI05), (pp. 18-26),
Irvin, California. IEEE CS Press.
Wang, Y. (2006, May 8-10). A unified mathematical model of programs. In Proceedings of the 19th
Canadian Conference on Electrical and Computer
Engineering (CCECE06) (pp. 2346-2349), Ottawa, Ontario, Canada.
Wang, Y. (2006, July). Cognitive informatics
towards the future generation computers that
think and feel. In Proceedings of the 5th IEEE
International Conference on Cognitive Informatics (ICCI06) (pp. 3-7),Beijing, China. IEEE CS
Press.
Wang, Y. (2006, July). Cognitive informatics and
contemporary mathematics for knowledge representation and manipulation (invited plenary talk).
In Proceedings of the 1st International Conference on Rough Set and Knowledge Technology
(RSKT06) (pp. 69-78), Chongqing, China. Lecture
Notes in AI (LNAI) 4062. Springer.
Wang, Y. (2006, July). On abstract systems and
system algebra. In Proceedingsof the 5th IEEE
International Conference on Cognitive Informatics (ICCI06), (pp. 332-343), Beijing, China. IEEE
CS Press.
Wang, Y. (2006, July). On concept algebra and
knowledge representation. In Proceedings of the
5th IEEE International Conference on Cognitive
Informatics (ICCI06) (pp. 320-331), Beijing,
China. IEEE CS Press.

58

Wang, Y. (2006). On the informatics laws and

deductive semantics of software. IEEE Transactions on Systems, Man, and Cybernetics (Part
C), 36(2), 161-171.
Wang, Y. (2006). Software engineering foundations: A transdisciplinary and rigorous perspective
(CRC Book Series in Software Engineering 2).
Boca Raton, FL: CRC Press.
Wang, Y. (2006, May). The OAR model for
knowledge representation. In Proc. of the 19th
IEEE Canadian Conference on Electrical and
Computer Engineering (CCECE06) (pp. 16961699), Ottawa, Canada.
Wang, Y. (2006, July). On the Big-R notation for
describing iterative and recursive behaviors. In
Proceedings of the 5th IEEE International Conference on Cognitive Informatics (ICCI06) (pp.
132-140), Beijing, China. IEEE CS Press.
Wang, Y. (2006, July). Cognitive complexity of
software and its measurement. In Proceedings
of the 5th IEEE International Conference on
Cognitive Informatics (ICCI06) (pp. 226-235),
Beijing, China. IEEE CS Press.
Wang, Y., Dong, L., & Ruhe, G. (2004, July).
Formal description of the cognitive process of
decision making. In Proceedings of the 3rd IEEE
International Conference on Cognitive Informatics (ICCI04) (pp. 124-130), Victoria, Canada.
IEEE CS Press.
Wang, Y., & Gafurov, D. (2003, August). The
cognitive process of comprehension. In Proceedings of the 2nd IEEE International Conference
on Cognitive Informatics (ICCI03) (pp. 93-97).
London: IEEE CS Press.
Wang, Y., Johnston, R., & Smith, M. (2002).
Cognitive informatics. In Proc. of the 1st IEEE
International Conference (ICCI02), Calgary,
Alberta, Canada. IEEE CS Press.
Wang, Y., & Kinsner, W. (2006, March).
Recent
advances in cognitive informatics. IEEE Trans-

The Theoretical Framework of Cognitive Informatics

actions on Systems, Man, and Cybernetics (Part

C), 36(2), 121-123.
Wang, Y., Liu, D., & Wang, Y. (2003). Discovering
rain and Mind:
the capacity of human memory. B
A Transdisciplinary Journal of Neuroscience and
189-198.
Neurophilosophy, 4(2),
Wang, Y., & Wang, Y. (2006, March).
On cognitive
informatics models of the brain. IEEE Transactions on Systems, Man, and Cybernetics, 36(2),
203-207.
Wang, Y., Wang, Y., Patel, S., & Patel, D. (2006,
March). A layered reference model of the brain
(LRMB). IEEE Transactions on Systems, Man,
and Cybernetics (Part C), 36(2), 124-133.

This work was previously published in International Journal of Cognitive Informatics and Natural Intelligence, Vol. 1, Issue
1, edited by Y. Wang, pp. 1-27, copyright 2007 by IGI Publishing, formerly known as Idea Group Publishing (an imprint of
IGI Global).

59

60

Chapter 1.5

Human Factors in Public

Information Systems
Karl W. Sandberg
Mid Sweden University, Sweden
Pan Yan
Lule University of Technology, Sweden

Introduction
News reports do not frequently mention many
problems or accidents caused by human error.
The specialty of human factors seeks to avoid
human error by making certain that computers
and all other equipment are designed to be easy to
understand and use; costly human errors are thus
minimised. This article provides a basic overview
of the subject of human factors as it pertains to
problem and error avoidance in computerised public information systems. When computer/system
design does not adequately consider human capability, the performance of the computer/system
and the user will be below desired levels.

Background: Human Factors,

the Discipline
Human factors has helped to make information
systems less costly and easier to use. Some background on this discipline will help to show the
relationship it has with information systems.
Human factors is the science of the relationships between humans, their activities, and the
context in which those activities take place. The
specialty of human factors came into being during
WWII when it was found that the sophisticated
equipment being designed did not perform as
expected; there was a strong and immediate need
to understand why many humans were unable to
effectively use equipment and/or systems. Teams
of researchers learned that the design strategies

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Human Factors in Public Information Systems

used did not consider human performance limitationsfor example, some designs presented too
much information at the same time or in the wrong
order for humans to be able to successfully operate
controls. Or, the arrangement of controls made
them difficult to reach quickly and easily. From
this discovery came the concept that human users, their work activities, and the contexts of the
activities had to be thought of as different parts
of a whole system and that each depends upon the
other for successful operation (Bailey, 1996).
After WWII the discipline of human factors
became a specialised knowledge area as it became
apparent that the human element of any system
had to be considered if the capabilities of new
technologies were to be efficiently exploited.
The older strategy of modifying designs over a
long period of time through user experiences was
inadequate; rates of change had become so rapid
that products were obsolete before improvements
could be added. Now, the strategy often used
by successful design environments is to include
human factors in design and development. When
properly managed, products or systems that use
human factors knowledge are more efficient,
safer, and more pleasing to use because they are
designed to accommodate human performance
capabilities (Norman, 1988).
Human factors is an extremely broad technical and scientific discipline; founders of the
first national and international human factors
organisations came from such diverse fields as
engineering, design, education, computer technology, psychology, and medicine. Through its
diversity human factors is able to draw upon and
combine knowledge from any area when working with human and system performance issues.
Due to the complexity of human behaviour, human factors specialists emphasise in their work
an iterative empirical approach. First, an initial
recommendation or interface design is made and
then laboratory or field studies are conducted to
test this initial design (the prototype). Changes
are made when deficits are identified; modifica-

tions are made; and further testing is then performed. This process continues until significant
problems are no longer found. Finally, validation
is achieved through observation in the field after
system deployment.
This emphasis on empirical work tends to shape
how human factors specialists perform their roles.
Irrespective of the specific methodology chosen
for gathering data about tasks, users, and the use
of products, human factors work tends to result
in product improvements likely to be economical,
easy, and efficient to use from the beginning of use;
the cost of and need to go back and fix problems
when human factors is not used is avoided.
Human factors can also be called ergonomics.
As the term human factors is in more common usage in the computer field, it is used for
this article.

Human Factors and Computers

As with other technologies, WWII helped to
stimulate the development of computers. The first
models of computer used to manage information
were primitive in comparison to the computers
of today. They were designed by scientists and
engineers to be used by scientists and engineers.
These computer systems were difficult to operate and had to be closely watched during operation; users had to understand the technology for
successful operation and tolerate problems, as
technology-advanced computers became easier
to operate. With todays systems, the typical user
does not need to understand technology; what is
important is that a user understands how to tell a
computer exactly what information is needed.
This progress in computer technology means
that average users of computers and information
technology (IT) are no longer just scientists or
engineers; instead, they can be anyone who understands how to operate the computer information system they are using. While computer and
IT systems are much easier to use, this does not

61

Human Factors in Public Information Systems

mean that there are not challenges faced. When

working with information stored electronically or
operating something with the help of a computer,
problems can occur when a user misunderstands
information presented to them, or they fail to correctly tell the computing system what is wanted.
Literature on computers and electronic technology
describes instances where there were serious negative consequences when information presented by
a computer was misunderstood or systems were
mistakenly told to do something different than
what was wanted. The shooting down of Korean
Airlines flight 007 in 1983 demonstrates some of
the problems that can occur between the operator
and a computer. Before take-off a crew member
entered wrong data into an on-board computer. A
warning was ignored as it appeared to be caused
by another factor other than the entry of incorrect
data. Subsequently the incorrect data caused the
pilot to fly off course and into a sensitive area
where the aircraft was shot down by a missile.
Had the system been designed differently, so that
the human activity context was better considered,
then it would be likely that a tragic chain of events
would not have been started.
To prevent such events from occurring, an
emphasis on understanding how to design systems,
where it is difficult for people to make mistakes or
misunderstand, has emerged as a specialty within
the field of human factors. The concept of human-computer interaction, or HCI, as a specialty
within human factors emerged. In simple terms,
HCI is the search for ways to make computers and
complex information services accessible, usable,
and acceptable to the non-specialist user.
At first, the emphasis of HCI was on physical
aspects such as the size, location, and qualities
of computer screens and controls. How people
think when using computing and IT systems
was added as it became more and more evident
that how users think and understand is extremely
critical in computer design. The subspecialties of
cognitive engineering and usability followedthe
former concerns understanding how people think

62

when understanding systems, and the latter

concerns how well people can understand and
use systems.
Human factors and its related disciplines are
now recognised as an important part of the computing system design process. Most computer and
related system design approaches try to consider
the needs and capabilities of users. With new
products and technologies constantly entering
the market, new problems are continually encounteredthis underscores a continuing need to
address human factors when designing computers
and computing systems (Cooper, 1999).
The migration of paper-based forms to computer-based public information systems is the type
of problem now faced. Work currently in progress
in Sweden has found that the way a form looks
on paper cannot be used on a computer screen.
A computerised system had been developed that
used exact copies of paper forms on computer
screens. User error rates were extremely high, and
the computer system was more costly to operate
than its paper-based predecessor. The computer
system had been designed without involvement of
human factors specialists. Human factors specialists were asked to help correct system problems.
The initial results of the human factors assessment
reduced error rates and costs; work continues to
bring performance to a desired level. The human
factors specialists questioned and observed users.
From those, it was possible to identify simple
solutions to performance problems.

Public Information Systems,

Computers, and Human Factors
Since public and private information systems use
the same technology and function to exchange
information between users and systems, is there
any difference between the two? Should they be
treated differently? Do the two have different
needs?

Human Factors in Public Information Systems

Questions about difference and how human

factors is connected are perhaps best answered by
looking at the populations served and the reasons
for the creation of any particular system. Private
information systems generally serve segments of
entire populations (exclusive) and make little or no
attempt to accommodate those who lack needed
skills. In contrast, public systems must at least
have the goal of being usable by every member
of a population (inclusive). The exclusive nature
of private information systems means that during
development and design there is a focus on meeting the needs of a target group that is part of an
entire population. In contrast, public information
systems generally must be developed and designed
to at least have the goal of accessibility to an entire
population. A second major difference between
the two is that private information systems are
oriented towards generating at least enough revenue to meet expenses, while public information
systems are oriented towards providing service
(which may or may not generate revenue).
While these differences may be subtle in some
circumstances (e.g., a system used by a private,
non-profit organisation or a system used by a
fee-based public entity), it can be argued that, in
some form, an inclusive or exclusive orientation
is present.
How is this difference important with respect
to human factors and public information systems? It is important because public information
systems must be designed to be usable by the
publicwhich includes everyone in a population.
The goal of accessibility to an entire population
means having to design for all levels of computer
skills, education, and ability. A private information system is designed to be used by persons
who are already familiar with computers and, if
specialised knowledge/skills are needed, training
is provided.
Any entity responsible for a public information
system benefits from incorporating human factors
knowledge and skills into any information system
development or operation. The multiple levels of

needs of users make understanding this diversity

essential. Without human factors there is a much
higher risk that the best design alternatives will
not be found and selectedsuch as trying to move
from paper forms to computer-based forms without considering the differences between entering
data on paper and entering the same data onto a
computer screen. As described in greater detail
in the following section, human factors practices
make certain that people can easily understand
and use a system.

Human Factors as a Key

Element in Public System
Design
Including human factors in the earliest stages of
any design process ensures that a product is defined
in terms of user needs. Human factors input at these
stages also helps project management to redirect
resources to approaches that are most likely to have
the greatest user benefit. Human factor specialist
involvement throughout design processes results
in catching problems early on; thus helping to
reduce overall project costs. Finally, the ability to
anticipate user responses continues to improve as
successive prototypes more closely approximate
the final product and actual field conditions. During design and development processes, human
factors specialists base user interface and system
design recommendations on an understanding of
user needs, goals, characteristics, and expectations. Systems designers then work with human
factors specialists to translate these stated needs
into specific technical design requirements that
may be implemented within cost targets. As noted
earlier, the human factors specialist can work
with public information system development and
design processes to balance usability requirements
with user constraints, system requirements, and
development costs. The result can be that the final
product functions as desired and in ways that are
easily usable (Bailey, 1996).

63

Human Factors in Public Information Systems

Planning for Human Factors

during Public Information
System Development
Information systems planning is a common
problem for organisations, and planning for human factors is a particular problem within the
overall problem of design (Kuark, 1996). The
traditional approach to planning IT systems is
called the systems development life cycle (SDLC)
method (Davis & Olson, 1985). This approach can
be summarised as a linear checklist of activities
undertaken in a series of phases that begins with
planning and ends with system termination. To
put it another way, SDLC is a planning process
that separates an IT system problem into its constituent parts, treats each step separately, and then
unifies the parts into a whole at the end (Langefors, 1973). This approach has been criticised as
too concentrated on technology at the expense
of users and product performance (e.g., Noyes
et al., 1996). To keep the valuable parts of the
SDLC method there is a movement to integrate
human factors into SDLC (Zhang, Carey, Teeni,
& Tremaine, 2005). Whether integrated in with
SDLC or another methodology, HCI development
and design is the process of creating quickly all
or part of a system in the form of a simulation or
prototype (e.g., on paper or a nonworking sample
displayed on a computer screen). The advantage
of this method is that it allows user feedback early
enough to accommodate major structural changes
before extensive development investments are
committed. This use of prototyping is a team
process which helps to overcome the limitations
SDLC. The early SDLC model has been described
as too linear, lacking a group planning orientation
and not an actual reflection of real world conditions (e.g., Wanninger & Dickson, 1992). Now the
emphasis is on using the systematic features of
SDLC from a human factors perspective; technical decisions are based on an understanding of
human needs and capabilities (e.g., Zhang et al.,
2005). The following section on user-centred

64

design describes the most commonly used model

promoted by human factors professionals and
user-oriented designers.

User-Centred Design
User-centred design (UCD) refers to the design
of interaction between users and the system,
called interaction design (Preece, Rogers, &
Sharp, 2002). It models a system from a user
perspective and focuses on the usability of an
interactive software system. The core objective
is to effectively support users in executing their
tasks (Earthy, 2001).
Usability is recognised as one of the most
important quality characteristics of software
intensive systems and products. Usability gives
many benefits including increased productivity,
enhanced quality of work, improved user satisfaction, reductions in support and training costs and
improved user satisfaction (ISO13407, 1999).
The prevailing paradigm of developing usable products and systems (or UCD) is that usable products are created through processes of
user-centred design. The UCD process model is
illustrated in Figure 1. Achieving quality in use
requires this type of user-centred design process
and the use of appropriate usability evaluation
techniques. Usability is defined as a high level
quality objective: to achieve effectiveness, efficiency, and satisfaction. This requires not only
ease of use but also appropriate functionality,
reliability, computer performance, satisfaction,
comfort, and so on.

UCD and Usability

The role of UCD in a development effort is to deal
with user needs in a way that highlights potential
problems and ensures that products, systems, and
services will succeed because they are based on
appropriate UCD and usability. The applied discipline of cognitive engineering leans heavily on

Human Factors in Public Information Systems

Figure 1. UCD process model (from ISO 13407, 1999)

the knowledge and methods developed by human

experimental psychology and, to a lesser extent,
on applied physiology and related human sciences
to carry this goal out.

Usability Facilities
Usability laboratories have become an important
tool. These laboratories, together with appropriate
field-kit evaluations, enable us to meet the growing
number of requests for usability support, chiefly
user-based evaluations of development prototypes
and competitors products. Together with the
appropriate development processes, laboratory
studies help predict and deal with specific usability
problems during product development. Such a
laboratory also provides an extremely powerful
tool to identify and publicise general usability issues that affect overall product use. Rubin (1994)
describes usability testing in detail and gives
examples of typical usability laboratories.

Implementation Issues
Although improved technology will always have
a critical role in system design, it is often the case
that human factors considerations will take up the

majority of the time of IT designers and managers. Much is known about the nature of human
error, the conditions that encourage error, and
hardware/software designs that are error resistant
(Norman, 1990), and much of this is now well
recognised in IT training literature (Salvendy,
1997). While human factors is recognised and
used by many information system developers and
designers, there are those who take the perspective
that users who want access must accommodate
themselves to an information system.
The attitude of the employees response for
administering a public information system is
also critical for success. New users or those with
problems may need to ask an employee (a humanhuman interface) for help. Employees who have
a negative attitude will too often give the user a
negative attitude towards the system (Cialdini,
1993). Including employees in processes will help
to avoid the problem of employees undermining
the operation of a public information system; following are some points to use in the avoidance of
employee resistance.

Cognitive Aspects of Implementation

Overcoming employee resistance is a problem in
cognitive change. Human factors texts and refer-

65

Human Factors in Public Information Systems

ence books describe how computerisation alters

the flow and content of information, changes
job content, or affects relationships between
organisational members (e.g., Bailey, 1996; Dix,
Finlay, Abowd, & Beale, 1998; Helander, 1997).
A key finding was that the primary issue was
not computing itself, but rather perceptions of
the method of implementing it. It is precisely in
this area that human factors become important
by including assessment of organisations during
public information system changes.

Cognitive Perception
In IT projects there are four common cognitive
perceptions that must be overcome by employees:
(1) automation is mandated arbitrarily; (2) the
new computer system will be unreliable; (3) the
new system will increase rather than decrease
the work burden; and (4) they (employees) will
neither understand a system nor be able to operate it. Free and open communications throughout
an IT implementation process is an important
avenue for reducing employee resistance coming
from these issues. Negative perceptions regarding the introduction of IT may be overcome to a
significant extent through the encouragement of
employee decision making during an IT introduction process.

Attitudes
Although user attitudes tend to be against change,
once new technology is seen as bringing desired
benefits, attitudes will begin to adapt. In particular,
acceptance is much more likely when computing
is presented as complementary to human skills,
enhancing rather than replacing them (Petheram,
1989; Rosenbrock, 1981). Changing employee
attitudes is all the more important since word of
mouth is a critical aspect of the process by which
technological innovations are spread (Czepiel,
1974). Dickson, a pioneer in the study of human
relations and IT noted that, when it comes to

66

management information systems, it is essential to

involve people as a component, knowledge of their
behaviour as a part of the system is important to
the success of the entire field (Dickson, 1968).
The human factor in IT involves both teamwork and leadership. If a leader attempts to bring
about IT change in an authoritarian manner, the
result may be resistance, lack of effectiveness,
and even outright failure. On the other hand, if a
leader is able to mobilise the cognitive pressure
of professionalism in favour of IT change, then an
organisational culture of acceptance is possible
(Prasad & Prasad, 1994).

Future Trends
When considering the future of human factors
in public information systems, it is useful to
first recognise that an astonishingly rapid rate of
technological change has been and is the norm for
computing systems. Whole buildings and highly
trained teams were once needed for a single computer; in contrast, more powerful systems can now
be found on the top of the desk of the least-skilled
person in an organisation. Given this rapid and
dramatic pace of change and development, it is
challenging to predict future directions.
Despite this dramatic pace which defies the
imagination it is possible to observe that whatever technology may bring the core emphasis of
human factors will remain unchanged. For any
technological change to be effective, it must be
readily understandable and usable by an average
user. For example, while nearly instantaneous
processing and presentation of data is becoming
the norm, that data are of no value if confusing
or overwhelming.
Any future system, regardless of any technological advance it may achieve, must include in its
design consideration how humans comprehend,
manage, and process information. For example,
Miller (1956) reported that the average person
is unable to remember more than seven pieces

Human Factors in Public Information Systems

of information at one time. A computing system

that requires that a user remember eight pieces of
information to successfully complete an operation
will therefore have high rates of user error. It is
this kind of simple, yet critical, understanding of
human capability that must always be included
in design.
Thus, technologies and designs of the future
will always need to include recognition of human
limitations in any interface design. Toolkits that
contain information about human capabilities will
make it possible for researchers and designers to
consistently include consideration of the human
user in any final product.

Conclusion
New technologies are often seen as an ultimate
solution. While this is essentially correct, it is also
necessary to recognise that improved technology
does not change human capability. Computers and
computing systems, as with public information
systems, can only be as effective as users are
capable. Designs that expect users to learn to perform beyond usual capabilities ignore the reality
that humans are limited and unlike technology,
cannot be improved or upgraded. For this reason
it is essential that the concept of user-centred
design be part of any public information system
design process.
As implied, the introduction into IT system
development of user-centred design procedures
ensures that success stories associated with
usability engineering will continue. Advanced
development projects that examine psychological factors underlying consumer reactions and
expectations include user-centred evaluations of
prototypes and match human capabilities with
system designs. These are most likely to result in
public information systems that are accepted by
both the public and governmental employees.
The inclusion of human factors into public
information system development and design

processes will, over the life of a system, reduce

costs through the avoidance of the need to correct
problems caused by errors or through difficultly
in using the system. To successfully achieve this
integration strong leadership that requires human factors elements in a system is necessary.
Strategies such as specifying ISO 13407 (1999)
as the standard during development and design
are recommended as a way to be sure that human
factors will be included.

References
Bailey, R. W. (1996). Human performance
engineering (3r d ed.). Upper Saddle River, NJ:
Prentice Hall.
Cialdini, R. (1993). Influence. New York: Quill.
Cooper, A. (1999). The inmates are running the
asylum. Indianapolis, IN: SAMS.
Czepiel, J. A. (1974). Word of mouth processes in
the diffusion of a major technological innovation.
Journal of Marketing Research, II, 172-180.
Davis, G. B., & Olsen, N. (1985). Management
information systems. New York: McGraw-Hill.
Dickson, G. W. (1968). Management information decision systems. Business Horizons, II(6),
17-26.
Dix, A., Finlay, J., Abowd, G., & Beale, R. (1998).
Human-computer interaction (2n d ed.). London:
Prentice Hall.
Earthy, J. (2001). The improvement of human-centered processes facing the challenge and reaping
the benefit of ISO 13407. International Journal
of Human-Computer Studies, 55, 553-585.
Helander, M. (1997). Handbook of human-computer interaction. Amsterdam: Elsevier Science
Publishers B.V.

67

Human Factors in Public Information Systems

International Ergonomics Association. (2000).

What is ergonomics? Retrieved February 14, 2006,
from http://www.iea.cc/ergonomics/

Rosenbrock, H. H. (1981). Engineers and work

that people do. IEEE Control Systems Magazine,
1(3), 4-8.

ISO13407. (1999). Human-centred design processes for interactive systems. International

Organization for Standardization, Genev,
Switzerland

Rubin, J. (1994). Handbook of usability testing.

New York: Wiley.

Kuark, J. K. (1996). Factors in the organizational

and technical development of CIM strategies and
CIM systems. Doctoral dissertation, Eidgenoessische Technische Hochschule, Zurich.
Langefors, B. (1973). Theoretical analysis of
information systems. Lund, Sweden: Lund University.
Miller, G. A. (1956). The magical number seven,
plus or minus two: Some limits on our capacity
for processing information. The Psychological
Review, 63, 81-97.
Norman, D. A. (1990). The problem of automation: Inappropriate feedback and interaction,
not over-automation. In D. E. Broadbent, A.
Baddeley, & J. T. Reason (Eds.), Human factors
in hazardous situations. Oxford: Oxford University Press.
Noyes, J. M., Starr, A. F., & Frankish, C. R. (1996).
User involvement in the early stages of the development of an aircraft warning system. Behaviour
& Information Technology, 15(2), 67-75.
Petheram, B. (1989). An approach to integrating
technology in human service applications. Computers in Human Services, 5(1/2), 187-195.
Prasad, P., & Prasad, A. (1994). The ideology of
professionalism and work computerization: An
institutionalist study of technological change.
Human Relations, 47(12), 1433-1458.
Preece, J., Rogers, Y., & Sharp, H. (2002). Interaction design. New York: John Wiley & Sons.

68

Salvendy, G. (1997). Handbook of human factors.

New York: John Wiley and Sons.
Wanninger, L. A., Jr., & Dickson, G. W.
(1992).GRIP-group requirement identification
process: The integration of GDSS technology
into the systems development life cycle. In J. F.
Nunamaker & R. H. Sprague (Eds.), Proceedings
of the Twenty-Fifth Hawaii International Conference on System Sciences (Vol. 3, pp. 310-319). Los
Alamitos, CA: Computer Society Press.
Zhang, P., Carey J., Teeni, D., & Tremaine, M.
(2005). Integrating human-computer interaction
development into the systems development life
cycle: A methodology. Communications of AIS,
15, 512-543.

Key Terms
Cognitive Engineering: Understanding and
predicting how changes in a task environment
will influence task performance.
Human-Computer Interaction (HCI):
The study of people, computer technology, and
the ways these influence each other (Dix et al.,
1998).
Human Factors (or Ergonomics): The scientific discipline concerned with the understanding
of interactions among humans and other elements
of a system, and the profession that applies theory,
principles, data, and methods to design in order
to optimise human well-being and overall system
performance (IEA, 2005).

Human Factors in Public Information Systems

Interface: A point in any human-computer

system where information is exchanged. Interface
designs that do not take into consideration human
capabilities will likely have higher than desirable
rates of human error.
Usability: The effectiveness, efficiency, and
satisfaction with which specified users can achieve
specified goals in particular environments.
User-Centred Design: During the design of
systems or products the capabilities of end users
(humans) are considered so that the final system
or product is readily usable and designed so that
the likelihood of error is minimised.

User Interface: The way that a human and

computer system exchange information. Both
hardware and software are part of the user interface. The screen, keyboard, and mouse are the
most common hardware combination; any other
hardware design that can be used to exchange
information is also a user interface. The software
arranges the information (e.g., sequence, quantity,
manner requested, etc.). Human factors focuses
on making certain that information is presented
in a way that facilitates understanding.

This work was previously published in Encyclopedia of Digital Government, edited by A. Anttiroiko and M. Malkia, pp. 940-946,
copyright 2007 by Information Science Reference, formerly known as Idea Group Reference (an imprint of IGI Global).

69

70

Chapter 1.6

Task Analysis at the Heart of

Human-Computer Interaction
Dan Diaper
Middlesex University, UK

INTRODUCTION
The history of task analysis is nearly a century
old, with its roots in the work of Gilbreth (1911)
and Taylor (1912). Taylors scientific management
provided the theoretical basis for production-line
manufacturing. The ancient manufacturing approach using craft skill involved an individual,
or a small group, undertaking, from start to finish, many different operations so as to produce a
single or small number of manufactured objects.
Indeed, the craftsperson often made his or her
own tools with which to make end products. Of
course, with the growth of civilisation came specialisation, so that the carpenter did not fell the
trees or the potter actually dig the clay, but still
each craft involved many different operations by
each person. Scientific managements novelty was
the degree of specialisation it engendered: each
person doing the same small number of things
repeatedly.
Taylorism thus involved some large operation,
subsequently called a task, that could be broken
down into smaller operations, called subtasks.
Task analysis came into being as the method

that, according to Anderson, Carroll, Grudin,

McGrew, and Scapin (1990), refers to schemes
for hierarchical decomposition of what people
do. The definition of a task remains a classic
and under-addressed problem (Diaper, 1989b).
Tasks have been differently defined with respect
to their scope: from the very large and complex,
such as document production (Wilson, Barnard,
& MacLean, 1986), to the very small, for example,
tasks that may involve only one or two activities
which take less than a second to complete, for
example, moving a cursor (Johnson & Johnson,
1987). Rather than trying to define what is a task
by size, Diapers (1989b) alternative is borrowed
from conversation analysis (Levinson, 1983). Diaper suggests that tasks always have well-defined
starts and finishes, and clearly related activities
in between. The advantage of such a definition
is that it allows tasks to be interrupted or to be
carried out in parallel.
Task analysis was always involved with the
concept of work, and successful work is usually
defined as achieving some goal. While initially
applied to observable, physical work, as the field
of ergonomics developed from World War II, the

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Task Analysis at the Heart of Human-Computer Interaction

task concept was applied more widely to cover

all types of work that refocused attention on the
information processing aspect of tasks and the
role of the human operator as a controller, planner, diagnostician and problem solver in complex
systems (Annett & Stanton, 1998). With some
notable exceptions discussed below, tasks are still
generally defined with people as the agents that
perform work. For example, Annett and Stanton
defined task analysis as [m]ethods of collecting, classifying and interpreting data on human
performance.

BACKGROUND
Stanton (2004) suggests that [s]implistically,
most task analysis involves (1) identifying tasks,
(2) collecting task data, (3) analyzing this data so
that the tasks are understood, and then (4) producing a documented representation of the analyzed
tasks (5) suitable for some engineering purpose.
While there are many similar such simplistic
descriptions, Stantons five-item list provides an
adequate description of the stages involved in
task analysis, although the third and fourth are,
in practice, usually combined. The following
four subsections deal with them in more detail,
but with two provisos. First, one should always
start with Stantons final item of establishing the
purpose of undertaking a task analysis. Second,
an iterative approach is always desirable because
how tasks are performed is complicated.

The Purpose of a Task Analysis

Task analysis has many applications that have
nothing to do with computer systems. Even
when used in HCI (human-computer interaction),
however, task analysis can contribute to all the
stages of the software-development life cycle. In
addition, task analysis can make major contributions to other elements associated with software
development, in particular the preparation of

user-support systems such as manuals and help

systems, and for training, which was the original
application of hierarchical task analysis (HTA;
Annett & Duncan, 1967; Annett, Duncan, Stammers, & Gray, 1971). HTA was the first method
that attempted to model some of the psychology
of people performing tasks.
Although infrequently documented, identifying the purposes for using task analysis in a
software project must be the first step (Diaper,
1989a) because this will determine the task selection, the method to be used, the nature of the
outputs, and the level of analysis detail necessary.
The latter is vital because too much detailed
data that does not subsequently contribute to a
project will have been expensive to collect, and
too high a level will require further iterations
to allow more detailed analysis (Diaper, 1989b,
2004). Decomposition-orientated methods such
as HTA partially overcome the level-of-detail
problem, but at the expense of collecting more
task data during analysis. Collecting task data
is often an expensive business, and access to the
relevant people is not always easy (Coronado &
Casey, 2004; Degen & Pedell, 2004; Greenberg,
2004). Within a software-development life cycle,
Diaper (2004) has suggested that one identify all
the stages to which a task analysis will contribute
and then make selections on the basis of where
its contribution will be greatest.

Identifying Tasks
In the context of task scenarios, which Diaper
(2002a, 2002b) describes as low fidelity task
simulations, Carroll (2000) rightly points out that
there is an infinity of possible usage scenarios.
Thus, only a sample of tasks can be analysed. The
tasks chosen will depend on the task analysis
purpose. For new systems, one usually starts
with typical tasks. For existing systems and welldeveloped prototypes, one is more likely to be
concerned with complex and difficult tasks, and
important and critical ones, and, when a system

71

Task Analysis at the Heart of Human-Computer Interaction

is in use, tasks during which failures or problems

have occurred. Wong (2004) describes his critical
decision method as one way of dealing with the
latter types of tasks.
Unless there are overriding constraints within a
software project, then task analysts should expect,
and accept, the need to be iterative and repeatedly
select more tasks for analysis. Since the coverage
of all possible tasks can rarely be complete, there
is a need for a systematic task selection approach.
There are two issues of coverage: first, the range
of tasks selected, and second, the range of different ways that tasks may be carried out, both
successfully and unsuccessfully.
One criticism of task analysis is that it requires extant tasks. On the other hand, all tasks
subjected to task analysis are only simulations as,
even when observed in situ, a Hiesenberg effect
(Diaper, 1989b) can occur whereby the act of observation changes the task. Often, it is desirable
to simulate tasks so that unusual, exceptional,
and/or important task instances can be studied
and, of course, when a new system or prototype
is not available.

Collecting Task Data

There are many myths about task analysis (Diaper
et al., 2003), and one of the most persistent involves
the detailed observation of people performing
tasks. Sometimes, task-analysis data do involve
such observation, but they need not, and often it
is inappropriate even with an existing system and
experienced users.
Johnson, Diaper, and Long (1984; see also
Diaper, 1989b, 2001) claim that one of the major strengths traditionally associated with task
analysis is its capability to integrate different
data types collected using different methods.
The critical concept is that of fidelity. According to Diaper (2002a, 2002b), fidelity, a close
synonym is validity, is the degree of mapping
that exists between the real world and the world
modelled by the (task) simulation, although as

72

he says parenthetically, N.B. slightly more accurately perhaps, from a solipsistic position, it is
the mapping between one model of the assumed
real world and another.
At one end of the task-fidelity spectrum there
is careful, detailed task observation, and at the
other, when using scenarios of novel future systems, task data may exist only in task analysts
imagination. Between, there is virtually every
possible way of collecting data: by interviews,
questionnaires, classification methods such as card
sorting, ethnography, participative design, and so
forth. Cordingley (1989) provides a reasonable
summary of many such methods. The primary
constraint on such methods is one of perspective, maintaining a focus on task performance.
For example, Diaper (1990) describes the use of
task-focused interviews as an appropriate source
of data for a requirements analysis of a new generation of specialised computer systems that were
some years away from development.

Task Analysis and Task

Representation
The main representation used by virtually all task
analysis methods is the activity list, although it
goes by many other names such as a task protocol
or interaction script. An activity list is a prose
description of one or more tasks presented as a list
that usually has a single action performed by an
agent on each line. Each action on an activity-list
line may involve one or more objects, either as the
target of the action or as support for the action,
that is, as a tool. An important component of an
activity list should be the identification of triggers
(Dix, Ramduny-Ellis, & Wilkinson, 2004). While
most tasks do possess some sequences of activity
list lines in which the successful completion of
an action performed on one line triggers the next,
there are many cases when some event, either
physical or psychological, causes one of two or
more possible alternatives to occur.

Task Analysis at the Heart of Human-Computer Interaction

Diaper (2004) suggests that an activity list

is sometimes sufficient to meet a task analysis
purposes. He suggests that one of the main reasons for the plethora of task analysis methods is
the volume of data represented in the activity list
format, often tens, if not hundreds, of pages. As
Benyon and Macaulay (2002) discuss, the role of
task analysis methods applied to activity lists is
not only to reduce the sheer amount of the data,
but to allow the data to be abstracted to create a
conceptual model for designers.
The two oldest and most widely cited taskanalysis methods are HTA (Annett, 2003, 2004;
Shepherd, 2001), and goals, operators, methods,
and selection rules (GOMS; Card, Moran, &
Newell, 1983; John & Kieras, 1996; Kieras, 2004).
HTA is often misunderstood in that it produces,
by top-down decomposition, a hierarchy of goals,
and these are often confused with physical or other
cognitive activities. HTA uses rules to allow the
goal hierarchy to be traversed. Analyses such as
HTA provide a basic analysis (Kieras) that can
then be used by methods such as GOMS. While
often perceived as too complicated, it is claimed
that GOMS provides good predictive adequacy
of both task times and errors.
There are between 20 and 200 task analysis
methods depending on how one counts them.
This presents a problem as different methods have
different properties and are suitable for different
purposes. An agreed taxonomy of methods for
method selection is still unavailable. In Diaper and
Stanton (2004a), there are half a dozen different
taxonomies. Diaper (2004), rather depressingly,
suggests, in practice, people either choose a task
analysis method with which they are familiar or
they use something that looks like HTA.
Limbourg and Vanderdonkt (2004) produced a
taxonomy of nine task analysis methods, abridged
in Table 1. The methods have been reorganized
so that they increase in both complexity and
expressiveness down the table. References and

further descriptions can be found in Diaper and

Stanton (2004a).
As can be seen from Table 1, there is no accepted terminology across task analysis methods. An exception, noted by Diaper and Stanton
(2004b), is that of goals and their decomposition
and generalisation.
A number of recent attempts have been made
to classify tasks into a small number of subtasks
(Carroll, 2000; Sutcliffe, 2003; Ormerod & Shepherd, 2004). The latters subgoal template (SGT)
method, for example, classifies all information
handling tasks into just four types: act, exchange,
navigate, and monitor. Underneath this level, they
have then identified 11 task elements. The general
idea is to simplify analysis by allowing the easy
identification of subtasks, which can sometimes
be reused from previous analyses.

TASK ANALYSIS AT THE

HEART OF HUMAN-COMPUTER
INTERACTION
Diaper and Stanton (2004b) claim that [t]oday,
task analysis is a mess. Introducing Diaper
(2002c), Kilgour suggests that Diaper should
consider the rise, fall and renaissance of task
analysis. While Diaper argues that really there
has been no such fall, Kilgour is right that there
was a cultural shift within HCI in the 1990s away
from explicitly referring to task analysis. Diapers
oft-repeated argument has been that whatever it
is called, analysing tasks has remained essential
and at the heart of virtually all HCI work. Diaper
(2002a, 2002b) comments, It may well be that
Carroll is correct if he believes that many in the
software industry are disenchanted with task
analysisIt may well be that the semantic legacy
of the term task analysis is such that alternatives
are now preferable.

73

Task Analysis at the Heart of Human-Computer Interaction

Table 1. An abridged classification of some task analysis methods (based on Limbourg & Vanderdonkt,
2004)
Method
HTA
GOMS
MAD*
GTA

MUSE

TKS

CTT
Dianne+

TOOD

Origin
Cognitive
analysis
Cognitive
analysis
Psychology
Computersupported
cooperative
work
Software
engineering
& human
factors
Cognitive
analysis &
software
engineering
Software
engineering
Software
engineering
& process
control
P rocess
control

Planning
Plans

Hierarchy
Leaves

Operators

Operational
Level
Tasks

Methods &
Unit tasks
selection rules
C onstructors P re- &
postconditions
Constructors
Basic tasks

Operators

Goals &
constructors

Tasks

Actions

Plans &
constructors

Procedures

Actions

Operators

Scenarios

Basic tasks

Goals

Procedures

Input/output
transitions

Task Analysis Today

Central to Diapers current definition of task
analysis, and the primary reason why task analysis
is at the heart of virtually all HCI work, is the
concept of performance. His definition (Diaper,
2004; Diaper et al., 2003) is as follows:
Work is achieved by the work system making
changes to the application domain. The application domain is that part of the assumed real world
that is relevant to the functioning of the work
system. A work system in HCI consists of one or
more human and computer components and usually many other sorts of thing as well. Tasks are
the means by which the work system changes the
application domain. Goals are desired future states
of the application domain that the work system
should achieve by the tasks it carries out. The

74

Operationalisation

Tasks
Actions &
system
operations

Actions
Operations

Task

work systems performance is deemed satisfactory

as long as it continues to achieve its goals in the
application domain. Task analysis is the study of
how work is achieved by tasks.
Most models and representations used in software engineering and HCI are declarative; that
is, they describe things and some of the relationships between things, but not the processes that
transform things over time. For example, data-flow
diagrams are atemporal and acausal and specify
only that data may flow, but not when and under
what circumstances. In contrast, it is essential,
for all successful task-analytic approaches, that
performance is modeled because tasks are about
achieving work.
Based on Dowell and Longs (1989) and
Longs (1997) general HCI design problem,
Diapers (2004) systemic task analysis (STA) ap-

Task Analysis at the Heart of Human-Computer Interaction

proach emphasizes the performance of systems.

While STA is offered as a method, it is more of
an approach in that it deals with the basics of
undertaking the early stages of a task analysis
and then allows other analysis methods and their
representations to be generated from its activity
list output. The advantage of STA over most other
task analysis methods is that it models systems
and, particularly, the performance of the work
system. STA allows the boundary definition of
a work system to change during a task so that
different constituent subtasks involve differently
defined work systems, and these may also be differently defined for the same events, thus allowing
alternative perspectives.
The novelty of STAs view of work systems is
threefold. First, as the agent of change that performs work, a work system in HCI applications
is not usually anthropocentric, but a collection of
things, only some of them human, that operate
together to change the application domain. Second,
it is the work system that possesses goals concerning the desired changes to the application domain
rather than the goals being exclusively possessed
by people. Third, STA is not monoteleological,
insisting that work is never achieved to satisfy
a single goal, but rather it states that there are
always multiple goals that combine, trade off,
and interact in subtle, complex ways.
STAs modeling of complex work systems has
recently been supported by Hollnagel (2003b)
in cognitive task design (CTD); he claims that
cognition is not defined as a psychological process unique to humans, but as a characteristic
of systems performance, namely the ability to
maintain control. The focus of CTD is therefore
the joint cognitive system, rather than the individual user. Hollnagels formulation of CTD is
more conservative than STAs; for example, CTD
is sometimes monoteleological when he refers to
a single goal, and he restricts nonhuman goals
to a limited number of things, albeit a growing
number of technological artefacts capable of
cognitive tasks. In STA, it is not some limited

number of technological artefacts that possess

goals and other cognitive properties, but the
work system, which usually has both human and
nonhuman components.

FUTURE TRENDS
While recognising the difficulty, perhaps impossibility, of reliably predicting the future, Diaper and
Stanton (2004b) suggest that one can reasonably
predict possible futures, plural. They propose that
[f]our clusters of simulated future scenarios for
task analysis organized post hoc by whether an
agreed theory, vocabulary, etc., for task analysis
emerges and whether task analysis methods become more integrated in the future. While not
predicting which future or combination will occur,
or when, they are however confident that [p]eople
will always be interested in task analysis, for task
analysis is about the performance of work, even
though they admit that [l]ess certain is whether
it will be called task analysis in the future.
Probably because of its long history, there is
an undoubted need for the theoretical basics that
underpin the task concept and task analysis to be
revisited, as Diaper (2004) attempts to do for the
development of STA. Diaper and Stanton (2004b)
also suggest that some metamethod of task analysis needs to be developed and that more attention
needs to be placed on a wide range of types of
validation, theory, methods, and content, and also
on methods predictive capability to support design and for other engineering purposes (Annett,
2002; Stanton, 2002; Stanton & Young, 1999). At
least two other areas need to be addressed in the
future: first, how work is defined, and second, the
currently ubiquitous concept of goals.
Task analysis has always been concerned with
the achievement of work. The work concept,
however, has previously been primarily concerned
with employment of some sort. What is needed, as
Karat, Karat, and Vergo (2004) argue, is a broader
definition of work. Their proposals are consistent

75

Task Analysis at the Heart of Human-Computer Interaction

with STAs definition of work being about the

work system changing the application domain.
They persuasively argue for nonemployment application domains, for example, domestic ones.
Thus, a home entertainment system, television, or
video game, for example, could be components of
a work system, and the goals to be achieved would
be to induce pleasure, fun, or similar feelings in
their users. That such application domains are
psychological and internal to such work systems
users, rather than the more traditional changes
to things separate and external to some work
systems components, is also consistent with STAs
conceptualisation of task analysis.
Finally, Diaper and Stanton (2004b) broach,
indeed they attempt to capsize, the concept of
goals. They question whether the goals concept
is necessary, either as what causes behavior or as
an explanation for behavior, which they suggest,
based on several decades of social psychological
research, is actually usually post hoc; that is,
people explain why they have behaved in some
manner after the event with reference to one or
more goals that they erroneously claim to have
possessed prior to the behavior. Not only in all
task analysis work, but in virtually every area of
human endeavour, the concept of goals is used.
Abandoning the concept as unnecessary and
unhelpful is one that will continue to meet with
fierce resistance since it seems to be a cornerstone
of peoples understanding of their own psychology and, hence, their understanding of the world.
On the other hand, academic researchers have a
moral duty to question what may be widely held
shibboleths. Currently, goal abandonment is undoubtedly a bridge too far for nearly everyone,
which is why STA still uses the goals concept, but
greater success, if not happiness, may result in
some distant future if the concept is abandoned.
At the least, it is time to question the truth and
usefulness of the goals concept.

76

CONCLUSION
Two handbooks (although at about 700 pages
each, neither is particularly handy) on task
analysis have recently become available: Diaper
and Stanton (2004a) and Hollnagel (2003a). Both
are highly recommended and, while naturally
the author prefers the former because of his personal involvement, he also prefers the Diaper and
Stanton tome because it provides more introductory material, is better indexed and the chapters
more thoroughly cross-referenced, comes with a
CD-ROM of the entire book, and, in paperback,
is substantially cheaper than Hollnagels book.
No apology is made for citing the Diaper and
Stanton book frequently in this article, or for the
number of references below, although they are a
fraction of the vast literature explicitly about task
analysis. Moreover, as task analysis is at the heart
of virtually all HCI because it is fundamentally
about the performance of systems, then whether
called task analysis or not, nearly all the published
HCI literature is concerned in some way with the
concept of tasks and their analysis.

REFERENCES
Anderson, R., Carroll, J., Grudin, J., McGrew,
J., & Scapin, D. (1990). Task analysis: The oft
missed step in the development of computer-human interfaces. Its desirable nature, value and
role. Human-Computer Interaction: Interact90,
1051-1054.
Annett, J. (2002). A note on the validity and reliability of ergonomic methods. Theoretical Issues
in Ergonomics Science, 3, 228-232.
Annett, J. (2003). Hierarchical task analysis. In
E. Hollnagel (Ed.), Handbook of cognitive task
design (pp. 17-36). Mahwah, NJ: Lawrence Erlbaum Associates.

Task Analysis at the Heart of Human-Computer Interaction

Annett, J. (2004). Hierarchical task analysis. In

D. Diaper & N. A. Stanton (Eds.), The handbook
of task analysis for human-computer interaction
(pp. 67-82). Mahwah, NJ: Lawrence Erlbaum
Associates.

Diaper, D. (1989a). Task analysis for knowledge

descriptions (TAKD): The method and an example. In D. Diaper (Ed.), Task analysis for human-computer interaction (pp. 108-159). West
Sussex, UK: Ellis Horwood.

Annett, J., & Duncan, K. D. (1967). Task analysis

and training design. Occupational Psychology,
41, 211-221.

Diaper, D. (1989b). Task observation for humancomputer interaction. In D. Diaper (Ed.), Task
analysis for human-computer interaction (pp.
210-237). West Sussex, UK: Ellis Horwood.

Annett, J., Duncan, K. D., Stammers, R. B., &

Gray, M. J. (1971). Task analysis (Training Information Paper No. 6). London: HMSO.
Annett, J., & Stanton, N. A. (1998). Special issue:
Task analysis [Editorial]. Ergonomics, 41(11),
1529-1536.
Benyon, D., & Macaulay, C. (2002). Scenarios
and the HCI-SE design problem. Interacting with
Computers, 14(4), 397-405.
Card, S., Moran, T., & Newell, A. (1983). The
psychology of human-computer interaction.
Mahwah, NJ: Lawrence Erlbaum Associates.

Diaper, D. (1990). Analysing focused interview

data with task analysis for knowledge descriptions (TAKD). Human-Computer Interaction:
Interact90, 277-282.
Diaper, D. (2001). Task analysis for knowledge
descriptions (TAKD): A requiem for a method.
Behaviour and Information Technology, 20(3),
199-212.
Diaper, D. (2002a). Scenarios and task analysis.
Interacting with Computers, 14(4), 379-395.
Diaper, D. (2002b). Task scenarios and thought.
Interacting with Computers, 14(5), 629-638.

Carroll, J. M. (2000). Making use: Scenariobased design for human-computer interactions.

Cambridge: MA: MIT Press.

Diaper, D. (2002c). Waves of task analysis. Interfaces, 50, 8-10.

Cordingley, E. (1989). Knowledge elicitation

techniques for knowledge-based systems. In D.
Diaper (Ed.), Knowledge elicitation: Principles,
techniques and applications (pp. 87-176). West
Sussex, UK: Ellis Horwood.

Diaper, D. (2004). Understanding task analysis for

human-computer interaction. In D. Diaper & N. A.
Stanton (Eds.), The handbook of task analysis for
human-computer interaction (pp. 5-47). Mahwah,
NJ: Lawrence Erlbaum Associates.

Coronado, J., & Casey, B. (2004). A multicultural approach to task analysis: Capturing user
requirements for a global software application. In
D. Diaper & N. A. Stanton (Eds.), The handbook
of task analysis for human-computer interaction
(pp. 179-192). Mahwah, NJ: Lawrence Erlbaum
Associates.

Diaper, D., May, J., Cockton, G., Dray, S., Benyon,

D., Bevan, N., et al. (2003). Exposing, exploring,
exploding task analysis myths. HCI2003 Proceedings (Vol. 2, pp. 225-226).

Degen, H., & Pedell, S. (2004). The JIET design

for e-business applications. In D. Diaper & N.
A. Stanton (Eds.), The handbook of task analysis
for human-computer interaction (pp. 193-220).
Mahwah, NJ: Lawrence Erlbaum Associates.

Diaper, D., & Stanton, N. A. (2004a). The

handbook of task analysis for human-computer
interaction. Mahwah, NJ: Lawrence Erlbaum
Associates.
Diaper, D., & Stanton, N. A. (2004b). Wishing
on a star: The future of task analysis for humancomputer interaction. In D. Diaper & N. A. Stanton

77

Task Analysis at the Heart of Human-Computer Interaction

(Eds.), The handbook of task analysis for humancomputer interaction (pp. 603-619). Mahwah, NJ:
Lawrence Erlbaum Associates.
Dix, A., Ramduny-Ellis, D., & Wilkinson, J.
(2004). Trigger analysis: Understanding broken
tasks. In D. Diaper & N. A. Stanton (Eds.), The
handbook of task analysis for human-computer
interaction (pp. 381-400). Mahwah, NJ: Lawrence
Erlbaum Associates.
Dowell, J., & Long, J. (1989). Towards a conception
for an engineering discipline of human factors.
Ergonomics, 32(11), 1513-1535.
Gilbreth, F. B. (1911). Motion study. Princeton,
NJ: Van Nostrand.
Greenberg, S. (2004). Working through task-centred system design. In D. Diaper & N. A. Stanton
(Eds.), The handbook of task analysis for humancomputer interaction (pp. 49-68). Mahwah, NJ:
Lawrence Erlbaum Associates.
Hollnagel, E. (2003a). Handbook of cognitive
task design. Mahwah, NJ: Lawrence Erlbaum
Associates.
Hollnagel, E. (2003b). Prolegomenon to cognitive
task design. In E. Hollnagel (Ed.), Handbook of
cognitive task design (pp. 3-15). Mahwah, NJ:
Lawrence Erlbaum Associates.
John, B. E., & Kieras, D. E. (1996). Using GOMS
for user interface design and evaluation: Which
technique? ACM Transactions on Computer-Human Interaction, 3, 320-351.
Johnson, H., & Johnson, P. (1987). The development of task analysis as a design tool: A method
for carrying out task analysis (ICL Report).
Unpublished manuscript.
Johnson, P., Diaper, D., & Long, J. (1984). Tasks,
skills and knowledge: Task analysis for knowledge based descriptions. Interact84: First IFIP
Conference on Human-Computer Interaction,
1, 23-27.

78

Karat, J., Karat, C.-M., & Vergo, J. (2004). Experiences people value: The new frontier for task
analysis. In D. Diaper & N. A. Stanton (Eds.), The
handbook of task analysis for human-computer
interaction (pp. 585-602). Mahwah, NJ: Lawrence
Erlbaum Associates.
Kieras, D. (2004). GOMS models for task analysis.
In D. Diaper & N. A. Stanton (Eds.), The handbook
of task analysis for human-computer interaction
(pp. 83-116). Lawrence Erlbaum Associates.
Levinson, S. C. (1983). Pragmatics. Cambridge,
MA: Cambridge University Press.
Limbourg, Q., & Vanderdonkt, J. (2004). Comparing task models for user interface design. In
D. Diaper & N. A. Stanton (Eds.), The handbook
of task analysis for human-computer interaction
(pp. 135-154). Mahwah, NJ: Lawrence Erlbaum
Associates.
Long, J. (1997). Research and the design of human-computer interactions or What happened
to validation? In H. Thimbleby, B. OConaill, &
P. Thomas (Eds.), People and computers XII (pp.
223-243). New York: Springer.
Ormerod, T. C., & Shepherd, A. (2004). Using task
analysis for information requirements specification: The sub-goal template (SGT) method. In D.
Diaper & N. A. Stanton (Eds.), The handbook of
task analysis for human-computer interaction (pp.
347-366). Lawrence Erlbaum Associates.
Shepherd, A. (2001). Hierarchical task analysis.
London: Taylor and Francis.
Stanton, N. A. (2002). Developing and validating theory in ergonomics. Theoretical Issues in
Ergonomics Science, 3, 111-114.
Stanton, N. A. (2004). The psychology of task
analysis today. In D. Diaper & N. A. Stanton
(Eds.), The handbook of task analysis for humancomputer interaction (pp. 569-584). Mahwah, NJ:
Lawrence Erlbaum Associates.

Task Analysis at the Heart of Human-Computer Interaction

Stanton, N. A., & Young, M. S. (1999). What price

ergonomics? Nature, 399, 197-198.
Sutcliffe, A. (2003). Symbiosis and synergy? Scenarios, task analysis and reuse of HCI knowledge.
Interacting with Computers, 15(2), 245-264.
Taylor, F. W. (1912). Principles of scientific management. New York: Harper and Row.
Wilson, M., Barnard, P., & MacLean, A. (1986).
Task analysis in human-computer interaction
(Rep. No. HF122). IBM Hursely Human Factors.
Wong, B .L. W. (2004). Critical decision method
data analysis. In D. Diaper & N. A. Stanton (Eds.),
The handbook of task analysis for human-computer interaction (pp. 569-584). Mahwah, NJ:
Lawrence Erlbaum Associates.

KEY TERMS

Application Domain: That part of the assumed real world that is changed by a work system
to achieve the work systems goals.
Goal: A specification of the desired changes
a work system attempts to achieve in an application domain.
Performance: The quality, with respect to
both errors and time, of work.
Subtask: A discrete part of a task.
Task: The mechanism by which an application
domain is changed by a work system to achieve
the work systems goals.
Work: The change to an application domain
by a work system to achieve the work systems
goals.
Work System: That part of the assumed real
world that attempts to change an application domain to achieve the work systems goals.

Activity List: A prose description of a task or

subtask divided into lines to represent separate
task behaviors and that usually has only one main
agent and one action per line.

This work was previously published in Encyclopedia of Human Computer Interaction, edited by C. Ghaoui, pp. 579-587, copyright 2006 by Information Science Reference, formerly known as Idea Group Reference (an imprint of IGI Global).

79

80

Chapter 1.7

Evaluating Mobile
Human-Computer Interaction
Chris Baber
The University of Birmingham, UK

Abstract
In this chapter the evaluation of human computer
interaction (HCI) with mobile technologies is
considered. The ISO 9241 notion of context of
use helps to define evaluation in terms of the
fitness-for-purpose of a given device to perform
given tasks by given users in given environments.
It is suggested that conventional notions of usability can be useful for considering some aspects
of the design of displays and interaction devices,
but that additional approaches are needed to fully
understand the use of mobile technologies. These
additional approaches involve dual-task studies in
which the device is used whilst performing some
other activity, and subjective evaluation on the
impact of the technology on the person.

INTRODUCTION
This chapter assumes that usability is not a feature of a product, that is, it does not make sense
to call a product itself usable. Rather, usability

is the consequence of a given user employing a

given product to perform a given activity in a given
environment. Holcomb and Tharp (1991) proposed
a model of interface usability, which is illustrated
by Table 1. The definitions presented in Table 1
arose from consideration of the user interface of
desk-based computers. However, it ought to be
apparent that the majority of the components are
defined in terms of an individuals perceptions of
features of the user interface.
The International Standards Organization has
a number of standards relevant to human-computer
interaction (Bevan, 2001). Current standards for
mobile devices tend to focus on product attributes,
for example, ISO 18021: Information Technology
User Interface for Mobiles (2001) provides
interface specifications for Personal Digital Assistants. Other Standards have recognized the
multifaceted nature of usability and have sought
to encourage an approach that is similar to Quality
Assessment (Earthey et al., 2001). Demonstrating
compliance with the standards requires analysts
to document their evaluation, demonstrating
how it meets the objectives of the standard. The

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Evaluating Mobile Human-Computer Interaction

Table 1. Holcomb and Tharps (1991) model of interface usability

Component

Term

Functional

Able to accomplish tasks for which software is intended

Perform tasks reliably and without errors

Consistent

Consistent key definitions

Show similar information at same place on screens
Uniform command syntax

Natural and Intuitive

Learnable through natural conceptual model

Familiar terms and natural language

Minimal memorization

Provide status information

Dont require information entered once to be re-entered
Provide lists of choices and allow picking from the lists
Provide default values for input fields

Feedback

Prompt before destructive operations like DELETE

Show icons and other visual indicators
Immediate problem and error notification
Messages that provide specific instructions for action

User help

Online help system available

Informative, written documentation

User control

Ability to undo results of prior commands

Ability to re-order or cancel tasks
Allow operating system actions to be performed within the interface

definition of usability offered by the International

Standards Organization, that is, in ISO9241, part
11, is, the extent to which a product can be
used by specified users to achieve specified goals
with effectiveness, efficiency and satisfaction in
a specified context of use. (ISO9241-11, 1998).
The implications are that, first, usability is the
consequence of a given user employing a given
product to perform a given activity in a given
environment (as stated) and, second, that it is
possible to measure aspects of this relationship
in terms of effectiveness, efficiency, and user
satisfaction. It is important to note that these
three aspects are inter-connected and that any
evaluation activity ought to try to measure some
aspect of each (Frkjr et al., 2000).

Defining Evaluation Targets

If one is able to speak of measures, then it makes
sense to be able to determine some criteria that
indicate good or poor performance on these

measures. Good et al. (1986) proposed that it is

important to define both evaluation targets and
metrics that relate to these targets. For example,
in a study of conferencing systems, Whiteside
et al. (1988) identified 10 attributes that they felt
reflected the use of the conferencing system, for
example, ranging from a fear of feeling foolish to
a number of errors made during task performance.
For each attribute, Whiteside et al. (1988) defined
a method for collecting data about that attribute,
for example, questionnaires, observation, and so
forth, and then set performance limits relating
to best, worst, and planned levels. A study of
a wearable computer for paramedics (Baber et
al., 1999) used this concept to produce Table 2.
In Table 2, three measures of performance were
undertaken, that is, predictive modeling (using
critical path analysis), user trials, and performance
improvement arising from practice. In addition,
three subjective evaluation methods were used.
Table 2 shows how the system met (or exceeded)
some of the target criteria but fell below the target

81

Evaluating Mobile Human-Computer Interaction

Table 2. Defining evaluation targets

Factors
Performance
Task
Practice
Subjective evaluation

Method

Metrics

Target

Best

Current

CPA

Time

-15%

+5%

-2%

User trials
1st vs. 3rd trial

Time
% change

-15%
1st > 3rd

0
3rd > 1st

+5%
0

-10%
3rd>1st

SUS1
SUMI2
Heuristics

Scale: 0-100
Scale: 0-100
Scale: 0-10

50
50
<6

60
60
6

70
70
>6

65
60
8

for time (although it is not within the worst case

range). One benefit of such a technique is to allow
the design team to decide whether there is a need
for more effort to refine the device, or whether,
having met (some or all of) the requirements, the
design process can be closed.
The ISO9241 notion of usability requires the
concept of evaluation targets, for example, one
could begin with a target of 66% of the specified
users would be able to use the 10 main functions
of product X after a 30 minute introduction.
Once this target has been met, the design team
might want to increase one of the variables, for
example, 85% of the specified users, or 20 main
functions, or 15 minute introduction, or might
want to sign-off that target.

Why Conduct Evaluation?

The concept of usability that is used in this chapter
(and in ISO9241) implies that changing any one of
the variables {user, activity, device, environment}
can have an impact on usability. This implication
points to the well-known assertion that an activity
that a designer of the product might find easy to
perform could prove problematic for a user who
has had little or no previous experience of the
product. It also points to potential issues relating
to the usability of mobile technology, particularly
through consideration of the environment. If
we think about sending a text-message from a
handheld device, such a mobile telephone or a
BlackberryTM, the activities involved could be

82

Worst

somewhat different while sitting on a train versus walking down a busy street. This change in
environmental setting will have a marked effect
on usability of the device. This does not necessarily result from the design of the device itself
but rather from the interactions between design,
use, and environment. As Johnson (1998) pointed
out, HCI methods, models and techniques will
need to be reconsidered if they are to address the
concerns of interaction on the move. (Johnson,
1998). The question for this chapter, therefore,
is how best to address the relationship between
user, activity, product, and environment in order
to evaluate the usability of mobile technology.
Related to this question is how evaluation might
capture and measure this relationship, and then
what can designers do to improve usability. This
latter point is particularly problematic if one
assumes that design is about creating a product
rather than about creating an interaction.
Before considering these questions, it is worth
rehearsing why one might wish to conduct evaluation. Baber (2005) notes that the primary reason
for conducting evaluation, in HCI, is to influence
design (ideally, to improve the product). This implies that evaluation ought never to be a one-off
activity to be conducted at the end of the design
lifecycle in order to allow a design to be signed-off
(Gould & Lewis, 1985; Johnson, 1992). Rather, it
means the following:
1.

Evaluation is a recursive activity that cuts

across the entire design lifecycle, for exam-

Evaluating Mobile Human-Computer Interaction

2.

3.

4.

ple, software engineers will run versions of

the code to debug and check; product designers will continually critique and refine their
concepts. What is not always apparent is the
manner in which these processes could (or
indeed ought) be made formal and to result
in something that can be communicated to
other members of the design team.
Evaluation should be incorporated into as
many stages of design as possiblethis
points to (i) but also raises that questions of
recording and communicating the results of
evaluation in a many that can be beneficial
to the design process.
Evaluation should be designed to maximize
the impact of the evaluation of the design
stage in which it is usedthe suggestion
is that, rather engaging in evaluation as a
mandated exercise to allow sign-off between
stages, it ought to be an activity that positively advances the design process.
Evaluation should guide and inform design
activitythe results of any evaluation
should be reported in a manner that can lead
to change in the design and can be reported in
a manner that is transparent and reliable.

A final point to note is that evaluation is a process of comparing the product against something
else, for example, other products, design targets,
requirements, standards. Thus, evaluation requires a referent model (Baber, 2005). It is nave
to believe that one can evaluate something in
a vacuum, that is, to think that one can take a
single product and evaluate it only in terms of
itself. In many ways this is akin the concept of
a control condition in experimental design; one
might be able to measure performance, but without
knowing what would constitute a baseline for the
measure, it is not possible to determine whether
it is good or bad.

Defining Referent Models

While it might be fairly clear as to why comparison
requires a referent model, there is a problem for
novel technologies. After all, the point of these
technologies is to move beyond the conventional
desk-bound personal computers and this will ultimately create new forms of interaction. However,
the move to different technologies makes it hard
to establish a sensible basis for evaluation. What
is the referent model for mobile HCI?
A common form of mobile technology is the
digital tour-guide that, knowing where the user
is (using Global Positioning Satellite (GPS) to
determine location) and what the user is doing,
can provide up-to-the-minute information to help
the user. There are few, if any, products that are
like these concepts, so what constitutes a referent?
At one level, this is simply because future HCI is
attempting to develop approaches to interaction
with technology for which there are no existing
models. The answer to this question, the author
suggests, comes from the assertion at the start of
this chapter: usability is not a characteristic of the
product, but the result of the interactions between
user, product, activity, and environment. If we assume that tourists have a variety of strategies and
artifacts that they currently use to find out where
they are or to find out interesting information
about a particular location, for example, maps,
books, leaflets, other people. One could ground
an initial evaluation of using the product to perform a given set of activities in comparison with
existing practices. Conducting evaluation against
other products in terms of a set of activities offers
the analyst the following benefits:
1.

The evaluation will cover a range of functions on the products. It is important to

ensure that the comparison provides a fair
and accurate view of the product. After all,
it is not really the point of evaluation to just
demonstrate the product X is better than
product Ypartly because there are bound

83

Evaluating Mobile Human-Computer Interaction

2.

3.

to be occasions when products X and Y are

similar, or where product Y is better than
product X, and partly because simply knowing that X > Y tells us very little about how
to improve X (or Y) or why X is superior.
The focus of the evaluation is less on product functioning than on user activity. This
might appear, at first glance, to be tautologicalsurely product evaluation is about
evaluating the product? This is, of course,
true in a technical sense. However, HCI is
about human-computer interaction, and
the defining feature of this relationship is
the interaction (rather than either human or
computer). If one is concerned with technical
evaluation then, perhaps some of the features
to be included in a comparison table (like the
one shown in Table 2) would be some of the
technical features, for example, processor
speed, RAM, memory, and so forth.
As the evaluation is concerned with user
activity (as opposed to product functioning),
the type of metrics that could be applied may
well change. When comparing user activity on two or more products, it is important
to decide what information is really being
sought. Do we want to know only that X>Y?
Or do we want to know that using product
X or Y have differing effects on user activity?

In the field of mobile and wearable computers,

much of the evaluation research has focused on
comparing performance on a wearable computer
with performance using other media. Thus, studies might compare performance using a wearable
computer, say to perform a task that involves following instructions, and find that sometimes performance is superior in the paper condition (Siegel
& Bauer, 1997; Baber et al., 1999) and sometimes
it is superior in the wearable computer condition
(Bass et al., 1995, 1997; Baber et al., 1998). This
highlights the potential problem of comparing

84

disparate technologies in an evaluation; it is not

clear that any differences in performance are due
to the experiment favoring one technology over
another or whether there are other factors at play
here. For example, a common observation is that
people using the wearable computer tend to follow
the instructions laid out on the display, whereas
people using paper tend to adopt a more flexible
approach (Siegel & Bauer, 1997; Baber et al., 1999).
The notion that technology influences the ways
in which people work is often taken as commonsense by Human Factors engineers. However, the
question of how and why such changes arise ought
to have a far deeper impact on evaluation than is
currently the case. As mentioned earlier, one way
to deal with this problem is to focus on activities
that people are performing using a variety of
products. However, this will only cope with part
of the problem. For instance, the electronic tourguide given could be evaluated in comparison
with other ways of performing activities, but this
does not tell us whether any differences between
the electronic tour-guide and the other products
are due to the concept or to the realization of the
concept or to changes in the activity arising from
the use of the device. In other words, if we find
that the electronic tour-guide performs less well
than speaking to someone, is this because the
tour-guide lacks information, or because it lacks
clear presentation of information, or because it
lacks speedy access to the information, or because
it lacks flexibility of response, or because of some
other reason (the evaluation could point to all of
these, not to specific reasons).
At one level, the evaluation of mobile HCI
calls for the application of current evaluation techniques. However, there are other aspects of future
HCI that call for rethinking of evaluation. In other
words, it might not be entirely appropriate to take
methods that have proven useful for evaluating
desktop HCI and apply these to future HCI. As
Wilson and Nicholls (2002) point out in discussing
the evaluation of virtual environments:

Evaluating Mobile Human-Computer Interaction

There are only a limited number of ways in which

we can assess peoples performance...We can
measure the outcome of what they have done, we
can observe them doing it, we can measure the
effects on them of doing it or we can ask them
about either the behavior or its consequences.
(Wilson & Nicholls, 2002)
The underlying assumption here is that human
behavior is measurable in a finite number of ways.
Combining this assertion with the need to study
the relationship between user, activity, device,
and environment, it becomes apparent that evaluation of user interaction with mobile activity can
be reduced to a small number of requirements.
Furthermore, the ISO 9241 notions of efficiency,
effectiveness, and satisfaction point to the approaches outlined by Wilson and Nicholls (2002).
For example, efficiency could be considered in
terms of the amount of resource expended in
order to achieve a goal (perhaps in terms of time
to complete a task), and effectiveness could relate
to the quality of this performance (perhaps in
terms of the amount of activity completed or the
quality of the outcome), and satisfaction would
relate to the user perception of the activity (perhaps in terms of a judgment relating to their own
performance or effort, perhaps relating to some
aspect of using the device). What is required is
not so much a battery of new measures, so much
as an adaptation of existing approaches that pay
particular attention to the relatively novel aspects
of the environment and activity that pertain to
mobile devices.

MAKING SENSE OF HUMAN

ACTIVITY WITH MOBILE
TECHNOLOGY
The argument so far is that what needs to be evaluated is not simply the product, but the interaction
between user, activity, device, and environment.
This raises the question of what can be defined as

appropriate forms of activity. The first issue for

mobile technology is the assumption that it is to
be used on the move, which raises two possibilities: (1) on the move means physically moving,
for example, walking, driving a car, traveling
as a passenger; (2) on the move means being
in different places away from normal office
environments. One problem relating to both of
these possibilities is the difficulty of collecting
data in the fieldthere are problems arising from
recording the data, managing the collection of
data, and controlling experimental conditions
that are far from trivial. However, if evaluation
studies involve managing the interactions between
user, activity, device, and environment, then it
might not be possible to concentrate efforts on
specific aspects of the interactions, for example,
comparing the use of the device under different
mobility conditions.

Interacting with Mobile Technology

while Walking
Consideration of interaction while moving immediately suggests that asking people to evaluate a
product whilst sitting down in a laboratory might
lead to different results than when using the product on the move. This is just what Kjeldskov and
Stage (2004) demonstrated. Indeed, they found
that having participants report usability problems
while sitting down in the laboratory led to more
usability problems being reported than when the
participants performed the evaluation while walking. They suggested that this result might have
arisen from different demands on attentionin
the seated condition there was little distraction
from the product and so participants were able
to devote most of their attention to it, but in the
walking conditions, attention needed to be divided between the device and the task of walking.
This effect can be compounded by variation in
other contextual factors, such as lighting levels
and complexity of the path that one is following
(Barnard et al., 2007).

85

Evaluating Mobile Human-Computer Interaction

It has been demonstrated that walking can impact

cognitive tasks (Ebersbach et al., 1995), and so the
use of a mobile device could be thought of in terms
of a dual-task. A common methodological approach
in Ergonomics/Human Factors involves asking
participants to perform one task while attending
to another, for example, tracking a line on a screen
while also performing mental arithmetic. There are
several reasons why this approach is useful, both in
terms of developing theory of human performance
and in terms of considering how combinations of
tasks can be modified. In broad terms, the assumption is that the human ability to process information
from several sources can be compromised under
conditions of increasing complexity. Complexity
might arise from the difficulty of one or both of the
tasks, from the quality of the signals being attended
to, from the amount of interference between the two
tasks, and so forth. By measuring performance on
the tasks under different levels of complexity, it is
possible to judge the persons ability to perform and
the amount of interference that could occur.
Taking the dual-task paradigm as a starting
point, one can consider many forms of mobile
technology to be used not only in different places
but also while the person is physically moving, for
example, walking down a busy street or following
a predefined route or walking on a treadmill. Thus,
one approach to studying mobile technology from
a dual-task perspective would involve measuring
some aspect of walking and some aspect of using the
technology. Barnard et al. (2005) compared reading
tasks on a Personal Digital Assistant (PDA) while
walking on a treadmill and walking along a defined
path. They found a reduction in walking speed (by
around 33%) compared to walking without performing the tasks on the device. This indicates that using
the device leads to measurable changes in walking
activity. They found no difference in comprehension
between conditions (although it is often difficult
to find measurable differences in comprehension
in experiments that involve reading from screens,
see Dillon, 1992), but they did find that word search
took significantly longer when walking along the

86

path than on the treadmill. This result suggests

that participants walking the path had more need
to divide their attention between the world and the
device, and indeed, path following correlated with
the use of scroll bars on the device, suggesting that
more attention on the world led to more need to scroll
the text to find ones place while reading. What is
interesting about this particular study is that it reports
objective results on both primary (using the device)
and secondary (walking under different conditions)
tasks, and shows some interactions between the
two. This shows that modifying the environment
has a bearing on activity which, in turn, affects user
performance (even with the same device requiring
the same activity).

Using Mobile Technology while

On-the-Move
Prototypical mobile technologies often address
scenarios related to tourists because this emphasizes the need to move around an unfamiliar
environment and the desire for information relating to the world around us, for example, routes to
places, interesting information about landmarks,
advice on traveling, and so forth. Considering the
scenario from the perspective of usability, evaluation could allow the design team to agree on
benchmark levels of performance using existing
practices, and to then consider what benefits might
accrue from modifying those practices through
the introduction of technology.
While the activity of walking can interact with
the use of the mobile device, there are other aspects
of use on-the-move that can also play a role. Duh
et al. (2006) compared the evaluation of a mobile
telephone, used to perform a set of activities, in the
laboratory and on a Mass Rapid Transit (MRT) train
in Singapore. The study showed that participants
encountered significantly more problems in the
train condition than in the laboratory. The authors
relate the problem to five primary areas: ambient
noise levels, movement of the train, issues relating
to privacy, increase in effort needed to perform

Evaluating Mobile Human-Computer Interaction

the activity, additional stress, and nervousness. Of

these factors, the main ones relate to aspects of the
environment, viz. noise and movement of the train,
and these, in turn, have a bearing of the ability of
participants to complete the activity. In addition to
the affect of movement on the performance of the
users, the impact on the performance of the technology is equally important, for example, what happens
when wireless networks do not cover the whole of
the area and the user encounters shadows, or what
happens when positioning systems have drift or
inaccuracies. One approach might be to attempt to
guarantee optimal delivery of service at all times by
modifying the infrastructure rather than the device,
for example, with boosters located in the environment. Another approach would be to provide ways
of informing the user about accuracy of the data on
which the system is working (Bell et al., 2006).
Devices can also be used while driving automobiles and there is evidence that interference between
the activity of driving and the activity of using the
device are more than simply physical (Boase et al.,
1988; Brookhuis et al., 1991; Svenson & Patten,
2005; Wikman et al., 1998). This means that using a hands-free kit will not eliminate all forms
of interference. For example, Nunes and Recarte
(2002) show that the more cognitively demanding
a telephone conversation, the greater the reduction
in the users ability to attend to the environment
while driving. This research further highlights
the problem of isolating the usability of the device
itself from the interactions between user, activity,
and environment.

SUBJECTIVE EVALUATION OF
TECHNOLOGY
[U]ltimately it is the users of a software system
[or any product] who decide how easy its user
interface is to manipulate (Holcomb & Tharp,
1991). Thus, one might feel that asking people
about the product would be the obvious and most
useful approach to take. However, there are several

problems with this approach, for example, people

might not always be able to articulate how they
feel about the product (so the reports might be
incomplete or inconsistent), people might use a
variety of previous experiences as their referent
models (so it might be difficult to generalize results
across respondents), people might not be able to
respond critically to the product (so there might be
a halo-effect with the participant responding to
the novelty of the device rather than considering
issues of usability). For these and other reasons,
it is common practice to provide some structure
to subjective evaluation, usually through some
form of procedure or checklist. Furthermore, it
would be suggested that subjective evaluation
should be used as a secondary measure as far as
practicable, with the primary focus on data collected from user trials.

Subjective Response to the Device

Participants could be asked to walk-through the
performance of a given activity using the device, by
explaining what they are doing and why. Monk et al.
(1986) presented a detailed set of guidelines on how
to use walk-through approaches to the evaluation
in their Cooperative Evaluation method. The main
aim of the approach is to capture problems that users
experience when using a product.
In terms of checklists, a great deal of research
effort from the late 1980s to the mid 1990s led to
the development of a number of usability surveys.
Some, like CUSI-Computer User Satisfaction Inventory (Kirakowski & Corbett, 1988) and QUISQuestionnaire for User Interface Satisfaction (Chin
et al., 1988), are designed to capture user response
to an interface, particularly in terms of affective
components (such as satisfaction). Others, like the
checklist of Ravden and Johnson (1989) or SUS
(Brooke, 1996), have been designed to cover both
aspects of the interface and characteristics of usability. While these surveys are based on sound
HCI principles, interpretation is left to the analyst
who could lead to potential bias or misinterpreta-

87

Evaluating Mobile Human-Computer Interaction

tion. The SUMI checklist (Kirakowski, 1996) was

developed using a rigorous approach to defining
appropriate components of usability and presents
results in terms of a comparison with a database of
previous evaluations.

such as frustration or perceived level of performance.

The NASA-TLX captures these responses and can
be used to compare perceptions of users with combinations of different devices or activities.

Subjective Responses to Using the

Device To Perform an Activity

DESIGNING AN EVALUATION
PROTOCOL

In addition to eliciting opinions from users regarding the device, researchers are also keen to obtain
reactions of some of the consequences of using
the device. By way of analogy, if we consider the
virtual reality research community, we can see
efforts to elicit reaction to either the physical effects of using virtual reality, for example, Cobb
et al.s (1999) Virtual Reality Induced Symptoms
and Effects (VRISE) or the measurement of
presence(Slater et al., 1994; Witmer & Singer,
1998). In the domain of wearable computers, physical effects have been evaluated using self-report
on a comfort rating scale (Knight et al., 2002).
In terms of performing an activity, researchers
often make use of the NASA-TLX (Hart & Staveland, 1988) which measure subjective response to
workload. The basic notion is that activities make
different demands on people in terms of time pressure
or mental effort, and can lead to different responses

Throughout this chapter, emphasis has been placed

on the notion of context of use and the concept
of usability defined by ISO 9241, pt. 11. The relationship between these concepts is illustrated by
Figure 1. In order to evaluate any item of technology, one needs to plan an appropriate campaign
of evaluationthis means consideration of the
evaluation from the start of the design process and
performance of evaluation as often as practicable
during the course of development. Assume that,
whatever design process is being followed, there
will be four primary phases: initial concept development, prototyping, specification and build.
At each phase, the form and type of evaluation
will change (depending on access to functionality on the product as much as anything else), but
the basic considerations remain constant, that is,
adequately defining context of use and applying
appropriate usability metrics.

Figure 1. ISO9241 usability evaluation process

Context of Use

Activity

Goal

User
Tasks
Usability Metrics

Environment

Efficiency

Other products

Product

88

Effectiveness

Outcome

Satisfaction

Evaluating Mobile Human-Computer Interaction

Before elaborating on Figure 1 as a process, it

is worth re-emphasizing the point made earlier that
usability evaluation always involves comparison
with the product being considered against some
referent model. The referent model could be other
products, but is equally likely to be a set of design
targets (see Table 2). In terms of comparison, a set
of usability metrics can be applied. The efficiency
metric relates to the manner in which resources
are applied during the activity in order to achieve
the outcome; the effectiveness metric relates to
the completion of the outcome; the satisfaction
metric relates to the users response to performing
the activity. Needless to say, all metrics apply to a
given user performing a given activity in order to
achieve a given goal in a given context of use with
a given product. In terms of what to measure, each
metric has several options. For the sake of brevity,
in this chapter, one quantitative and one qualitative measure for each metric will be considered
(the reader is encouraged to review ISO 9241, pt.
11 as a starting point for considering alternatives).
For efficiency, a quantitative metric could be the
number of mistakes a person made when using the
product, and a qualitative metric could be a subjective workload (using the NASA-TLX mentioned);
for effectiveness, a quantitative metric could be
time to achieve the goal and a qualitative metric
could be a subjective rating of performance; for
satisfaction, a quantitative metric could be time
spent using the device (over the course of several
days) and a qualitative metric could be a self-report
of how pleasant the product was to use. It should be
apparent that the distinction between efficiency, effectiveness, and satisfaction is somewhat arbitrary,
which is why is important to make sure that all three
metrics are applied during evaluation.
The idea that evaluation requires a protocol
is meant to imply that one ought to approach it in
much the same way that one approaches the design
of an experiment, that is, by defining independent
variables, which are the goal, activity, and context
of use, and by defining dependent variables, which
are the usability metrics. The notion of the referent

model also makes sense in terms of experimental

design because the hypothesis under test is that
the outcome will be equal to or better than the
referent model.

Initial Concept Development

During the initial concept development phase, it
is possible that one of the components of context
of use will dominate the others. For example, a
designer might have an idea about how the product
will function or how to perform a particular task
or how to help a particular user. In order to explore
this concept, designers make use of scenarios
in various forms, for example, storyboarding,
sketching, rich pictures, illustrative stories, and
so forth. From Figure 1, it can be argued that a
good scenario would include (as a minimum) some
consideration of the type of person who would
be likely to use the product, the tasks that the
person would perform in order to achieve specific
goals (as well as any other tasks that might need
to be performed concurrently), the environment
in which they might be performing these tasks,
and the presence or use of other products to support this activity. In the domain of traditional
computer systems, the environment can be
assumed to be more or less constant, that is, the
computer would be used on a desk in an office. In
mobile computing, the environment will have
a significant impact on how the product will be
used, as will the range of tasks that the person
will be performing. It is this impact of the environment and the increasing range of concurrent
tasks that makes evaluating mobile technology
different from other computer applications. One
way in which these aspects can be considered is
to develop a scenario in which a person achieves
the defined goal using no technology, another in
which they use contemporary technology and
another in which they use the concept product.
By storyboarding these different scenarios, the
design team gets a feeling for the main benefits
to be gained from using the product (and an

89

Evaluating Mobile Human-Computer Interaction

appreciation as to whether or not to pursue its

development). During this stage, the usability
metrics can be defined in terms of what measures
can sensibly differentiate the product from any
alternative ways of performing the task.

Prototyping
During prototyping different versions of the
product are developed and tested. The prototype
need not be a fully-functioning product. Indeed,
Nilsson et al. (2000) shows how very simple
models can be used to elicit user responses and
behaviors. Their study involved the development
of a handheld device (the pucketizer for use
in water treatment plants and the initial studies
had operators walking around the plant with a
non-functioning object to simulate the device.
From this experience, the design team went on to
implement a functioning prototype based on an
8-bit microcontroller, wireless communications,
and a host computer running a JAVA application).
This work is interesting because it illustrates
how embedding the evaluation process in the
environment and incorporating representative
end-users lead to insights for the design team.
Taking this idea further, it is feasible for very
early prototyping to be based on paper versions.
For example, one might take the form factor of the
intended device (say a piece of wood measuring
5 x 3 x which is approximately the size of
Personal Digital Assistant) and then placing 3 x
2 paper overlays to represent different screen
stateschange the screens is then a matter of
the user interacting with buttons on the product
and the evaluator making appropriate responses.
Of course, this could be done just as easily using
an application in WinCE (or through the use of
a slideshow on the device), but the point is that
initial concepts can be explored well before any
code is written or any hardware built.

90

Specification and Build

Specification and build is the phase that one
might traditionally associate with evaluation.
Evaluation activity at this phase of the design
process would be summative (i.e., occur at the
summation of the process), and would usually
be used to confirm that the design was acceptable prior to committing to manufacture. At this
stage, the main concerns regarding hardware and
software would have been dealt with and so any
usability evaluation that would call for significant
change to hardware or software is likely to be
ignored (unless the product is scrapped and the
process started again, or unless these recommendations are filed for the next version of the
product). However, usability evaluation can play
an important role in this phase because it will form
part of the acceptance testing of end-users and
could, if positive, play a role in defining marketing activity or informing training requirements.

CONCLUSION
While the concept of usability as multi-faceted
might seem straightforward, it raises difficult
problems for the design team. The design team
focuses its attention on the device, but the concept
of usability used in this chapter implies that the
device is only part of the equation and that other
factors relating to the user and environment can
play significant roles. The problem with this, of
course, is that these factors lie outside the remit
of the design team. One irony of this is that a
well-designed device can fail as the result of
unanticipated activity, user characteristics, and
environmental features.
The issue raised in this chapter is that evaluating
mobile technology involves a clear appreciation of
the concept of usability, in line with ISO standard
definitions. The ISO9241 concept of usability emphasizes the need to clearly articulate the context
of use of the device, through consideration of user,

Evaluating Mobile Human-Computer Interaction

activity, device, and environment. This means that

evaluation has to take account of the interactions
between user, activity, device, and environment.
What is essential is that evaluation is conducted in
a way that ensures a good fit between the context
of use in the real-world and that simulated in the
laboratory. This does not mean that one needs to
include all aspects of the real-world in the laboratory but that one is able to reflect key variables and
that the evaluation is designed to ensure a balanced
comparison. It would be easy to prove that a given
device was superior to any other device simply by
ensuring that the test favored the device in question. It is equally easy to prove that evaluation in
the laboratory do not reflect performance in the
real-world. However, such studies often reflect a
limited grasp of adequate experimental design and,
ultimately, a poor understanding of science. One is
not proving that a product is well designed through
evaluation. Rather one is demonstrating fitness-forpurpose under a well-defined context of use.

REFERENCES
Baber, C. (2005). Evaluation of human-computer
interaction. In J. R. Wilson & E. N. Corlett (Eds.),
Evaluation of human work (pp. 357-388). London:
Taylor and Francis.
Baber, C., Arvanitis, T. N., Haniff, D. J., & Buckley, R. (1999). A wearable computer for paramedics: Studies in model-based, user-centered and
industrial design. In M. A. Sasse & C. Johnson
(Eds.), Interact99 (pp. 126-132). Amsterdam:
IOS Press.
Baber, C., Haniff, D. J., Knight, J., Cooper, L., &
Mellor, B. A. (1998). Preliminary investigations
into the use of wearable computers. In R. Winder
(Ed.), People and computers XIII (pp. 313-326).
Berlin: Springer-Verlag.
Baber, C., Haniff, D. J., & Woolley, S. I. (1999).
Contrasting paradigms for the development of

wearable computers. IBM Systems Journal, 38(4),

551-565.
Barnard, L., Yi, J. S., Jacko, J. A., & Sears, A.
(2005). An empirical comparison of use in- motion evaluation scenarios for mobile computing
devices. International Journal of Human-Computer Studies, 62, 487-520.
Barnard, L., Yi, J. S., Jacko, J. A., & Sears, A.
(2007). Capturing the effects of context on human
performance in mobile computing systems. Personal and Ubiquitous Computing, 11(2), 81-96.
Bass, L., Kasabach, C., Martin, R., Siewiorek, D.,
Smailagic, A., & Stivoric, J. (1997). The design of
a wearable computer. In Proceedings of the CHI
97 (pp. 139-146). New York: ACM.
Bass, L., Siewiorek, D., Smailagic, A., & Stivoric,
J. (1995). On site wearable computer system. In
Proceedings of the CHI 95 (pp. 83-88). New
York: ACM.
Bell, M., Chalmers, M., Barkhuus, L., Hall, M.,
Sherwood, S., Tennent, P., Brown, B., Rowland, D.,
Benford, S., Hampshire, A., & Capra, M. (2006).
Interweaving mobile games with everyday life.
In Proceedings of the CHI 2006 (pp. 417-426).
New York: ACM.
Bevan, N. (2001). International standards for HCI
and usability. International Journal of Human
Computer Interaction, 55, 533-552.
Boase, M., Hannigan, S., & Porter, J. M. (1988).
Sorry, cant talk nowjust overtaking a lorry: The
definition and experimental investigation of the
problem of driving and hands-free car phone use.
In E. D. Megaw (Ed.), Contemporary ergonomics.
London: Taylor and Francis.
Brooke, J., (1996). SUS: a quick and dirty usability scale. In P. W. Jordan, B. Weerdmeester, B.
A. Thomas, & I. L. McLelland (Eds.), Usability
evaluation in industry (pp. 189-194). London:
Taylor and Francis.

91

Evaluating Mobile Human-Computer Interaction

Brookhuis et al, (1991). The effects of mobile

telephoning on driving performance. Accident
Analysis and Prevention, 23
Chin, J. P., Diehl, V. A., & Norman, K. L. (1988).
Development of an instrument measuring user
satisfaction of the human-computer interface In
Proceedings of the CHI 88 (pp. 213-218). New
York: ACM.
Cobb, S. V. G., Nichols, S. C., Ramsey, A. D.,
& Wilson, J. R. (1999). Virtual reality induced
symptoms and effects (VRISE). Presence: Teleoperators and Virtual Environments, 8, 169-186.
Dillon, A. (1992). Designing usable electronic
text. London: Taylor and Francis.
Duh, H. B-L., Tan, G. C. B., & Chen, V. H-H.
(2006). Usability evaluation for mobile devices:
a comparison of laboratory and field tests. In Proceedings of Mobile HCI 06 (pp. 181-186). Berlin:
Springer-Verlag.
Earthey, J., Sherwood Jones, B., & Bevan, N.
(2001). The improvement of human-centered
processesFacing the challenges and reaping
the benefit of ISO 13407. International Journal
of Human Computer Studies, 55, 553-585.
Ebersbach, G., Dimitrijrvic, M. R., & Poewe, W.
(1995). Influence of concurrent tasks on gait: A
dual task approach. Perceptual and Motor Skills,
81, 107-113.
Frkjaer, E., Hertzum, M., & Hornbaek, K.
(2000). Measuring usability: are effectiveness,
efficiency and satisfaction really correlated? In
Proceedings of the CHI 2000 (pp. 345-352).
New York: ACM.

Communications of the ACM, 28(3), 300-311.

Gray, W. D., & Salzman, M. C. (1998). Damaged merchandise? A review of experiments that
compare usability evaluation methods. HumanComputer Interaction, 13(3), 203-261.
Hart, S. G., & Staveland, L. E. (1988). Development of NASA-TLX (Task Load Index): Results
of empirical and theoretical research. In P. A.
Hancock & N. Meshkati (Eds.), Handbook of
mental workload. Amsterdam: North Holland.
Holcomb, R., & Tharp, A. L. (1991). What users
say about software usability. International Journal
of Human-Computer Interaction, 3(1), 49-78.
ISO 18021. (2001). Information technologyUser
interface for mobiles. ISO9241-11 1998.
ISO 9241. (1998). Ergonomics of office work with
VDTsGuidance on usability. Geneva: International Standards Office.
Jeffries, R., Miller, J. R., Wharton, C., & Uyeda,
K. M. (1991). User interface evaluation in the real
world: A comparison of four techniques (CHI91)
(pp. 119-124). New York: ACM.
Johnson, P. (1992). Human-computer interaction:
psychology, task analysis and software engineering. London: McGraw-Hall.
Johnson, P. (1998). Usability and mobility. Paper
presented at the 1st Workshop on Human-Computer Interaction with Mobile Devices, Glasgow.
Kirakowski, J., & Corbett, M. (1988). Measuring
user satisfaction. In D. M. Jones & R. Winder
(Eds.), People and computers IV (pp. 329-430).
Cambridge: Cambridge University Press.

Good, M., Spine, T. M., Whiteside, J., & George,

P. (1986). User-derived impact analysis as a tool
for usability engineering. In Proceedings of the
CHI 86 (pp. 241-246). New York: ACM.

Kirakowski, J., & Corbett, M. (1993). SUMI:

the software usability measurement inventory.
British Journal of Educational Technology, 24,
210-214.

Gould, J. D., & Lewis, C. H. (1985). Designing for

usability: Key principles and what designers think.

Kjeldskov, J., & Stage, J. (2004). New techniques

for usability evaluation of mobile systems. Inter-

92

Evaluating Mobile Human-Computer Interaction

national Journal of Human-Computer Studies,

60, 599-620.
Knight, J. F., & Baber, C. (2005). A tool to assess the comfort of wearable computers. Human
Factors, 47, 77-91.
Knight, J. F., Baber, C., Schwirtz, A., & Bristow, H.
(2002). The comfort assessment of wearable computers. Digest of Papers of the 6th International
Symposium on Wearable Computing (pp. 65-74).
Los Alamitos, CA: IEEE Computer Society.
Monk, A. F., Wright, P. C., Haber, J., & Davenport, L. (1986). Improving your human-computer
interface: A practical technique. London: Prentice-Hall.
Nielsen, J. (1993). Usability engineering. London:
Academic Press.
Nilsson, J., Sokoler, T., Binder, T., & Wetcke, N.
(2000). Beyond the control room: Mobile devices
for spatially distributed interaction on industrial
process plants. In P. Thomas & H. W. Gellerson
(Eds.), Handheld and ubiquitous computing:
Second international symposium (pp. 30-45).
Berlin: Springer.

Whiteside, J., Bennett, J., & Holtzblatt, K.

(1988). Usability engineering: Our experience
and evolution. In M. Helander (Ed.), Handbook
of human-computer interaction (pp. 791-817).
Amsterdam: Elsevier.
Wichansky, A. M. (2000). Usability testing
in 2000 and beyond. Ergonomics, 43(7), 9981006.
Wikman, A., Nieminen, T., & Summala, H. (1998).
Driving experience and time sharing during in-car
tasks on roads of different widths. Ergonomics
4, 358-372.
Wilson, J. R., & Nichols, S. C. (2002). Measurement in virtual environments: Another dimension
to the objectivity/subjecitivity debate. Ergonomics, 45, 1031-1036.
Witmer, B. G., & Singer, M. J. (1998). Measuring
presence in virtual environments: a presence questionnaire. Presence: Teleoperators and Virtual
Environments, 7, 225-240.

Key Terms

Nunes, L., & Recarte, M. A. (2002). Cognitive

demands of hands-free phone conversation while
driving. Transportation Research Part F: Traffic
Psychology and Behavior, 5, 133-144.

Context of Use: The combination of user, task,

product, and environment during the achievement
of a desired goal

Ravden, S. J., & Johnson, G. I. (1989). Evaluating

usability of human-computer interfaces. Chichester: Ellis Horwood.

Dual-Task: The performance of two (or more)

tasks at the same time. This could involve simultaneous performance or could involve some form
of time-sharing between the tasks.

Slater, M., Usoh, M., & Steed, A. (1994). Depth

of presence in virtual environments. Presence:
Teleoperators and Virtual Environments, 3,
130-144.
Svenson, O., & Patten, C. J. D. (2005). Mobile
phones and driving: A review of contemporary
research. Cognition, Technology and Work, 7,
182-197.

Effectiveness: The ability of a given user to

employ a given product to achieve a desired goal
in a given context of use
Efficiency: The optimal expenditure of resources by a given user in using a given product to
achieve a desired goal in a given context of use
Referent Model: A product (or set of metrics)
against which a given product can be compared

93

Evaluating Mobile Human-Computer Interaction

Satisfaction: The subjective response of a

user to interacting with a product
Usability: the extent to which a product
can be used by specified users to achieve specified
goals with effectiveness, efficiency and satisfaction in a specified context of use. (ISO9241-11,
1998)

Endnotes

SUS: Software Usability Scale (Brooke,
1996)
2

SUMI: Software Usability Metrics Inventory, Kirakowski and Corbett (1993)
1

This work was previously published in Handbook of Research on User Interface Design and Evaluation for Mobile Technology, edited by J. Lumsden, pp. 731-744, copyright 2008 by Information Science Reference, formerly known as Idea Group
Reference (an imprint of IGI Global).

94

95

Chapter 1.8

An Overview of Multimodal
Interaction Techniques and
Applications
Marie-Luce Bourguet
Queen Mary University of London, UK

INTRODUCTION
Desktop multimedia (multimedia personal computers) dates from the early 1970s. At that time,
the enabling force behind multimedia was the
emergence of the new digital technologies in
the form of digital text, sound, animation, photography, and, more recently, video. Nowadays,
multimedia systems mostly are concerned with
the compression and transmission of data over
networks, large capacity and miniaturized storage devices, and quality of services; however,
what fundamentally characterizes a multimedia
application is that it does not understand the data
(sound, graphics, video, etc.) that it manipulates.
In contrast, intelligent multimedia systems at the
crossing of the artificial intelligence and multimedia disciplines gradually have gained the ability
to understand, interpret, and generate data with
respect to content.

Multimodal interfaces are a class of intelligent

multimedia systems that make use of multiple
and natural means of communication (modalities), such as speech, handwriting, gestures, and
gaze, to support human-machine interaction.
More specifically, the term modality describes
human perception on one of the three following
perception channels: visual, auditive, and tactile.
Multimodality qualifies interactions that comprise
more than one modality on either the input (from
the human to the machine) or the output (from
the machine to the human) and the use of more
than one device on either side (e.g., microphone,
camera, display, keyboard, mouse, pen, track
ball, data glove). Some of the technologies used
for implementing multimodal interaction come
from speech processing and computer vision;
for example, speech recognition, gaze tracking,
recognition of facial expressions and gestures,
perception of sounds for localization purposes,
lip movement analysis (to improve speech rec-

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

An Overview of Multimodal Interaction Techniques and Applications

ognition), and integration of speech and gesture

information.
In 1980, the put-that-there system (Bolt, 1980)
was developed at the Massachusetts Institute of
Technology and was one of the first multimodal
systems. In this system, users simultaneously
could speak and point at a large-screen graphics
display surface in order to manipulate simple
shapes. In the 1990s, multimodal interfaces started
to depart from the rather simple speech-and-point
paradigm to integrate more powerful modalities
such as pen gestures and handwriting input (Vo,
1996) or haptic output. Currently, multimodal
interfaces have started to understand 3D hand
gestures, body postures, and facial expressions
(Ko, 2003), thanks to recent progress in computer
vision techniques.

BACKGROUND
In this section, we briefly review the different
types of modality combinations, the user benefits brought by multimodality, and multimodal
software architectures.

Combinations of Modalities
Multimodality does not consist in the mere
juxtaposition of several modalities in the user
interface; it enables the synergistic use of different
combinations of modalities. Modality combinations can take several forms (e.g., redundancy and
complementarity) and fulfill several roles (e.g.,
disambiguation, support, and modulation).
Two modalities are said to be redundant when
they convey the same information. Redundancy is
well illustrated by speech and lip movements. The
redundancy of signals can be used to increase the
accuracy of signal recognition and the overall robustness of the interaction (Duchnowski, 1994).
Two modalities are said to be complementary
when each of them conveys only part of a message but their integration results in a complete

96

message. Complementarity allows for increased

flexibility and efficiency, because a user can select
the modality of communication that is the most
appropriate for a given type of information.
Mutual disambiguation occurs when the
integration of ambiguous messages results in
the resolution of the ambiguity. Let us imagine
a user pointing at two overlapped figures on a
screen, a circle and a square, while saying the
square. The gesture is ambiguous because of the
overlap of the figures, and the speech also may
be ambiguous if there is more than one square
visible on the screen. However, the integration of
these two signals yields a perfectly unambiguous
message.
Support describes the role taken by one modality to enhance another modality that is said
to be dominant; for example, speech often is accompanied by hand gestures that simply support
the speech production and help to smooth the
communication process.
Finally, modulation occurs when a message
that is conveyed by one modality alters the content of a message conveyed by another modality.
A persons facial expression, for example, can
greatly alter the meaning of the words he or she
pronounces.

User Benefits
It is widely recognized that multimodal interfaces,
when carefully designed and implemented, have
the potential to greatly improve human-computer
interaction, because they can be more intuitive,
natural, efficient, and robust.
Flexibility is obtained when users can use
the modality of their choice, which presupposes
that the different modalities are equivalent (i.e.,
they can convey the same information). Increased
robustness can result from the integration of
redundant, complementary, or disambiguating
inputs. A good example is that of visual speech
recognition, where audio signals and visual signals
are combined to increase the accuracy of speech

An Overview of Multimodal Interaction Techniques and Applications

recognition. Naturalness results from the fact that

the types of modalities implemented are close to
the ones used in human-human communication
(i.e., speech, gestures, facial expressions, etc.).

Software Architectures
In order to enable modality combinations in the
user interface, adapted software architectures
are needed. There are two fundamental types
of multimodal software architectures, depending on the types of modalities. In feature level
architectures, the integration of modalities is performed during the recognition process, whereas
in semantic level architectures, each modality
is processed or recognized independently of the
others (Figure 1).
Feature-level architectures generally are
considered appropriate for tightly related and
synchronized modalities, such as speech and lip
movements (Duchnowski et al., 1994). In this
type of architecture, connectionist models can
be used for processing modalities because of
their good performance as pattern classifiers and
because they easily can integrate heterogeneous
features. However, a truly multimodal connectionist approach is dependent on the availability of
multimodal training data, and such data currently
is not available.
When the interdependency between modalities
implies complementarity or disambiguation (e.g.,
speech and gesture inputs), information typically

is integrated into semantic-level architectures (Nigay et al., 1995). In this type of architecture, the
main approach for modality integration is based
on the use of data structures called frames. Frames
are used to represent meaning and knowledge and
to merge information that results from different
modality streams.

MAIN ISSUES IN MULTIMODAL

INTERACTION
Designing Multimodal Interaction
Recent developments in recognition-based interaction technologies (e.g., speech and gesture
recognition) have opened a myriad of new possibilities for the design and implementation of
multimodal interfaces. However, designing systems that take advantage of these new interaction
techniques are difficult. Our lack of understanding of how different modes of interaction can
be combined best into the user interface often
leads to interface designs with poor usability.
Most studies to understand natural integration of
communication modes are found in the experimental psychology research literature, but they
tend to qualitatively describe human-to-human
communication modes. Very few attempts have
been made so far to qualitatively or quantitatively
describe multimodal human-computer interaction
(Bourguet, 1998; Nigay, 1995; Oviatt, 1997). Much
more work is still needed in this area.

Figure 1. Multimodal software architectures

97

An Overview of Multimodal Interaction Techniques and Applications

Implementing Multimodality
Developers still face major technical challenges for
the implementation of multimodality, as indeed,
the multimodal dimension of a user interface raises
numerous challenges that are not present in more
traditional interfaces (Bourguet, 2004). These
challenges include the need to process inputs from
different and heterogeneous streams; the coordination and integration of several communication
channels (input modalities) that operate in parallel
(modality fusion); the partition of information
sets across several output modalities for the
generation of efficient multimodal presentations
(modality fission); dealing with uncertainty and
recognition errors; and implementing distributed
interfaces over networks (e.g., when speech and
gesture recognition are performed on different
processors). There is a general lack of appropriate
tools to guide the design and implementation of
multimodal interfaces.
Bourguet (2003a, 2003b) has proposed a simple
framework, based on the finite state machine
formalism, for describing multimodal interaction
designs and for combining sets of user inputs of
different modalities. The proposed framework can
help designers in reasoning about synchronization patterns problems and testing interaction
robustness.

Uncertainty in Multimodal Interfaces

Natural modalities of interaction, such as speech
and gestures, typically rely on recognition-based
technologies that are inherently error prone.
Speech recognition systems, for example, are sensitive to vocabulary size, quality of audio signal,
and variability of voice parameters (Halverson,
1999). Signal and noise separation also remains
a major challenge in speech recognition technology, as current systems are extremely sensitive to
background noise and to the presence of more than
one speaker. In addition, slight changes in voice
quality (due, for example, to the speaker having a

98

cold) can significantly affect the performance of a

recognizer, even after the user has trained it.
Several possible user strategies to prevent or
correct recognition errors have been uncovered.
Oviatt (2000) shows that in order to avoid recognition errors, users tend to spontaneously select
the input mode they recognize as being the most
robust for a certain type of content (modality
selection strategy). When recognition errors occurr, Suhm (2001) suggests that users be willing
to repeat their input at least once, after which they
will tend to switch to another modality (modality
switching strategy). Finally, Oviat (2000) reports
cases of linguistic adaptation, where users choose
to reformulate their speech in the belief that it
can influence error resolutiona word may be
substituted for another, or a simpler syntactic
structure may be chosen. Overall, much more
research is still needed to increase the robustness
of recognition-based modalities.

APPLICATIONS
Two applications of multimodal interaction are
described.

Augmented Reality
Augmented reality is a new form of multimodal
interface in which the user interacts with realworld objects and, at the same time, is given
supplementary visual information about these
objects (e.g., via a head mounted display). This
supplementary information is context-dependent
(i.e., it is drawn from the real objects and fitted to
them). The virtual world is intended to complement
the real world on which it is overlaid. Augmented
reality makes use of the latest computer vision
techniques and sensor technologies, cameras, and
head-mounted displays. It has been demonstrated,
for example, in a prototype to enhance medical
surgery (Dubois, 1999).

An Overview of Multimodal Interaction Techniques and Applications

Tangible Interfaces
People are good at sensing and manipulating
physical objects, but these skills seldom are
used in human-computer interaction. Tangible
interfaces are multimodal interfaces that exploit
the tactile modalities by giving physical form to
digital information (Ishii, 1997). They implement physical objects, surfaces, and textures as
tangible embodiments of digital information. The
tangible query interface, for example, proposes
a new means for querying relational databases
through the manipulation of physical tokens on
a series of sliding racks.

FUTURE TRENDS
Ubiquitous Computing
Ubiquitous computing describes a world from
which the personal computer has disappeared
and has been replaced by a multitude of wireless,
small computing devices embodied in everyday
objects (e.g., watches, clothes, or refrigerators).
The emergence of these new devices has brought
new challenges for human-computer interaction.
A fundamentally new class of modalities has
emergedthe so-called passive modalitiesthat
corresponds to information that is automatically
captured by the multimodal interface without any
voluntary action from the user. Passive modalities
complement the active modalities such as voice
command or pen gestures.
Compared with desktop computers, the screens
of ubiquitous computing devices are small or
non-existent; small keyboards and touch panels
are hard to use when on the move, and processing
powers are limited. In response to this interaction
challenge, new modalities of interaction (e.g.,
non-speech sounds) (Brewster, 1998) have been
proposed, and the multimodal interaction research
community has started to adapt traditional multimodal interaction techniques to the constraints

of ubiquitous computing devices (Branco, 2001;

Schaefer, 2003; Schneider, 2001).

CONCLUSION
Multimodal interfaces are a class of intelligent
multimedia systems that extends the sensorymotor capabilities of computer systems to better
match the natural communication means of human beings. As recognition-based technologies
such as speech recognition and computer vision
techniques continue to improve, multimodal interaction should become widespread and eventually
may replace traditional styles of human-computer
interaction (e.g., keyboard and mice). However,
much research still is needed to better understand
users multimodal behaviors in order to help designers and developers to build natural and robust
multimodal interfaces. In particular, ubiquitous
computing is a new important trend in computing
that will necessitate the design of innovative and
robust multimodal interfaces that will allow users
to interact naturally with a multitude of embedded
and invisible computing devices.

REFERENCES
Bolt, R. A. (1980). Put-that-there: Voice and gesture at the graphics interface. Proceedings of the
7th Annual Conference on Computer Graphics and
Interactive Techniques. Seattle, Washington.
Bourguet, M. L. (2003a). Designing and prototyping multimodal commands. Proceedings
of the IFIP TC13 International Conference on
Human-Computer Interaction, INTERACT03,
Zurich, Switzerland.
Bourguet, M. L. (2003b). How finite state machines can be used to build error free multimodal
interaction systems. Proceedings of the 17th British
Group Annual Conference, Bath, UK.

99

An Overview of Multimodal Interaction Techniques and Applications

Bourguet, M. L. (2004). Software design and

development of multimodal interaction. Proceedings of the IFIP 18th World Computer Congress
Topical Days.
Bourguet, M. L., & Ando, A. (1998). Synchronization of speech and hand gestures during multimodal human-computer interaction. Proceedings
of the Conference on Human Factors in Computing Systems, Los Angeles, California.
Branco, P. (2001). Challenges for multimodal interfaces towards anyone anywhere accessibility:
A position paper. Proceedings of the 2001 Workshop on Universal Accessibility of Ubiquitous
Computing: Providing for the Elderly. Alccer
do Sal, Portugal.
Brewster, S., Leplatres, G., & Crease, M. (1998).
Using non-speech sounds in mobile computing
devices. Proceedings of the 1st Workshop on Human Computer Interaction with Mobile Devices,
Glasgow, Scotland.
Dubois, E., Nigay, L., Troccaz, J., Chavanon, O.,
& Carrat, L. (1999). Classification space for augmented surgery: An augmented reality case study.
Proceedings of Seventh IFIP TC13 International
Conference on Human-Computer Interaction,
Edinburgh, UK.
Duchnowski, P., Meier, U., & Waibel, A. (1994).
See me, hear me: Integrating automatic speech
recognition and lipreading. Proceeding of the
International Conference on Spoken Language
Processing, Yokohama, Japan.
Halverson, C., Horn, D., Karat, C., & Karat, J.
(1999). The beauty of errors: Patterns of error
correction in desktop speech systems. Proceedings of the Seventh IFIP TC13 International
Conference on Human-Computer Interaction,
Edinburgh, Scotland.
Ishii, H., & Ullmer, B. (1997). Tangible bits:
Towards seamless interfaces between people,
bits and atoms. Proceedings of the ACM Confer-

100

ence on Human Factors in Computing Systems,

Atlanta, Georgia.
Ko, T., Demirdjian, D., & Darrell, T. (2003). Untethered gesture acquisition and recognition for a
multimodal conversational system. Proceedings
of the 5th International Conference on Multimodal
Interfaces, Vancouver, Canada.
Nigay, L., & Coutaz, J. (1995). A generic platform
for addressing the multimodal challenge. Proceedings of the Conference on Human Factors
in Computing Systems.
Oviatt, S. (2000). Taming recognition errors with
a multimodal interface. Communications of the
ACM, 43(9), 45-51.
Oviatt, S., De Angeli, A., & Kuhn, K. (1997).
Integration and synchronisation of input modes
during multimodal human-computer interaction. Proceedings of the ACM Conference on
Human Factors in Computing Systems, Atlanta,
Georgia.
Schaefer, R., & Mueller, W. (2003). Multimodal
interactive user interfaces for mobile multi-device
environments. Proceedings of the Ubicomp 2003
Workshop Multi-Device Interfaces for Ubiquitous
Peripheral Interaction, Seattle, Washington.
Schneider, G., Djennane, S., Pham, T. L., &
Goose, S. (2001). Multimodal multi device UIs
for ubiquitous access to multimedia gateways.
Proceedings of the 17th International Joint Conference on Artificial Intelligence and Workshop
Artificial Intelligence In Mobile Systems, Seattle,
Washington.
Suhm, B., Myers, B., & Waibel, A. (2001). Multimodal error correction for speech user interfaces.
ACM Transactions on Computer-Human Interaction, 8(1), 60-98.
Vo, M. T., & Wood, C. (1996). Building an application framework for speech and pen input
integration in multimodal learning interfaces.
Proceedings of the IEEE International Confer-

An Overview of Multimodal Interaction Techniques and Applications

ence on Acoustics, Speech, and Signal Processing.

Atlanta, Georgia.

Modality Fusion: Integration of several modality inputs in the multimodal architecture to

reconstruct a users command.

KEY TERMS

Mutual Disambiguation: The phenomenon

in which an input signal in one modality allows
recovery from recognition error or ambiguity in
a second signal in a different modality is called
mutual disambiguation of input modes.

Active Modality: Modality voluntarily and

consciously used by users to issue a command
to the computer; for example, a voice command
or a pen gesture.
Feature-Level Architecture: In this type of
architecture, modality fusion operates at a low
level of modality processing. The recognition
process in one modality can influence the recognition process in another modality. Feature-level
architectures generally are considered appropriate
for tightly related and synchronized modalities,
such as speech and lip movements.
Haptic Output: Devices that produce a tactile
or force output. Nearly all devices with tactile
output have been developed for graphical or
robotic applications.
Modality Fission: The partition of information sets across several modality outputs for the
generation of efficient multimodal presentations.

Passive Modality: Information that is captured automatically by the multimodal interface;

for example, to track a users location via a microphone, a camera, or data sensors.
Semantic-Level Architecture: In semantic
level architectures, modalities are integrated at
higher levels of processing. Speech and gestures,
for example, are recognized in parallel and independently. The results are stored in meaning
representations that then are fused by the multimodal integration component.
Visual Speech Recognition: Computer vision
techniques are used to extract information about
the lips shape. This information is compared with
information extracted from the speech acoustic
signal to determine the most probable speech
recognition output.

This work was previously published in Encyclopedia of Human Computer Interaction, edited by C. Ghaoui, pp. 451-456, copyright 2006 by Information Science Reference, formerly known as Idea Group Reference (an imprint of IGI Global).

101

102

Chapter 1.9

Multimodal and Federated

Interaction
Frankie James
SAP Research, USA
Rama Gurram
SAP Research, USA

Abstract

The Basics of Multimodality

This chapter introduces the concepts of multimodal and federated interaction. Because multimodality means, simply, the combination of
multiple modalities (or types of input and output),
the authors first introduce some of the various
modalities available for computer interaction. The
chapter then discusses how multimodality can be
used both in desktop and mobile computing environments. The goal of the chapter is to familiarize
scholars and researchers with the range of topics
covered under the heading multimodality and
suggest new areas of research around the combination of modalities, as well as the combination
of mobile and stationary computing devices to
improve usability.

As was discussed in the introduction to Liability,

interaction with a computer (or computer-based
device) can take place using a variety of different
forms or modalities. On the input side, information
can be transferred from a human operator to the
computer via keyboards, keypads, touch screens,
mice, joysticks, spoken language, or even gesture
and motion sensors. Information can be output
through visual displays (large and small), audio
displays (including spoken text and non-speech
sounds; see also Mobile Speech Recognition),
and tactile displays (such as Braille or raised-line
displays), as well as more exotic forms, such as
force-feedback (haptic) joysticks and mice, and
olfactory (aroma-based) displays.
Each of the input and output channels mentioned above have their own benefits and limita-

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Multimodal and Federated Interaction

tions. For example, large visual displays are able

to present a great deal of information at the same
time, which users can quickly scan to find the data
that is relevant to their current needs. However,
visual displays are not appropriate for blind users, and they are also inappropriate for mobile
users who are not able to stand or sit in one place
to read the display (or carry it with them while
traveling).
The purpose of this chapter is to discuss the
concepts of multimodal interaction, where two
or more modalities (such as vision or audio) are
combined. Multimodal interaction is frequently
used to compensate for limitations in one interaction modality by providing a second one. For
example, the limited visual display capabilities of
a mobile device can be augmented by providing
audio output, or speech input can be provided
to a user with limited typing ability to increase
data entry speed. Multiple modalities can also
be combined within a single input (or output) to
increase efficiency; a seminal example here is the
combination of a spoken action command (e.g.,
color this red) with mouse or touch-screen
selection of the object to be acted upon. There
are, of course, many other reasons for using
multimodality, which will be discussed later in
this chapter.
We will begin in this introduction with a discussion of the different forms of multimodality,
as well as the different purposes to which multimodal interactions can be applied. In the next two
sections, we address the use of multimodality for
desktop applications and on mobile devices, where
the reasons and methods for using multimodality
can be quite different. We then discuss the concept of device federation, where multiple devices
(each with their own available modalities) can be
combined within a single interaction. Finally, we
conclude with the chapter summary.

Different Forms
The basic definition of multimodality is the use
of more than one modality within a single interface. The availability of both keyboard and voice
input is one of the most common examples of
multimodality, as is the use of both visual (text
or graphical) and audio output. Most of the five
classical human senses (sight, hearing, touch,
smell, and taste) can be used for both the input
and output sides. Each sense allows a broad range
of possibilities.
Table 1 gives a brief list of the types of input
and output that are associated with the senses.
The most common use of the sense of sight
is in the visual presentation (output) of information on small and large displays. Sight can also
be used for input: eye tracking can be used for
selection or to gauge interest in a particular area
of a screen, and retinal scanning can be used to
identify the user.
Input options based on the sense of hearing include speech, for entering text or giving
commands, speaker identification (to identify or
authenticate the user), and even humming (Ghias
et al., 1995). Audio output can be used for presenting written text (using text-to-speech), recorded
audio files, document and interface structures,
and sonifications of graph data [see the chapter
Mobile Speech Recognition and James (1998)
for an overview]. Speech input and audio-based
output are useful in a variety of contexts, including mobile and vehicle-based scenarios, as well
as accessibility.
The sense of touch is already commonly found
in computer inputs today, through the use of
keyboards, pointing devices, and touch screens.
In addition to detecting simply that a key, button,
or screen area has been clicked or touched, more
advanced devices (such as game controllers and
track pads) can also detect the amount of pressure
exerted by the user. Handwriting and gesture are
also gaining in popularity within certain contexts,
along with the use of fingerprints for user iden-

103

Multimodal and Federated Interaction

Table 1. Human senses and associated input and output types

Sense

Input Types

Output Types

Sight

Eye tracking
Retinal scan

Visual displays (small and

large) of graphics and text

Speech

Text-to-speech

Speaker identification

Recorded audio

Humming

Auditory icons and earcons

Hearing

Sonification
Touch/Gesture

Keyboard

Tactile display

Mouse/pointing device

Braille display

Handwriting

Haptic/force feedback display

Gesture and motion

Pressure sensing
Heat sensing
Fingerprint recognition
Smell

Chemical odor-detecting
sensors

Olfactory output

Taste

(currently none)

Gustatory output

tification. In the near future, we can expect to

see heat and motion-sensing inputs, possibly for
tracking people and objects. Tactile and haptic displays are also used for output, especially to make
computers accessible to disabled users. Tactile
displays include Braille and raised-line displays,
which can be felt with the fingertips or hands to
recognize shapes and text. Haptic displays use
force feedback, generally to the hands and fingers,
to present textures and boundaries that represent
elements within the user interaction.
Smell and taste have spawned the creation
of some new output possibilities, including the
intriguing, if somewhat irreverent, Edible Bits
(Maynes-Aminzade, 2005) gustatory display. The
applicability of these senses to human-computer
interaction, however, appears to be limited to
specialized applications and artistic contexts.
Multimodal applications differ in which modalities and techniques they choose to combine,
as well as in the ways that the modalities are used
together. Some applications use modalities in a
redundant way, so that input can be given in any

104

of the available modalities, and the same output

is rendered simultaneously in multiple modalities.
Other applications use the modalities to complement one another, combining multiple inputs or
outputs to produce a single command or piece of
output (also called multimodal fusion). The choice
of whether to use redundant or complementary
modalities is dependent on the goals, context of
use, and target user group for the application.
Examples and use cases for both of these methods
are described below.

Different Purposes
Applications are designed to use multimodal interactions for a wide variety of reasons. Having
several modalities available can allow designers
to create applications that are easier or more
natural to use, or that can be used by more users
within more contexts. Table 2 shows the different
purposes of using multimodality, categorized by

Multimodal and Federated Interaction

Table 2. Purposes for using multimodality

Type of Multimodality

Purpose

Complementary Modalities
(Multimodal Fusion)

o
o
o

natural interaction; good mapping to task

avoid clutter/overload (output)
disambiguation (input) non-mobile

Replacement Modalities

o
o
o

natural interaction; good mapping to task

reduce screen clutter
compensate for limited input (or output) capabilities mobile

Redundant Modalities

o
o
o
o

accessibility
improve efficiency
compensate for hands- or eyes-busy tasks mobile
compensate for environment mobile

the way the modalities are combined in the interaction. In many cases, the purposes are valid both
for mobile and non-mobile (i.e., desktop) devices;
in other cases, there is special value in using multimodality for specific device categories.
The most common usage of complementary
modalities (or multimodal fusion) is to attempt
to make the interaction more natural, where modalities are chosen that are the best fit to different
parts of the task. For example, pointing and other
gestures are appropriate for selection, while voice
may be more efficient for entering text. Complementary modalities can also be used to spread
the input or output across different modalities,
reducing the load on any one channel. One example here is on the output side, where the most
important or urgent information is presented in
audio while supporting information is presented
visually. The benefit is that the user can get both
the key and supporting information simultaneously by listening to the audio and attending to
the visuals at the same time.
Replacement modalities, as the name implies,
replace one modality for another. Replacement
can produce a more natural interaction, when,
for example, keyboard entry of a password is replaced by speaker identification in a system that
already allows voice input. On a mobile device,
replacement can also be used to compensate for
the limited capabilities of the devicelong por-

tions of text output are difficult to read and scroll

through on a small screen, and may be easier to
comprehend using an audio presentation.
Redundant modalities provide users with the
ability to choose from different interaction options. By allowing both voice and keyboard input,
for example, an application is more accessible to
users who cant type well or who have a physical
disability. Redundant output, presented both visually and auditorially, can benefit not only users
with visual disabilities, but also users performing
tasks requiring their visual attention to be focused
elsewhere (for example, while driving). The remainder of this chapter presents these purposes
in more detail within the context of example applications and research prototypes.

Multimodality on the
Desktop
Because desktop computers are still the primary
computing environment for most users, we begin
our discussion of multimodality on this platform.
People frequently see multimodality as unnecessary or even inadvisable on desktop computersI dont want to talk to my computer, or
have it talk back to me, in my officesomeone
might hear some of my private information!
This section will show that there are many cases

105

Multimodal and Federated Interaction

where the uses of multimodal input or output

(even speech and audio) are indeed very useful
tools for desktop interaction.
In desktop settings, there are several uses of
multimodality:
1.

2.
3.

To make the interaction more natural, by

mapping information to be input or output
to the most appropriate modality;
To provide better and more usable security;
To increase accessibility for disabled users.

This section discusses each of these uses,

and concludes with the presentation of an area of
possible research around the use of multimodal
output in desktop computing.

Using Multimodality to Make

Interaction More Natural
Standard desktop systems use keyboard and
mouse for input, with visual displays and some
(limited) audio output. For many interactions,
these tools and modalities are appropriate; for
example, selecting an icon on a desktop or some
text in a document is straightforward using a
mouse or other pointing device. However, other
interactions involving pointing devices require
far more effort on the part of the user, when the
mapping requires users to search long menus
or complicated tool bars for desired commands
(for example, when specifying the attributes of
a shape or icon placed on a canvas). Keyboards,
while useful for entering text into a document, are
not as useful for issuing commands or entering
passwords, where arcane syntax and structures
must be remembered. By providing more input
modalities to a desktop computer user, interactions can be optimized to be more natural, and
may also reduce the amount of visual clutter on
the display.

106

Speech is perhaps the most commonly available input modality after keyboards and pointing
devices (although as Oviatt [1999] points out, it
should by no means be considered the primary
information carrying modality within a multimodal system). Speech recognition systems use
grammars to define the commands or phrases
that can be accepted by the system. Speech recognition grammars can range from types found
in standard computer theory (regular grammars,
context-free grammars) to statistical grammars
(e.g., n-gram grammars, where the system uses
the preceding n-1 words spoken by the user to
help identify the nth word).

Replacement Modalities
Within the desktop context, speech input can be
used to make an interaction more natural and
also more visually appealing, by eliminating the
need to display complicated tool bars and menus.
A striking example of this is the CommandTalk
system, which enabled users to create, control,
and modify battlefield missions and control the
map display of the ModSAF battlefield system
using spoken language and mouse input (Moore
et al., 1997). Through the addition of the CommandTalk speech interface, the researchers were
able to reduce the screen clutter of ModSAF,
eliminating almost all of the tool bars that had
previously been required for specifying details
of checkpoints and ordinances, controlling the
display, and specifying mission movements.
The architecture of CommandTalk included
many natural language processing components
to create an intuitive command language. The
speech recognition grammar was designed to
allow the same commands and object referents as
would be used in a human-to-human interaction.
Complications regarding the resolution of noun
phrases (the M1 platoon) to specific units shown
on the display, and the mapping of predicates
to interface actions (move, which can map to
one of two different commands depending on

Multimodal and Federated Interaction

the type of platoon to be moved) were handled

automatically and made transparent to the user. By
designing the grammar using a corpus of sample
human-to-human dialogues, the interaction is
made extremely natural for users who already
understand the domain, and the system can be
used without much initial training.
In CommandTalk, the spoken language interface provides an alternative means of input.
CommandTalk also changed the visual display
of the underlying application, although this is not
strictly necessary when creating a multimodal
version of an existing application. The available
input modality, speech, was in this case used as
a replacement modality. Replacing one modality with another should, however, be done with
caution. The spoken language input for CommandTalk was designed to be natural for expert
users in the domain, based on sample dialogues.
Non-expert users who are unfamiliar with the
domain will require some time to learn the command structure that is used for the task. Toolbars
and buttons can act as visible reminders of the
available commandsthese reminders allow
users (especially non-experts) to build up their
understanding of the domain using recognition
memory. Removing the visible reminders forces
users to rely on recall memory (Lodding, 1982).
In multimodal systems designed for non-expert
users, it may be more appropriate in many cases
to add a new modality without removing the old;
in this case, the speech input would have been a
redundant modality.

Complementary Modalities
Input modalities can also be used in a complementary way, where the inputs from several modalities
are fused to produce a single input command.
This is commonly known as multimodal fusion.
Multimodal fusion allows users to mix and match
modalities as they choose, and the system attempts
to find a coherent meaning for the set of inputs.
For example, when speech and gesture input

are combined, spoken commands can be used

to specify actions, while mouse or pen gestures
specify the objects or action details. This can be
much more efficient or more usable than using the
mouse to both select the object and the command,
since the two parts can be done simultaneously
and because the user does not have to search
the visual menus or tool bars for the command.
Combining voice and a pointing device is also
more natural than using voice alone, since object
selection and many other tasks (such as resizing
and moving) can be tedious and error-prone to
accomplish using voice alone (see, for example,
the Voice-Enabled Portal project, described later)
(Grasso et al., 1998; Tyfa & Howes, 2000).
Many multimodal fusion systems focus on the
use of speech with other modalities. One of the
first systems to combine voice and gesture was the
Put-That-There system (Bolt, 1980). The system
focused on the basic concept of combining voice
and gesture into what Bolt called a concerted,
natural user modality. The system allowed users
to place shapes on a large screen display, when
seated facing the screen. Gesture sensors were
used to determine the placement of the shape,
while speech commands gave additional information, including the type of shape, color, and size.
The gestures disambiguated deictic references
(references whose interpretation is relative to the
context of the utterance) in the spoken commands,
such as that and there.a As Oviatt (1999)
points out, deictic commands tend to dominate
multimodal input, since they are also widely used
in human-to-human communication.
The combination of speech with gestures has
most commonly been demonstrated in map-related tasks. Koons

et al.
(1993) describe a system
that integrates speech, gesture, and eye gaze input
to query objects on a map. Their focus is on the
interpretation of multimodal commands from the
three modalities, using reasoning to combine the
modalities and determine the objects and actions
requested. The Multimodal Maps application
designed by Adam Cheyer and Luc Julia (Cheyer

107

Multimodal and Federated Interaction

& Julia, 1998) is another example of a system

that fuses gesture and voice inputs to produce a
natural, less error-prone application (their paper is
also a good reference to other publications related
to multimodal fusion). The application is a travel
planning tool that shows locations of interest on a
map display. Users can ask questions using either
modality, or both in combination. For example,
users can give the spoken command show restaurants within one mile of this hotel, specifying
the hotel by circling it on the map.
Multimodal commands involving speech can
be formed and used in a variety of ways, and
systems built to understand them have to be able
to handle these differences. For example, while
some commands may be formed by actions that
co-occur, such as a circling gesture or pointing
action that occurs simultaneously with the speech,
many commands that should be combined are
temporally separated: the gesture or pointing action may occur either before or after the spoken
command. It is extremely important to properly
associate the actions, since the information given
in the spoken command may be impossible to
fully interpret without the gesture, and vice
versa. Standard algorithms pay attention to the
temporal ordering of actions in various modalities,
evaluating whether two actions are related based
on how long the gap is between the actions,
whether the individual actions themselves can
be well-formed commands, and other criteria.
Even systems that are designed with multimodal
fusion in mind must be able to support unimodal
commands (independent gestures or speech), since
users typically intermix unimodal and multimodal
commands (Oviatt, 1999).

Multimodality and Security

Computer security and usability often find
themselves in conflict. For example, computer
security suggests that user passwords should
contain digits, non-alphanumeric characters, be

108

more than six characters long, and not listed in

a standard dictionary. Usability findings suggest
that passwords that conform to these rules will
be difficult to remember, causing users to compromise security by writing the passwords down
for future reference. New modalities, in the form
of biometrics and other techniques, may provide
a solution to this problem.
Biometric identification systems include
speaker identification, fingerprint recognition,
and retinal scanning. These systems rely on obtaining a unique identifier for a user that is based
on who the user is. In addition to reducing the
need for users to remember an arcane security
code, biometric identification can make a system
easier to use by combining the identification task
with other interaction tasks. For example, voice
identification could theoretically be done while
the user is giving a voice interaction command.
The user experience is that the extra identification
step is being skipped, when in fact it is being
processed in the background.
Because current biometric identification systems are not 100% accurate, identifiers based on
what the user knows (similar to passwords) will
likely continue to be part of computer security.
Multimodal techniques in this category include
picture recognition tasks, where users must
identify a familiar picture from among a set of
unfamiliar ones (Weinshall & Kirkpatrick, 2004).
Other applications are beginning to emerge within
the picture recognition space, such as the SiteKey
feature of the Bank of America website,b where
users select a picture to be displayed after they
have logged in to online banking. The display of
the user-selected picture reassures users that the
pages and information they are viewing have indeed come from the bank. Interestingly, decadesold research (Bakst, 1988) suggests that the best
computer security will require a combination of
who the user is, what the user knows, and something the user carries (for example, a keycard).
By definition, a system that includes all of these
would have to be multimodal.

Multimodal and Federated Interaction

range of users, this is an example of universal

design.d Although there is some argument about
whether it is in fact possible to create a design
that is universally usable (Mace, 1998), especially given the difference between input modes
and the kinds of information that they can most
easily express (Oviatt, 1999), universal design is
frequently applied to systems that are to be used
in a public context. Legislation regarding equal
accessibility to information (such as the Americans with Disabilities Act and Section 508 of the
Rehabilitation Act in the United States) encourage
the designers of kiosks, automated teller machines
(ATMs), and other public computing sources to
take this approach.
Some applications are designed early on to be
accessible, but with a focus on a specific group of
users. For example, several commercial systems
and research projects [such as pwWebSpeake,
ASTeR (Raman, 1993), and AHA (James, 1998)]
have addressed the issue of giving blind users
better access to WWW-based content, taking
different approaches to presenting structured
documents in audio. One of the interesting things
about this area of research is that the underlying
WWW documents are not altered, but are instead
interpreted differently by the new application to
achieve accessibility.

Multimodality and Accessibility

New, often redundant, modalities can be added
to an application to simply provide an alternative
means to enter or receive information. One of
the major reasons for doing this is to support the
goal of accessibility, where computer systems are
made to be usable by people with disabilities. By
providing speech input capabilities, for example,
users with physical disabilities that affect their
ability to type or handle a pointing device can have
access to the same applications that non-disabled
users do. Similarly, audio (both speech and nonspeech) output can be provided in addition to a
visual screen to make an application or system
accessible to blind users.
Accessible applications can be created in
several ways, depending on when in the design
process the issue is considered and the function
of the application. Table 3 gives an overview of
approaches that can be taken to make an application accessible, and lists benefits and drawbacks
for each approach.
If accessibility is considered early in the
process and a design is created that incorporates
multiple input and output options to support a wide

Table 3. Approaches to accessibility

Approach Taken

Example

Benefits

Drawbacks

Create new app.

Kiosk using universal design

o
o

accessible to all user groups

avoid traps by considering
accessibility early

increase accessibility of
existing content
no changes required to
existing app.

Create app. targeted to a

group of disabled users

Audio-based Web browser for

blind/visually-impaired users

o
o

requires design planning for

all potential user groups
must be started at initial
design phase
new app. may not be
accessible to all
collaboration hard for
disabled/non-disabled users

Add new modality to

existing app.

Mac OS X VoiceOverc

increase accessibility of
existing content

requires original
manufacturer to decide to
make product change

Create helper app.

Screen readers, Voice-Enabled

Portal

original application
unchanged

need access to underlying

app.

109

Multimodal and Federated Interaction

Accessibility can also be added to products

later on, as in the case of voice-enabled portal.
The voice-enabled portal (VEP) [originally called
Voice over Workplace in James & Roelands
(2002)] is another example of a speech interface
that was developed as a redundant means of input
for an application, in this case a business portal.
The system was designed to support physically
disabled users by adding speech capabilities to
multiple applications within a portal, without
requiring those applications to be changed. To do
this, the VEP interface supported navigation and
basic text entry based on the applications syntax.
Users selected items by speaking their names or
by referring to elements by their type (e.g., text
box). A major problem for this system was in
disambiguating commands that could refer to
multiple targets: if there are four text boxes on
the screen, to which one does the command text
box refer? Because the main goal was to provide
access for physically disabled users, the solution
chosen uses voice input exclusively: VEP used
a predefined prioritization scheme to prioritize
targets based on their location on the screen,
choosing the target with the highest priority. If
more than one target had equally high priorities,
VEP numbered the targets using semi-transparent icons and required the user to specify the
desired one.
Screen readers are another type of application
where accessibility is added after an application
has already been designed and built. Screen
readers give blind users access to graphical interfaces using audio feedback and text-to-speech
systems to read back text on menus, within open
documents, and related to icons. These systems
typically use access hooks provided by the applications and the operating system to find text,
icons, and other graphical elements. For physically
disabled users, third-party speech recognition
systems (as well as those offered by the operating
systems vendors themselves) also use the hooks
to generate command grammars that allow users
to control applications as well as enter text.

110

Accessibility and design for users with disabilities may seem like a specialized area of
work, with a relatively small target population.
However, as pointed out in Perry et al. (1997)
and other sources, users can be handicapped, or
unable to accomplish a task, merely because of
the circumstances in which they try to perform
it. So-called temporary disabilities, such as the
inability of a user to direct his visual attention
to a computer monitor, can be mitigated through
the use of alternative modalities.

Multimodal Output and Transferring

Tasks between Modalities
Much of this section has focused on the use of
multimodal input in desktop settings. Multimodal
output can also be used in desktop computing,
especially for accessibility, as in the example of
screen readers for blind users. Another potential
use for multimodal output in a desktop setting is
related to computing tasks that may be continued
in a non-desktop setting, such as reading and
reviewing documents while mobile.
Imagine the case where a sales representative
is in her office reviewing her previous interactions
with a client whom she will be visiting later that
day. She reads through meeting minutes, e-mails,
and other text-based documents to refresh her
understanding of the situation. At the meeting
that afternoon, the client says something that reminds the sales representative of a note in one of
the documents she read, so she wants to go back
to the document to get more details. Because the
sales representative is out of her office, she is only
carrying a mobile device, which does not have
a large enough screen to display the document.
The device is, however, capable of presenting the
information via an audio interface. Will an audio
review of information that had been previously
only viewed visually be as helpful as a reminder
as would a visual glance?
The intriguing question rising from this and
similar scenarios is based on the concept of source

Multimodal and Federated Interaction

memory (Johnson et al., 1993), or the memory of

the qualities of the information presented (e.g.,
was it spoken or read, who said it, was it new
information, etc.). If information is presented in
one modality, but then must be retrieved using
a different modality, does this affect the users
success rate, efficiency, and other retrieval factors? If research shows the answer to be yes, it
may turn out to be beneficial to support additional
output modalities (e.g., speech and audio) in a
desktop computing environment. Information
can initially be presented in several modalities
simultaneously, so that later review or search in
any of the modalities is made easier.

Multimodality for Mobile

Devices
Increasingly, computer usage is moving from
desktop-based systems towards mobile devices,
including personal digital assistants (PDAs) and
mobile smart phones. New business and consumer
applications for mobile devices are being created
on an almost daily basis, to provide services and
information to mobile users. However, the design
of mobile applications presents many challenges
from the standpoint of usability.
Mobile devices are limited by their very nature
in terms of the user experience they can provide.
In order to remain portable, mobile devices are
constrained to a small form factor, which means
that their input and output devices are also small.
Tiny keypads and touch screens make input a
challenge, especially when the user is walking or
moving through the world. Small output screens
are difficult to read and cannot present much
information at one time. This section describes
how multimodal input and output support can
enhance the usability of mobile devices.

Input Modalities for Mobile Devices

Mobile devices come in several varieties, two
of the most common being PDAs and mobile
phones. The default input for these two device
groups differs: mobile phones typically use a
keypad (usually a 12-button telephone keypad,
with perhaps a few additional function buttons),
whereas PDAs (especially consumer-based models) as well as certain smart phones (for example,
the Palm-based Handspring series) typically rely
on handwriting input on a pressure-sensitive
screen or small keyboards. It is interesting to note
that although phones were originally designed to
facilitate voice communication, interaction with
mobile phone devices for tasks other than talking
to another person (e.g., messaging, changing settings, and dialing) still rely heavily on non-speech
input and output.
The small keypads used by most mobile phones
make it difficult for the user to efficiently enter
text and perform other input functions. Telephone keypad mappings requiring multi-tap or
predictive text entry can be error-prone and add
additional steps to the user interaction. QWERTYf
keyboards on mobile devices are so small that
the buttons can be difficult to accurately select.
Initial attempts to mitigate these problems were
taken by the device manufacturers themselves:
PDA designers offered external keyboards that
could be attached to the devices. The problem
with this solution, as well as related research into
keyboards that can be projected on a tabletop, is
that keyboards are difficult (if not impossible) to
use while standing, walking, or moving. If the
user is forced to sit and remain stationary to use
the keyboard, it defeats much of the purpose of
the mobile device.
Keyboard alternatives have been proposed for
mobile and wearable device input. These include
a variety of gesture-sensing devices, such as
the FingeRing device (Fukumoto & Tonomura,
1997), which senses chord-style typing using a
set of wearable rings. Chord typing is based on

111

Multimodal and Federated Interaction

particular combinations of fingers tapped on a

surface simultaneously; for example, tapping a
combination (or chord) of index, middle, and
ring finger describes a single letter of input. FingeRing allows users to tap chords on any surface,
including a leg or arm, making it more useful
while walking or moving. Other chord keyboards,
such as the Twiddlerg, are small hand-held devices
with strategically placed keys that can be selected
in groups to form the chords. The limitations of
chording, though, are a small input alphabet (for
FingeRing: five fingers, with two positions each
per chord, yields only 32 different chords) and a
fairly high cost of learning effort.
On the PDA side, handwriting recognition via
stylus has been available since the early days of
the Apple Newton. Unfortunately, problems with
the quality of the recognition using this system
were at least partially responsible for the devices
demise. More recently, the Graffiti system offered
with Palm devices has seen better success, with
its stroke-based (rather than word-based) recognition. Although Graffiti (or, rather, the underlying
Unistrokes alphabet) was specifically designed to
minimize recognition errors, Shneiderman (2002)
also attributes the success of Graffiti to the ability
of users to more easily see the locality and cause
of errors than with word-based recognition.
Stylus input, then, seems to be a fairly good
input modality for small devices. However,
styli, as well as the small keyboards found on
Blackberry-style devices, face limitations in
many mobile settings. These include situations
where the users hands are busy or encumbered
by gloves, or where the users eyes are busy. Both
the stroke-based recognition used by Graffiti
and small keyboards can be a bit slow and rather
cumbersome for entering large amounts of text.
Other modalities, such as voice, can alleviate
some of these problems.
Voice input is a common and well-accepted
alternative to stylus or keypad input for mobile
devices. (A Conversation with Jordan Cohen,
2006) Mobile phones are especially well adapted

112

to voice input, given that they already contain a

microphone to support the telephony functions
of the device. Voice input can free up the users
hands to perform the other functions necessary
when mobile, such as carrying packages, opening doors, steering a car, or wearing gloves. Of
course, voice input has its limitations in a mobile
context, as well. In extremely noisy environments,
recognition quality can deteriorate to the point
that voice is unusable. Quieter environments
may also be problematic for voice input, when
it is inappropriate to talk or when it would be
undesirable to be overheard. Technologically
speaking, speech recognition via mobile phone
often has a higher error rate than desktop-based
recognition, both due to potential background
noise and the quality of the devices microphone.
And, limitations in device memory can force
speech recognition grammars to be quite small,
which impacts usability.
Another way to improve user input to a mobile
device is to eliminate or reduce the need for the
input. With new capabilities to determine the location of mobile phones, such as global positioning
systems (GPS) and cell towers, mobile devices
can automatically gather information that would
otherwise require human input. For example, if
a user conducts a search for restaurants, he is
likely to want to find a restaurant close by. With
an application that is not context-aware, the search
query needed for this task will be something
like restaurant Palo Alto. If the device is able
to detect the users location, the query from the
user would only have to be restaurant, and the
application could use the devices location to
filter the results.
Context-awareness can be used in even more
clever ways, as well. The MICA (Multimodal
Interaction in Context-Adaptive Systems) project
used context awareness to sense actions by the
user that could be used as inputs (Lorenz et al.,
2005). The warehouse-based prototype designed
in the MICA project can sense when the user was
passing an item that needed to be picked from a

Multimodal and Federated Interaction

shelf and placed on the users trolley, and indicate

this to the user. In addition, the system was able
to recognize when the user was looking for, but
unable to find, an object on a shelf and provide
guidance to the exact location.
Finally, mobile device input can be augmented
using technologies that sense movements (e.g.,
tilting or lateral motion) of the device itself
(Hinckley et al., 2005). For example, a phone can
detect when it has been picked up and placed by
the users ear to automatically activate the speaker.
Or, a device can detect that it has been tilted, to
scroll down on a menu. Such interactions can be
achieved by augmenting the device with rotational
and gyroscopic sensors, or by new techniques
that make use of mobile phone cameras (Wang
& Canny, 2006).

Output Modalities for Mobile Devices

The small screens available on mobile devices are
not really adequate to display all of the complex
information that is requested by mobile users, such
as maps, directions, and documents. The most
common way of addressing this is through the
use of audio output to augment the small-screen
display. For example, many navigation systems
show a small portion of a map view on the phone
screen, with turn-by-turn directions presented
using voice. This type of interaction also benefits
mobile users by allowing them to focus more of
their visual attention on locomotion tasks, such
as walking and driving, which are common when
using mobile devices. The next section goes into
more detail on the use of audio output in divided
attention scenarios.
Mobile devices can also make use of haptic
feedback as an output modality. The canonical
example here is a vibration alert on a mobile
phone. These alerts do not require either visual
or auditory attention, making them useful in very
noisy or very quiet environments while the user
is mobile. Haptic feedback can be used in mobile
applications, such as navigation, to support the

movement-based input described above (e.g.,

tilting and rotation). Haptic output can provide
the user with more feedback and control during
tilting input, for example, by providing some level
of resistance to help the users actions stay in sync
with the action taken on the visual display (Oakley
& OModhrain, 2005). More experimental uses
of haptics include the expression of presence by
a remote participant to enhance social interaction
[for example, Rovers & van Essen (2004) and Teh
et al. (2006), where hand-holding, petting, or hugging by the remote participant is expressed to the
user through vibration on the local device].

Designing for Divided Attention

As mentioned above, mobile devices are used in
many contexts where the user is unable to devote
his or her full attention to the mobile device interface. These contexts include device interaction
while walking or driving: situations where the
mobile device task is secondary, and the task of
locomotion or navigation through the world is
primary. The users attention must be divided
between the primary and secondary task, and
the mobile device interaction must be sensitive to
the fact that it is secondary. This means that the
interaction must be designed to be usable without
requiring undue amounts of attention, and must
be especially sensitive to the modalities used to
avoid overloading any of the sensory channels.
Multimodality can be used effectively in divided attention situations, most frequently by providing information input and output capabilities on
channels that are not as important to the primary
task. For example, visual attention requirements
can be reduced by providing redundant output
using the audio channel. This also reduces the
need for the user to be in close physical proximity
with the device, since audio can be heard over a
greater distance than small visual displays can
be used. Prototypes for warehouse picking have
been designed in this way: the MICA system, for
example, provides redundant visual and audio

113

Multimodal and Federated Interaction

information so that the worker can move away

from the visual display on the warehouse trolley to retrieve an item from the shelf, while still
receiving information about the item to retrieve
(Lorenz et al., 2005).
An important caveat, however, is that simply
changing modality does not always produce an
interaction that can be used effectively in dividedattention scenarios, especially vehicle-based
scenarios. A study on speech-based interaction
in a driving context showed that menu navigation
produced competition for attentional resources
with the spatial aspects of navigating the vehicle (Lee et al., 2001). The central processing
theory of attention suggests that this is because
central resources are being overloaded, even if
none of the individual modalities are (Moray,
1999). Therefore, divided-attention interactions
should also be careful about the sheer amount of
information that is to be presented, perhaps by
adapting to the current driving situation based
on information about traffic conditions, driver
stress, and so forth.
Finally, investigations of vehicle interfaces
have shown that the style of information presentation has a strong impact on driving ability. Lee

et
al. (1999)

found that drivers responded differently,

in terms of response time and action taken, depending on several factors. These factors included
not only modality, but also display location and
message style (e.g., whether the system advised of
a situation or gave explicit instructions for action).
Other studies have also shown that the phrasing
of messages, as well as the emotions expressed
by the voice used for output, have an impact on
driving performance and attention (Jonsson et
al., 2004; Jonsson et al., 2005). Voice gender
can also make a difference, especially for certain
tasks such as navigation and feedback on driving
performance (Nass & Braves, 2005).

Device Federation
An emerging area of research in human-computer
interaction involves the combination of small,
portable devices with ambient computing and
interaction resources in the users environment.
This concept, which is cropping up in industrial
and academic research projects at various locations, is an attempt to balance the dual problems
of portability and usability through a new model
for mobile interaction, which we will call here
device federation.
The idea of device federation is to augment
small, portable devices such as smart phones
(called personal devices) with ambient computing resources, such as large displays, printers,
computers, PDAs, and keyboards. The personal
device can be used to establish the users identity
(see the discussion above related to improving
security by combining something the user carries with what the user knows and who the user
is), run applications, or connect to back-end
databases and servers, while ambient resources
are leveraged to provide usable input and output.
Conversely, personal devices can be connected
to external sensors that have minimal or no user
interfaces of their own, allowing the user to view
or manipulate data that would otherwise be hidden
within the environment.
Device federation is a broad concept, which
covers a wide range of federation types. An obvious example is federating large displays with small
mobile devices. Other types include federating
sensors and other resources with limited human
interaction capabilities with mobile devices, federating portable user devices with audio output,
and federating portable input devices with ambient
computers to provide accessibility.

Ad Hoc Wireless Federation

Device federation promises the ability to augment
the large amount of computing resources found in
todays portable devices using the ambient devices

114

Multimodal and Federated Interaction

found nearly everywhere in the Western world.

More and more offices, coffee shops, airports, and
other locations contain large displays, speakers,
and keyboards that could be used to make up
for the limited input and output capabilities of
mobile devices. Within the context of business
usage, goals include allowing users to federate
keyboards or other input devices with a personal
mobile device to enhance input, or allowing voice
input through a mobile device to be recognized
and sent to a federated large-screen display.
It is already possible today to connect input and
output devices to mobile phones and PDAs. USB
and other wired connections provide a standard
interface for keyboards, mice, displays, and other
devices to connect to personal devices in a fairly
seamless manner. However, wired connections
require device proximity, which can be problematic for some use cases, such as connecting a
mobile phone to a large-screen display. It would
be preferable to be able to connect to the display
from anywhere in the room, without having to be
right next to it. In addition, some phones may not
have standard wired connection points and instead
use proprietary ports and connectors. Therefore,
device federation holds wireless connectivity as
one of its main goals.
The second main goal of device federation
is to allow ad hoc connectivity. Many common wireless standards, such as Bluetooth, use
handshaking and pairing protocols that are far
too cumbersome to be of value when a personal
device is to be connected to a particular ambient
resource only once or twice in its lifetime. These
protocols are also not very user-friendly, especially
for inexperienced users.
One concern regarding ad hoc wireless federation is, of course, security. Devices that can
easily connect and disconnect can also be easily
compromised. Unfriendly devices could be used to
snoop on data connections, or to grab computing
resources. However, the need for security must
be balanced with usability, so that the benefits of
federation are not compromised.

Input and Output Federation

One of the most basic scenarios for device federation is display federation. Mobile devices often
have screens that are too small to adequately
display information, especially if the information
involves complicated visualizations, graphics,
or large quantities of text. Output federation can
solve this problem by connecting with a nearby
display to present the information. Additionally,
input mechanisms to large screen displays have
been investigated by many research groups [see
Bezerianos & Balakrishnan (2005) for a recent
example]. Federation offers another alternative,
by allowing input through a mobile device to
control a large-screen display.
The implementation of display federation is
not as future reaching as it may seem. Intels
Ubiquity project is currently prototyping a device called the Personal Server, which could
be used for this purpose. The personal server is
a small lightweight computer with high-density
data storage capability.h The primary purpose of
the device is to store data and provide computing power that can be leveraged by other devices
(such as a laptop, desktop, or projection displays)
using a local wireless connection (Want et al.,
2002). In fact, some of the original incarnations
of the Personal Server have included no direct
input and output capabilities: the very use of this
device depends on display federation (as well
as federation to input devices). WinCuts from
Microsoft Research allows users to push shared
areas of live screen content to other machines,
which are controlled locally (Tan et al., 2004).
PdaReach by June Fabricsi displays the contents
of a PDA screen on a computer connected via the
PDA synchronization cable.
What remains to be investigated are usage
scenarios and design for display federation. For
example, should all content automatically be
routed to a federated large display, or should users
explicitly control the portions of the output to be

115

Multimodal and Federated Interaction

displayed? Such considerations likely depend on

the applications involved, the security or privacy
of the environment, and other factors.
The federation of input devices would allow users to federate the input devices of their
choice with other resources. Users could, for
example, federate a speech input device to a
computer, thereby allowing multimodal input
to the computers applications. This would not
only be beneficial to mobile users, but was also
proposed more than ten years ago as a solution to
the accessibility problem (Perry et al., 1997). The
basic idea was to provide users with a portable
device that exactly met their needs in terms of
input capabilities: low-mobility users could use
single-switch or voice entry, blind users could use
Braille entry devices, and so forth. This would be
an example of creating a helper application to
support disabled users (as discussed earlier in this
chapter): the new element here is that the helper
application would not need to reside on the same
computer as the original application.
The Apple iPod is also a good example of input
federation. The iPod is portable and allows selfcontained functionality, but its input and output
functionality increase when federated with other
devices such as computers and car stereos. Federation with car stereos is especially interesting
in regards to the distribution of I/O: some products federate the iPod to a car stereos speakers,
handling input via the iPod itself, while others
use the cars output and input. User satisfaction
with the different products can provide insight
into federation for other consumer or business
applications.
Because flexible connection and disconnection is a goal for device federation, it is possible
that users may want to federate (or de-federate)
within the course of performing a task. This again
raises the possibility of using redundant outputs
to facilitate task transfer, as described above.

116

Data Source Federation

Data sources, in the form of sensors, RFID tags,
and other smart items are becoming more
prevalent in todays environment. Sensors and
tags can be used to track objects and provide realtime information. Federated devices can provide
new opportunities to access these nearly invisible
information sources by federating them with
other devices (such as PDAs or laptop computers) that can display their state, or even modify
their behavior.
Federating with data sources can also be used
in ubiquitous computing environments to provide
support for context-aware applications. Knowing
what data sources are currently in proximity to the
users personal device can give a good sense of
location, and data such as noise level, temperature,
and other environmental factors can be used to
modify the interaction. For example, if the noise
level reported by nearby sensors is very high, a
mobile device can make the decision to present
information using the visual and haptic channels,
rather than using audio.
Research projects that investigate software
updates for embedded and networked devices
point the way for data source federation. The
OSGi Alliance is creating an open service delivery and management platform so that software
components can be installed, updated, or removed
on the fly without having to disrupt the operation
of the device.j Similarly, Sun Labs is researching
Small Programmable Object Technology, or
Sun SPOTk, which allows running Java programs
to be moved between sensors. The flexibility
inherent in these projects opens the possibility
of federating devices to access data from nearby
sensors or tags.
In data source federation, it is important to
investigate the process of connecting and disconnecting devices. This topic of course also arises
in input and output federation scenarios, but the
sheer volume of sensors and tags available within a
local area increases the magnitude of the problem.

Multimodal and Federated Interaction

If the system requires users to search for nearby

devices and explicitly grant pairing rights, selecting a few sensors (or a nearby display) from a list
of thousands of possible devices becomes overwhelming. On the other hand, totally automatic
pairing with thousands of nearby devices will
generally be undesirable, as the user is unlikely
to want to interact with all of them and may be
uncomfortable not knowing which devices are
connecting at any given time.

Social Impact of Voice

Technologies
A discussion of multimodality would be incomplete without paying some consideration to the
social impact of voice technologies. As mentioned
earlier in the chapter, speech is one of the most
common input technologies after keyboards and
pointing devices. However, communicating with
computer devices using speech raises many issues
regarding how people respond to technologies that
engage in this very human activity. The Media
Equation, as it was coined by Reeves & Nass,
can be summarized by the idea that people treat
interactions with computers and other media the
same way that they treat interactions in real life
(Reeves & Nass, 1996). Among other things, users
react to and expect the same social phenomena in
human-computer communication as in humanhuman communication, including following the
rules of politeness and interpersonal difference.
Nass & Braves ( 2005) extend these findings to
state that humans are Wired for Speech, and have
even stronger expectations about how systems
that can speak or listen should behave.
In this section, we will reflect on one fundamental issue related to multimodal interactions
that is implied by this work: the mismatch of modalities between input and output (James, 1998).
When a device (or desktop application, for that
matter) provides voice output, but does not allow
voice input, this puts the user in a subordinate

position to the computer, and may make the user

feel out of control. As Brenda Laurel puts it:
We...tend to expect symmetry between the input
and output modalities of a system; that is, we expect that they are operating in the same sensory
universe as the rest of the representation. If a
computer talks to us, we want to be able to talk
back, and vice versa.... In most systems, our side
of the I/O equation is severely impoverished. The
system can present images, sounds, movements,
words, and possibly even speech, but we must
act inside a straitjacket of menu items and mouse
clicks. No wonder we often feel that computers
are always in controlthe system is holding all
the cards! Working toward symmetry in input and
output channels in human-computer activities can
vastly improve our experience of engagement and
agency. (Laurel, 1993)
Control is not the only issue. Mismatching
modality can also be perceived as impoliteness.
The polite response to a letter is another letter;
answering a letter with a phone call is less polite.
Computer systems that require an input in a different modality than the modality used for output,
according to Reeves & Nass (1996), force the user
to make an impolite response.
Due to the factors discussed above, care should
be taken when adding a modality to make sure
that it is available both for input and output, if it
is technically possible to do so. There is, however,
every possibility that new generations of computer
users who have been exposed to interfaces with
modality mismatches since birth may not have the
same control and politeness responses [changes in
politeness behavior related to technology use have
been shown to some extent in relation to mobile
phones, for example (Ling, 2004)]. Although computer interfaces evolve at a much faster pace than
humans, people who grow up using multimodal
interfaces that may not conform to human-human
social interaction standards may adapt their reactions to the computers and be more accepting of
seemingly impolite behavior.
117

Multimodal and Federated Interaction

Summary
This chapter has described some of the basic
principles of multimodality. We began with a
description of some of the ways that the human
senses can be used to interact with a computer,
and discussed the idea that modalities can be used
redundantly, where more than one modality is used
to either present or gather the same information,
or complementarily, where information from
more than one modality must be combined to
produce the whole input or output message. Each
of these methods can be used for either desktop
or mobile interactions, although generally for
different reasons.
Multimodal interfaces to desktop computers
strive to provide natural mappings between the
information to be input or output and the modality
used. This more frequently leads to interfaces that
use complementary modalities, or multimodal fusion. Desktop applications also use multimodality
to support users with disabilities or to provide more
usable security. Another interesting research area
for multimodality is around the use of redundant
output modalities to support users who move from
desktop to mobile computers during the course
of an interaction.
Mobile applications use multimodality to
improve the input and output capabilities of the
devices. Because mobile devices must be small
to be portable, they often have very small visual
displays and keypads. Voice and audio are obvious
choices here, and have been used widely. Finally,
because mobile devices are often used in contexts
where the user is carrying out another (more
primary) task, such as walking or driving, it is
important to design interfaces that do not require
too much of the users attention or overload any
of the senses.
Next, we described an emerging area of human-computer interaction that seeks to combine
the portability of mobile devices with the interaction capabilities of larger ambient devices, called
device federation. Input and output federation can,

118

and have already begun to, be used to support

mobile users with both business and consumer
applications. Federation with data sources, such as
sensors and RFID tags, could also benefit business
users and support context-awareness.
The final section of this chapter discussed an
important consideration when using voice interactions, namely, the social implications. Based
on continuing research from sociology, it is clear
that people treat media in many of the same ways
that they treat other people, and have many of
the same expectations. This is especially true for
voice-based systems, so care must be taken when
using voice input and output.

future research directions

The importance of multimodal interaction design
will continue to increase as users move more and
more towards interaction with mobile devices,
and away from traditional desktop computing
environments. Increasing wireless connectivity
is shifting users toward an always-on, alwaysavailable mentality that will be supported by
newer and better mobile devices.
The widespread use of mobile devices will
require a focus on supporting work that is done
within the context of other activities. This work has
already begun to some extent, but there are open
questions around the types of tasks users choose
to engage in while busy doing other things and
how to manage interruptions within an interaction. For example, in the vehicle context, it will be
important to understand what kinds of tasks are
of interest to users, and then create interfaces that
will help users complete them safely and without
creating an undue mental burden. Even without
considering interruptions, it will be important to
investigate the confirmation of actions taken by
users whose attention is divided between interaction with a mobile device and interaction with the
rest of the world.

Multimodal and Federated Interaction

The federated device concept discussed in

this chapter certainly falls under the area of future research. Initial projects have shown that it
should be possible to (wirelessly and seamlessly)
interconnect mobile and ambient devices; what
remains to be seen is how the interactions with
device federations should be structured. What
information should be shown on a personal
display, as opposed to a public display? Should
the personal device be used for input to a device
federation, and if so, how should it operate? Will
the personal device act as a remote control, or
will users respond to it simply as one component
of the larger system? These and other questions
are important to understand as the world moves
toward ubiquitous computing environments.
Finally, this chapter listed a wide range of
modalities available for human-computer interactions. Some of these modalities are already part
of many computer systems, such as voice and
audio, while others are less widely used. Haptics
and other modalities could be valuable additions
to new interaction designs, but more research is
required to understand where they would be most
appropriate for everyday use.

References
Bakst, S. (1988). The future in security methods.
The Office, 108(19-20).
Bezerianos, A., & Balakrishnan, R. (2005).
The vacuum: Facilitating the Manipulation of
Distant Objects In Proceedings of the SIGCHI
Conference on Human Factors in Computing
Systems, Portland, OR (pp. 361-370). New York:
ACM Press.
Bolt, R.A. (1980). Put-that-there: Voice and
gesture at the graphics interface. In Proceedings
of the Seventh Annual Conference on Computer
Graphics and Interactive Techniques (pp. 262270). New York: ACM Press.

Cheyer, A., & Julia, L. (1998). Multimodal maps:

An agent-based approach. In H. Bunt, R.-J. Beun,
& T. Borghuis (Eds.), Multimodal human-computer communication (pp. 111-121). Springer.
A conversation with Jordan Cohen. (2006, July/
August). ACM Queue, pp. 14-23.
Fukumoto, M., & Tonomura, Y. (1997). Body
coupled FingeRing: Wireless wearable keyboard.
In Proceedings of the SIGCHI Conference on
Human Factors in Computing Systems, Atlanta,
GA (pp. 147-154). New York: ACM Press.
Ghias, A., Logan, J., Chamberlin, D., & Smith,
B.C. (1995). Query by Humming: Musical Information Retrieval in an Audio Database.In
Proceedings of the Third ACM International
Conference on Multimedia, San Francisco, CA
(pp. 231-236). New York: ACM Press.
Grasso, M.A., Ebert, D.S., & Finin, T.W. (1998).
The integrality of speech in multimodal interfaces. ACM Transactions on Computer-Human
Interaction, 5(4), 303-325.
Hinckley, K., Pierce, J., Horvitz, E., & Sinclair,
M. (2005). Foreground and background interaction with sensor-enhanced mobile devices. ACM
Transactions on Computer-Human Interaction,
12(1), 31-52.
James, F. (1998). Representing structured information in audio interfaces: A framework for
selecting audio marking techniques to represent
document structure. Unpublished Ph.D. Dissertation, Stanford University, Stanford, CA.
James, F., & Roelands, J. (2002). Voice over
Workplace (Vowp): Voice Navigation in a Complex Business Gui. In V.L. Hanson & J.A. Jacko
(Eds.), Proceedings of the ACM Conference on Assistive Technologies, ASSETS 2002, Edinburgh,
Scotland (pp. 197-204).
Johnson, M.K., Hashtroudi, S., & Lindsay, D.S.
(1993). Source

monitoring. Psychological Bulletin, 114(1), 3-28.

119

Multimodal and Federated Interaction

Jonsson, I.-M., Nass, C., Endo, J., Reaves, B.,

Harris, H., Ta, J.L., et al.
(2004). Dont blame me,
I am only the driver: Impact of blame attribution
on attitudes and attention to driving task. In CHI
04 Extended Abstracts on Human Factors in
Computing Systems, Vienna, Austria (pp. 1219122). New York: ACM Press.
Jonsson, I.-M., Zajicek, M., Harris, H., & Nass,
C. (2005). Thank You, I Did Not See That: In-Car
Speech Based Information Systems for Older
Adults. In CHI 05 Extended Abstracts on Human
Factors in Computing Systems, Portland, OR (pp.
1953-1956). New York: ACM Press.
Koons, D.B., Sparrell, C.J., & Thorisson, K.R.
(1993). Integrating simultaneous input from
speech, gaze, and hand gestures. In M. Maybury
(Ed.), Intelligent multimedia interfaces (pp. 257276). Menlo Park, CA: MIT Press.
Laurel, B. (1993). Computers as theatre (2nd
ed.). Reading, MA: Addison-Wesley Publishing
Company.
Lee, J.D., Caven, B., Haake, S., & Brown, T.L.
(2001). Speech-based interaction with in-vehicle
computers: The effect of speech-based e-mail on
drivers attention to the roadway. Human Factors,
43, 631-640.
Lee, J.D., Gore, B.F., & Campbell, J.L. (1999).
Display alternatives for in-vehicle warning and
sign information: Message style, location, and
modality. Transportation Human Factors, 1(4),
347-375.
Ling, R. (2004). The mobile connection: The cell
phones impact on society. San Francisco, CA:
Morgan Kaufmann.
Lodding, K.N. (1982). Iconics: A visual manmachine interface. In Proceedings of the Third
Annual Conference and Exhibition of the National
Computer Graphics Association, Inc. (Vol. 1, pp.
221-233).
Lorenz, A., Zimmermann, A., & Eisenhauer, M. (2005). Towards natural interaction by

120

approaching objects. Paper presented at ABIS

2005, Saarbrcken.
Mace, R. L. (1998). A perspective on universal
design. Paper presented at Designing for the
21st Century: An International Conference on
Universal Design.
Maynes-Aminzade, D. (2005). Edible bits: Seamless interfaces between people, data and food.
Paper presented at Conference on Human Factors
in Computing Systems, Portland, OR.
Moore, R.C., Dowding, J., Bratt, H., Gawron, J.M.,
Gorfu, Y., & Cheyer, A. (1997). Commandtalk:
A Spoken-Language Interface for Battlefield
Simulations. In Proceedings of the Fifth Conference on Applied Natural Language Processing
(pp. 1-7). San Francisco: Morgan Kaufmann
Publishers Inc.
Moray, N. (1999). Commentary on Goodman,
Tijerina, Bents, and Wierwille, Using cellular
telephones in vehicles: Safe or unsafe. Transportation Human Factors, 1(1), 43-46.
Nass, C., & Braves, S. (2005). Wired for speech:
How voice activates and advances the humanCambridge, MA: MIT
computer relationship.
Press.
Oakley, I., & OModhrain, S. (2005). Tilt to
scroll: Evaluating a motion based vibrotactile
mobile interface. In Proceedings of the First
Joint Eurohaptics Conference and Symposium
on Haptic Interfaces for Virtual Environment and
Teleoperator Systems (pp. 40-49). IEEE.
Oviatt, S.L. (1999). Ten

myths of multimodal
interaction. Communications of the ACM, 42(11),
74-81.
Perry, J., Macken, E., Scott, N., & McKinley, J.L.
(1997). Disability, inability and cyberspace. In B.
Friedman (Ed.), Human values and the design of
computer technology (Number 72 Ed., pp. 65-89).
Stanford, CA: CSLI Publications.

Multimodal and Federated Interaction

Raman, T.V. (1993). Audio system for technical

readings. Unpublished Ph.D. Thesis, Cornell
University, Ithaca, NY.
Reeves, B., & Nass, C. (1996). The media equation: How people treat computers, television and
new media like real people and places. New York:
Cambridge University Press.
Rovers, A.F., & van Essen, H.A. (2004). HIM: A
framework for haptic instant messaging. In CHI
04 Extended Abstracts on Human Factors in
Computing Systems, Vienna, Austria (pp. 13131316). New York: ACM Press.
Shneiderman, B. (2002). Leonardos laptop: Human needs and the new computing technologies.
Cambridge, MA: MIT Press.
Tan, D.S., Meyers, B., & Czerwinski, M. (2004).
Wincuts: Manipulating Arbitrary Window Regions for More Effective Use of Screen Space.
In CHI 04 Extended Abstracts on Human Factors in Computing Systems, Vienna, Austria (pp.
1525-1528). New York: ACM Press.
Teh, K.S., Lee, S.P., & Cheok, A.D. (2006). Poultry.Internet: A Remote Human-Pet Interaction
System. In CHI 06 Extended Abstracts on Human
Factors in Computing Systems, Montreal, Quebec,
Canada (pp. 251-254). New York: ACM Press.
Tyfa, D.A., & Howes, M. (2000). Speech recognition for command entry in multimodal interaction. International Journal of Human-Computer
Studies, 52, 637-667.
Wang, J., & Canny, J. (2006). Tinymotion: Camera Phone Based Interaction Methods. In CHI
06 Extended Abstracts on Human Factors in
Computing Systems, Montreal, Quebec, Canada
(pp. 339-344). New York: ACM Press.
Want, R., Pering, T., Borriello, G., & Farkas, K.
(2002). Disappearing hardware. IEEE Pervasive
Computing, 1(1), 36-47.

Weinshall, D., & Kirkpatrick, S. (2004). Passwords Youll Never Forget, but Cant Recall.In
CHI 04 Extended Abstracts on Human Factors
in Computing Systems, Vienna, Austria (pp. 13991402). New York: ACM Press.

additional reading
Arons, B. (1991). Hyperspeech: Navigating in
Speech-Only Hypermedia. In Proceedings of
the Third Annual ACM Conference on Hypertext, San Antonio, TX (pp. 133-146). New York:
ACM Press.
Blattner, M.M. (1992). Metawidgets: Towards a
theory of multimodal interface design. In Proceedings of the Sixteenth Annual International
Computer Software and Applications Conference,
Chicago, IL (pp. 115-120). IEEE Press.
Bly, S. (1982). Presenting information in sound.
In Human factors in computer systems (pp. 371375). Gaithersburg, MD.
Buxton, W.A.S. (1994). The three mirrors of interaction: A holistic approach to user interfaces.
In L.W. MacDonald & J. Vince (Eds.), Interacting
with virtual environments. New York: Wiley.
Edwards, W.K., Mynatt, E.D., & Stockton, K.
(1994). Providing access to graphical user interfacesnot graphical screens. In Proceedings of
ASSETS 94: The First Annual ACM Conference
on Assistive Technologies, Marina del Rey, CA
(pp. 47-54). New York: ACM Press.
Flowers, J.H., Buhman, D.C., & Turnage, K.D.
(1996). Data sonification from the desktop: Should
sound be part of standard data analysis software?.
ACM Transactions on Applied Perception, 2(4),
467-472.
Gaver, W.W. (1990). The Sonicfinder: An interface
that uses auditory icons. In E.P. Glinert (Ed.),
Visual programming environments: Applications

121

Multimodal and Federated Interaction

and issues (pp. 561-581). Los Alamitos, CA: IEEE

Computer Society Press.
Gilman, A.S., Vanderheiden, G., & Zimmermann,
G. (2000). Universal design and the grid. Paper
presented at Grid Forum 5.
Gong, L. (2003). Multimodal interactions on mobile devices and users behavioral and attitudinal
preferences. In C. Stephanidis (Ed.), Universal
Access in HCI: Inclusive Design in the Information
Society (pp. 1402-1406). Mahwah, NJ: Lawrence
Erlbaum Associates.
Oviatt, S.L. (1996). User-centered

design of spoken language and multimodal interfaces. IEEE

Multimedia, 3(4), 26-35.
Smith, A., Dunaway, J., Demasco, P., & Peischl,
D. (1996). Multimodal Input for Computer Access
and Augmentative Communication. In Proceedings of the 2nd Annual ACM Conference on Assistive Technologies, Vancouver, BC (pp. 80-85).
New York: ACM Press.
Wise, G.B., & Glinert, E.P. (1995). Metawidgets
for multimodal applications. In Proceedings of
the RESNA 95 Conference, Vancouver, Canada
(pp. 455-457). Washington, DC: RESNA Press.

Endnotes

e
f

g
h

k

i
j

Deictic references can also include reference to

times, such as now or later and people (such
as you or me).
http://www.bankofamerica.com/privacy/index.
cfm?template=privacysecur_olb
ht t p://w w w.apple.com /macosx /feat u res/
voiceover/
See http://trace.wisc.edu/world/gen_ud.html for
a general discussion of the concept of universal
design.
http://www.soundlinks.com/pwgen.htm
QWERTY is the name for a standard U.S.
keyboard layout, and comes from the first six
letter keys in the top row starting from the left.
See Wikipedia for a more detailed explanation
(http://en.wikipedia.org/wiki/QWERTY ).
http://www.handykey.com/site/twiddler2.html
http://www.intel.com/research/exploratory/personal_server.htm
http://www.junefabrics.com/index.php
http://www.osgi.org/
http://research.sun.com/projects/dashboard.
php?id=145

This work was previously published in Handbook of Research on Ubiquitous Computing Technology for Real Time Enterprises,
edited by M. Mhlhuser and I. Gurevych, pp. 487-507, copyright 2008 by Information Science Reference, formerly known as
Idea Group Reference (an imprint of IGI Global).

122

123

Chapter 1.10

The Arab World, Culture and

Information Technology
Mohamed El Louadi
University of Tunis, Tunisia

Introduction
In 801, Harun Rashid offered Charlemagne a
water clock, the like of which was inexistent in all
of Europe at that time; the Kings court thought
that a little devil was hidden inside the clock. In
the 1930s, King Abdulaziz of Saudi Arabia had
to convince his people that the radio was not the
making of the devil and that it could in fact be used
to broadcast and spread the Quran. In 2003, the
Arab region is found to be still lagging in modern
technologies adoption (UNDP, 2003). Thus, in
a little more than 11 centuries, the Arabs were
transformed from leaders to adopters, then to late
adopters as far as technologies are concerned.
The Arab world is taken to mean the 22 members of the Arab League, accounting for more than
300 million people with an economy of 700 billion
dollars. Although most Arabs practice Islam, they
represent less than one third of all Muslims. The
Arab world is often thought of as economically
prosperous due to its oil resources; yet its total
GDP is lower than that of Spain (UNDP, 2003).

Arab countries share language and culture but

differ in size, geography, demographics, government type, etc. In terms of spending in and use of
IT, statistics portray the Arab world as far from
being up-to-date (UNDP, 2003).
This article raises the question of whether existing research models are appropriate to explain
this delay. If certain characteristics proper to the
Arab culture play an important role in explaining the delay in IT adoption, then we anticipate
the answer to be that no existing model is able
to adequately study the IT adoption phenomena
in the Arab world.

The Arab World and It

It is a recognized fact that the Arab world lags
behind in terms of IT (Aberdeen, 2001; UNDP,
2003). (See Figure 1.) In 2001, Ajeeb reported that
the number of Arab Web sites reached 1.24% of
the Arab population (Ajeeb, 2001), it predicted
that the number of Internet users would equal 5
million by 2001, and reiterated DITnets March

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

21

The Arab World, Culture and Information Technology

magne a water
n all of Europe
at a little devil
King Abdulaziz
e that the radio
it could in fact
n. In 2003, the
ng in modern
hus, in a little
e transformed
opters as far as

22 members of
an 300 million
n dollars. Alrepresent less
world is often
due to its oil
n that of Spain

lture but differ

ment type, etc.
atistics portray
o-date (UNDP,

hether existing
n this delay. If
culture play an
n IT adoption,
at no existing
adoption phe-

lags behind in
003). (See Figumber of Arab
ulation (Ajeeb,
Internet users
rated DITnets

Figure 1. The MENA region is under-performing in terms of IT spending even when compared
with other developing regions (Source: Aberdeen,
Figure
2001) 1. The MENA region is under-performing in
terms of IT spending even when compared with other
developing regions (Source: Aberdeen, 2001)

The Arab Woman

120000

90000

60000

30000

0
2001

Latin America

2002

2003

MENA

2004

2005

Sub-Saharan Africa

2000 prediction that Internet users would number

ten to twelve million by the end of 2002. In fact,
there were a reported 8.2 million users by the
March
prediction that Internet users would number
end of2000
2002.
ten to twelve million by the end of 2002. In fact, there were
To date, no Arab country has been able to
a reported 8.2 million users by the end of 2002.
develop
itsnoelectronic
commerce
capabilities
to
To date,
Arab country
has been able
to develop its
electronic
capabilities
the extentcommerce
seen in the
West. to the extent seen in
the West.

Reasons FOR
For THE
TheLAG
Lag
REASONS
The
be be
explained
by the
with which
Thelag
lagcan
canpartly
partly
explained
bydelay
the delay
with
1
technologies
have
traditionally
reached
Arab
countries
which technologies have traditionally reached.
Davison et al. (2000)
suggested several other reasons:
1
. Davisonbetween
et al. (2000)
suggested
countries
aArab
perceived
incompatibility
local cultures
and
several other
reasons:aapreference
perceivedfor
incompatibility
imported
technologies,
autonomy and
independence
respect
to imported
technology,
and a lack of
between localwith
cultures
and
technologies,
economic resources to acquire technology.
a preference
for autonomy and independence with
The first two of are plausible as is it often the case
respect
to
technology,
andnot
a lack
of of
economic
that IT stumbling blocks occur
because
technical
reasons
but rather
because
of human and social obstrucresources
to acquire
technology.
tions.
Thefirst
thirdtwo
reason
canplausible
be excluded
six Gulf
The
of are
as for
is ittheoften
the
countries which claim per capita revenues of nearly five
case that
IT stumbling
blocks
not because
times
the average
of the rest
of theoccur
Arab countries.
The
of
technical
reasons
but
rather
because
of human
rate of adoption of the Internet for these countries
is up

and social obstructions. The third reason can be

t or electronic forms without

written permission
IGI Gulf
is prohibited.
excluded
for theofsix
countries which claim per

capita revenues of nearly five times the average of

the rest of the Arab countries. The rate of adoption
of the Internet for these countries is up to fifteen
times that of the rest of the Arab world.

124

Other factors also explain the low rate of Internet penetration in Arab nations as compared
to the rest of the world. In these nations, the rate
of penetration is essentially measured based on
only one part of society: men.

Two thirds of the 65 million illiterate Arabs are

women. Women make up only 4% of all Arab
Internet users while in Europe women make
up 42% on average2. The UNDP states that the
condition of women is one of the three main
factors explaining the current state of economic
development in the Arab world.
In that more than 40% of women in Arab countries are denied participation to the technological
revolution, Arab nations are failing to integrate
a considerable part of their human resources in
their future economies.

Information and IT
When Arab countries invest in IT, they do so
mainly in hardware. While this may be a characteristic of developing countries, it may also be
viewed as Arabs distrust of anything immaterial.
Software on the other hand is associated with
innovation, creativity, and the free flow of information and knowledge, qualities that the Arabs
have been found lacking (UNDP, 2003). Thus, not
only Arabs are increasing their dependence to the
West being consumers of hardware, they seem to
be passive users of the software and intelligence
produced elsewhere.
This issue leads to the tight relationship between information (and not IT, let alone hardware)
and democracy and freedom. If Arab countries
are truly information shy (Henry, 1998), then
what information is to be shared and circulated
by IT? Therefore, the Arab does not see what
use he could make of IT and would therefore not
consider it an instrument of development.

Arabs areArabs
increasing
are increasing
their dependence
their dependence
to the West
to the
being
West being
pulling) pulling)
could becould
used to
be lower
used to
thelower
levelthe
of uncertainty
level of uncertainty
consumers
consumers
of hardware,
of hardware,
they seem
they
to be
seem
passive
to beusers
passive
of users
during
of business
during business
transactions.
transactions.
Even when
Evenmore
whendirect
more direct
the software
the software
and intelligence
and intelligence
producedproduced
elsewhere.
elsewhere.means are
means
available,
are available,
Arabs prefer
Arabswasta
prefer
because
wasta because
of the of the
This issue
This
leads
issueto leads
the tight
to the
relationship
tight relationship
between between
human contact
humanitcontact
offers. itWasta
offers.
is Wasta
a way of
is alife
way
that
of builds
life that builds
information
information
(and
not
(and
IT,
let
not
alone
IT,
let
hardware)
alone
hardware)
and
democand
democupon
human
upon
interactions,
human
interactions,
a
major
part
a
major
of
an
part
Arabs
of
an
life,
Arabs life,
The Arab World, Culture and Information Technology
racy andracy
freedom.
and freedom.
If Arab countries
If Arab countries
are trulyare
informatruly informawhich she
which
or he
shemay
or he
notmay
be willing
not be to
willing
sacrifice
to sacrifice
to
to
tion shytion
(Henry,
shy 1998),
(Henry,then
1998),
what
then
information
what information
is to be is totechnology.
be
technology.
shared and
shared
circulated
and circulated
by IT? Therefore,
by IT? Therefore,
the Arab the
does
Arab
not does not

Figure 2b. it will continue to represent more

Figure 2a. While hardware is projected to rep70%
allit
IT
investments
in the
resent
less
than 35%
all IT
the Figurethan
Figure 2a.
Figure
While
2a.hardware
While
hardware
is of
projected
is investments
projected
to represent
to inrepresent
2b.
Figure
itof
2b.
will
continue
will continue
to represent
to MENA
represent
moreregion
than
more than
(Source:
Aberdeen,
2001).
world
less thanless
35%
than
of all
35%ITofinvestments
all IT investments
in the world
in the world
70% of all
70%
IT of
investments
all IT investments
in the MENA
in theregion
MENA(Source:
region (Source:
Aberdeen,
Aberdeen,
2001). 2001).

100%

100%

100%

100%

80%

80%

80%

80%

60%

60%

60%

60%

40%

40%

40%

40%

20%

20%

20%

20%

0%

0%
2001

0%
2001
2002

2003
2002

Software+IT Services
Software+IT Services
Hardware

22

2004
2003

2005
2004

Hardware

22

Culture

Culture is [the] collective programming of the

mind that distinguishes the members of one group
of people from those of another (Hofstede, 1991).
For Ibn Khaldun, 14th century Arab sociologist,
man is son to his habits and his environment, not
to his nature and his moods3.
Hill et al. (1998) claim that the Arab culture is
one of the most complex of the world. It is complex
for several reasons. Though the majority of Arabs
are Muslim, many are Jewish or Christian while
others are agnostic and even atheists. Many actions
and symbols assume a particular significance.
Often in traditional Arab cultures a handshake or
the spoken word are seen as more legally binding
than a signed contract and the written word has
less significance than in other cultures4.
For Arabs bartering is very different from
what the West calls negotiating. The numerous
steps involved in bartering hold social meanings,
the least of which is to learn about and get more
acquainted with the other so as to build trust in
terms of the product quality. Selling and buying
are social acts not replicated by technology.
In his study of Qatari and Emirati cultures,
Solberg (2002) found that wasta (social connections and string-pulling) could be used to

2005

0%
2001

2001
2002

2003
2002

Software+IT Services
Software+IT Services
Hardware

2004
2003

2005
2004

2005

Hardware

lower the level of uncertainty during business

transactions. Even when more direct means are
available, Arabs prefer wasta because of the human contact it offers. Wasta is a way of life that
builds upon human interactions, a major part of
an Arabs life, which she or he may not be willing
to sacrifice to technology.
Arabs tend to be more talkative than Westerners; they are fond of sentences with several
adjectives, the number of which, combined with
the length of sentences and the tone of the voice,
make up an integral part of the message. For
example, when a guest is offered a cookie, it is
customary to not accept too keenly or quickly. It
is up to the host to insist (invoking God) to convert the initial refusal into a yes5. Arabs rarely
say yes or no, rather, they are inclined to answer
using less committed terms such as usually and
Inshallah. The design of computer interfaces
with only the yes, no and cancel options is
clearly rooted in other cultural norms.

Theories And Research

Models
Many research models are available to study IT
adoption. Flowers et al. (1975), for example, built

125

The Arab World, Culture and Information Technology

upon the framework of Graves (1970) in which

people would evolve through levels of psychological existence reflecting their personal values
and lifestyles, which vary with their cultural
programming.
Research using this framework suggests
that dominant value systems of Middle-Eastern
managers are conformist, sociocentric and existential and their dominant decision-making style
is consultative (Amirshahi, 1997).
Another important stream is represented by
Hofstedes work (1991) which considers culture a
national trait. The results of Hofstedes research
is a typology of cultures based on five cultural
dimensions: power distance, individualism, masculinity, uncertainty avoidance, and long-term
orientation6.
In research pertaining to IT, some dimensions
have been invoked more than others: individualism
(Jarvenpaa et al., 1999), power distance (Simon,
2001) and masculinity (Hofstede, 2001b). Findings from Hofstede (2001b) show a relationship
between masculinity and fax machines, and
femininity and cellular phones. Like authority, a
masculine trait, a fax is not interactive; and like
exchange and dialogue, which are feminine traits,
a cellular phone is interactive.
Certain Arab values could be mapped onto
Hofstedes cultural dimensions7. What one calls
maslaha el amma (public interest) happens
to mean the exact opposite of individualism. In
the Arab culture, public interest could even be
extended to the notion of Umma (nation). Family
commitments and responsibility towards authority figures and political leaders are often held as
priorities over other matters. Arabs from the Gulf
swear allegiance to the family, clan, tribe, religion,
and nation8 (Abbas, 1987). Usunier (1993) suggests
that Arab cultures tend to resemble cultures that
are focused on the present and for whom the future
(destiny) is inevitably predestined (Mektoub) by
God (Inshallah: God willing). For Patai (1973),
fatalism is a predominant trait of the Arab culture. If past-oriented cultures are more passive

126

than future-oriented cultures, the Arabs should

have an attitude that can be summarized as if
it is written then it will happen henceforth they
will not feel the urge to go toward the technology
and will expect it to come to them9.

Limitations of Existing Models

The Absence of Arab Dimensions
If in Hofstedes model, the first four dimensions
are applicable to other cultures, it would be useful to more closely study Arab values in order to
include them in existing models.
Cultures are likely to diverge rather than come
together as we would assume on the basis of
globalization and the democratization of media.
Hofstede (2001b) has shown that cultures are
diverging rather than converging. More specifically, current world events do not lend support
to the hypothesis that Arab and Western cultures
are closing in.
Current research models would therefore benefit from including Arab dimensions; otherwise
they would risk being applicable only to those
cultures for which they were developed or for the
Arab culture to continue being perceived through
the lens of another culture. This could be done
following the example of the Far East cultures.
These cultures were really only taken into account
into Hofstedes model with the addition of the
fifth dimension, Confucian dynamism, thereafter called long-term orientation. It was added in
order to complete Hofstedes model since it was
the dimension that distinguished western and
far-eastern cultures.

The Implicit Cultural Homogeneity

Differences exist among Arab countries (Solberg,
2002), and even between different regions of a
same country. This is even more noticeable as

The Arab World, Culture and Information Technology

the borders between Arab countries were often

drawn arbitrarily during colonization.
Fundamental cultural differences exist between Sunnis and Shiites, among them masculinity (Noland, 2003). Sunnis are considered more
triumphant (sic.) while Shiites focus more
on suffering. The two strains of Islam could
therefore be considered at the extremes of the
masculinity dimension although scores for this
dimension place the Arab world at the midpoint
of the scale. However, even Sunnis bear certain
feminine traits such as seeking good relationships
with leaders and cooperating with colleagues in
order to foster equally good relationships with
them (Weir, 2001).
Differences can also be found between the
cultures of the Gulf and other nations of the Arab
world. The discovery of oil in the 1930s resulted
in an odd mix of traditional and tribal cultures
with modern cultures. This mix did not manifest
itself in North Africa resulting in the Arab world
composed of countries generally having either
people or oil10. Furthermore, North African countries have not always had Arab identity. Unlike the
Gulf countries, they became Arabs at the same
time as Muslims.

The Cultural Barrier

Arabs, Arab culture, and Arab religion have
always been a black box for the Westerner. For
instance, some westerners have a near-Manichean
tendency to dichotomize (Good vs. Evil, Them
vs. Us, etc.) and are fond of clash-of-civilizations theories (see Huntington, 1996). El-Solh
and Mabro (1994, p. 14) show how the application of over-simplistic dichotomies to the study
of gender relations in the Middle East have left
out much of the flexibility women might have in
actual terms. The simple idea that the sexes have
a complementary role in Arab society has often
been misunderstood in Western studies as a binary
opposition with a tacit hierarchy.

Furthermore, if by fatalism it is understood that

humans have no say in their destiny, then the concept of time orientation of Patai (1973) originates
from a fundamentally inaccurate understanding
of a large part of the Arab world. Islam is founded
on the promise of salvation (Jenna) or the threat of
damnation (Jehenna). Debates on the kadha and
kadar11 (destiny and predestination) continue to
fuel the controversy that divides theologians on
the issue. Suffice it to mention that Arab societies do not necessarily neatly fit within cultures
anchored in either the past or the present.

Conclusion
Significant contradictions may exist between
how Arabs perceive themselves and how they are
perceived by others. For example, Hill et al. (1998)
argue that the adoption of technologies is rarely
in quest of imitating the West. But Ibn Khaldun
maintains that imitation is characteristic of Arabs. Abbas and Al-Shakti, (1985) even suggested
that Arab executives are the product of cultural
values that tend to produce more followers than
leaders. Yet, imitation is sometimes suspicious in
the Muslims eye as some believe that imitation
of the non-Muslims is haram (sinful) While it
is also believed that imitating non-Muslims is
permissible, the average Arab Muslim sometimes
wonders when imitation stops being haram and
starts being hallal (allowed)12. How to reconcile
these points of view?
Two theories seem promising in that they may
complement the research models we reviewed
here. Social Identity Theory recognizes that cultural layers exist that describe different levels of
programming (social, national, regional, religious,
contextual, organizational, etc.).
Abdul-Gader and Kozar (1995) borrowed
the construct of technological alienation from
psychosociology to explain certain purchase and
use decisions of IT. They showed that alienated

127

The Arab World, Culture and Information Technology

individuals resist any kind of technology adoption. More generally, Value-Expectancy Theory
(Feather, 1995) promises to enrich the debate on
IT adoption by Arabs since it addresses the issue
of the value attributed to things by individuals and
their expectations, founded or not, such as their
resistance to the possible danger of technological
and cultural dependence. This is all the more valid
that Arabs view IT as a technology, not as a medium of knowledge and of accessing knowledge,
something they need direly as evidenced by the
conclusions of the UNDP (2003).

References
Abbas, A. (1987). Value systems as predictors of
work satisfactions of Arab executives. International Journal of Manpower, 8(2), 3-7.
Abbas, A., & Al-Shakti, M. (1985). Managerial
value systems for working in Saudi Arabia: An
empirical investigation. Group & Organizational
Studies, 10(2), 135-152.
Abdalla, I.A., & Al-Homoud, M.A. (2001) Exploring the implicit leadership theory in the Arabian
gulf states. Applied Psychology An International
Review, 50(4), 506-531.

the managers in Iran. Unpublished PhD dissertation, Curtin University of Technology.

Davison, R., Vogel, D., Harris, R., & Jones, N.
(2000) Technology leapfrogging in developing
countries-An inevitable luxury? The Electronic
Journal on Information Systems in Developing
Countries, 1(5), 1-10.
El-Solh, C.F., & Mabro, J. (1994). Introduction:
Islam and Muslim women. In C.F. El-Solh & J.
Mabro (Eds.), Muslim womens choices. Religious
belief and social reality. Cross-cultural perspectives on women: Vol. 12. Oxford/Providence:
Berg.
Feather, N.T. (1995). Values, valences and choice:
The influence of value on the perceived attractiveness and choice of alternatives. Journal of Personality and Social Psychology, 68(6), 1135-1151.
Flowers, V.S., Myers, C.L., Myers, M.S., & Myers,
S.S. (1975). Managerial values for working (An
AMA Survey Report). New York: Amacom.
Graves, C.W. (1970). Levels of existence: An open
systems theory of values. Journal of Humanistic
Psychology, (2), 131-155.
Gupta, V., Hanges, P.J., & Dorfman, P. (2002).
Cultural clusters: Methodology and findings.
Journal of World Business, 37(1), 11-15.

Abdul-Gader, A.H., & Kozar, K.A. (1995) The

impact of computer alienation on information
technology investment decisions: An exploratory
cross-national analysis. MIS Quarterly, 19(4),
535-559.

Hill, C.E., Loch, K.D., Straub, D.W., & El-Sheshai,

K. (1998) A qualitative assessment of Arab culture
and information technology transfer. Journal of
Global Information Management, 6(3), 29-38.

Aberdeen Group. (2001, May). Worldwide IT

spending 2001-2005: Forecast and analysis for
hardware. Software and Service, 1-70.

Hofstede, G. (1991). Cultures and organizations,

software of the mind: Intercultural cooperation
and its importance for survival. McGraw-Hill.

Ajeeb. (2001, March 25). Over 3.5 million Arabs

accessing the Net. Retrieved from http://eit.ajeeb.
com/ViewArticle.asp?Article_ID=27643

Hofstede, G. (2001a). Cultures consequences:

Comparing values, behaviors, institutions and
organizations across nations (2nd ed.) Thousand
Oaks: Sage Publications.

Amirshahi, M. (1997). An empirical of managerial

value systems and decision-making styles among

128

The Arab World, Culture and Information Technology

Hofstede, G.J. (2001b). Adoption of communication technologies and national culture. Systmes
dInformation et Management, 6(3), 55-74.

UNDP. (2003, October 20) Arab Human Development Report 2003. United National Development
Program.

House, R., Javidan, M., Hanges, P., & Dorfman, P.

(2002). Understanding cultures and implicit leadership theories across the globe: an introduction
to project GLOBE. Journal of World Business,
37(1), 3-10.

Usunier, J-C. (1993) International marketing: A

cultural approach. Prentice Hall.

Huntington, S.P. (1996). The clash of civilizations

and the remaking of world order. New York:
Simon and Schuster.
Jarvenpaa, S.L., Tractinsky, N., Saarinen, L., &
Vitale, M. (1999). Consumer trust in an Internet
store: A cross-cultural validation. Journal of
Computer-Mediated Communications, 5(2). Retrieved from http://www.ascusc.org/jcmc/vol5/issue2/jarvenpaa.html
Kabasakal, H. & Bodur, M. (2002). Arabic cluster: A bridge between East and West. Journal of
World Business, 37, 40-54.
Khalid, A. (1994). Printing, publishing and reform
in tsarist central Asia. International Journal of
Middle East Studies, 26, 187-200.
Noland, M. (2003) Religion, culture, and economic performance, WP 03-8, The Institute for
International Economics. Retrieved from www.
iie.com/publications/wp/2003/03-8.pdf

Weir, D. (2002, May 8-11) Management in the

Arab world: A fourth paradigm? The European
Academy of Management (EURAM 2002), Stockholm, Sweden.

KEY Terms
Arab World: The Arab world is taken to
include all 22 countries members of the Arab
League.
Culture: According to hofstede (1991), it is
[the] collective programming of the mind that
distinguishes the members of one group of people
from those of another. For the 14th century arab
scholar, ibn khaldun, man is son to his habits
and his environment, not to his nature and his
moods. In all the literature about culture, there
is a common understanding that culture is an
abstraction from concrete behaviour but is not
behaviour itself. Hofstedes typology includes
five cultural dimensions:

Patai, R. (1973). The Arab mind. New York:

Scribners.
Simon, S.J. (2001). The impact of culture and
gender on Web sites: An empirical study. Database for Advances in Information Systems,
32(1), 18-37.
Solberg, C.A. (2002, December 5-7). Culture
and industrial buyer behavior: The Arab experience. Proceedings of the 18th IMP Conference
de lIMP, Dijon.

Individualism: as opposed to collectivism,

describes the degree to which a culture relies on and has allegiance to the self or the
group.
Power Distance: reflects the degree to which
a culture believes how institutional and organizational power should be distributed.
Masculinity: indicates the degree to which
a culture values such behaviours as assertiveness, achievement, acquisition of wealth,
social supports and the quality of life.
Uncertainty Avoidance: refers to the extent to which a culture feels threatened by

129

The Arab World, Culture and Information Technology

ambiguous, uncertain situations and tries

to avoid them.
Long-Term Orientation: reflects the degree to which a culture is oriented toward
the search for virtuous behaviour rather
than being oriented towards the search for
truth.

Information Technology: Hardware, software, network and services related to the use and
operation of equipment with the aim of processing and communication of analogue and digital
data, information, and knowledge. These include
computers and computer applications such as the
Internet, Intranets, Extranets, Electronic Data
Interchange, electronic commerce, mobile and
fixed lines, etc.

Endnotes
1

130

Most technological innovations are adopted

by the Arabs with some delay. For example,
if the printing press was invented around
1450, the first Arab book to have been printed
was the Quran, in 1537 in Venice. Arabic
character-based printing reached the Arabs
only in 1728 under the reign of Ahmed III
(1703-1730) when the Arabs were occupied
by the Ottoman Empire. Bukhara (todays
Uzbekistan) delayed the introduction of
printing until 1917 (See Khalid, 1994). It is
as though in the Muslim world, the advent of
printing broke the ulamas long-established
control of knowledge. Similarly to the Bible
and the printing press, no longer could
knowledge be obtained through someone
more knowledgeable. It could be read and
understood - or misunderstoodin libraries, schools, and even homes. Later, while
e-mail was invented in 1971, the first Arabic
Web-based mail solution was released in
December 1998 (see http://web1.maktoob.

com/maktoob/press1998/press1998-1.html).
See more on Kaitlin Duck Sherwoods site,
http://www.webfoot.com/advice/WrittenArabic.html.
Nielsen//NetRatings (May 2003), www.
journal dunet.com/cc/01_internautes/inter_profil_eu.shtml, accessed August 31,
2003.
Ibn Khaldun, The Muqaddimah, An Introduction to History, Translated from French
by F. Rosenthal, Princeton, 1958; 1967.
For Muslim Arabs, this may be explained
historically and religiously by the fact that
when the Divine message was delivered, the
Angel Gabriel dictated the Quranic verses to
the Prophet Muhammad. In all pre-modern
times, documents were not copied; they were
memorized, where there was no other way
to preserve them.
In his book, The Saddam Years (Fayard,
2003), Saman Abdul Majid, personal interpreter to the deposed dictator, explains
how, in 1993, President Clinton sent a secret
agent to Iraq to suggest that a new leaf be
turned over and that discussions be resumed.
Saddam did not immediately answer, an act
that Clinton took as a refusal. That file was
then closed. In fact, Saddam was expecting a
more solid and thought-out proposition to be
put forward, and was surprised that Clinton
did not come through with one. This miscommunication between two men of very
different cultures has had the now all-too
known consequences.
It is assumed that most readers are familiar with Hofstedes work. Due to space
limitations, details of his work will not be
elaborated here. For more information, the
reader is referred to Hofstede (2001b).
Also of interest is the GLOBE (Global Leader
and Organizational Behavior Effectiveness)
project which seeks to determine the relationship between leadership and societal
culture (House et al., 2002). GLOBE uses

The Arab World, Culture and Information Technology

nine dimensions, six of which are the same

as Hofstedes: uncertainty avoidance, power
distance, institutional collectivism, in-group
collectivism, gender egalitarianism, assertiveness, future orientation, performance
orientation, and humane orientation. Using
these dimensions, Gupta et al. (2002) identified 10 cultural clusters: Asia, Anglo, Arab,
Germanic Europe, Latin Europe, Eastern
Europe, Confucian Asia, Latin America,
Sub-Sahara Africa and Nordic Europe.
Unfortunately, often in GLOBE papers,
the Arab cluster is composed of very few
countries (see Abdalla and Al-Homoud,
2001). Other countries, such as Turkey,
have disconcertingly been found in such
Arab clusters (see Kabasakal & Bodur,
2002) ignoring the fact that Turkey is not a
member of the Arab League and overlooking
its claim to European Union membership.
Perhaps even in this order, especially if
one takes into account the famous proverb
in which an Arab states: Me, against my
brother; my brother and I against our cousin;
my brother, my cousin and I against all
others. Was it not written of Ishmael (and
by extension of all Arab children) in the
Bible that And he will be a wild man; his
hand will be against every man, and every
mans hand against him; and he shall dwell

in the presence of all his brethren (Genesis

16:12). This, however, does not seem to hold
any longer given the events that took place
between Iraq and Kuwait in 1990-1991 as
well as since then.
9
Judging by the lack of cultural sensitivity
of todays graphical user interfaces and
icons, that technology is all but coming to
the Arab culture.
10
The only exception being Iraq.
11
This highly philosophical issue revolves
around the overly simplified question that if
God had already decided on much of what
we humans will do on Earth even before
we were created, what is the Last Judgment
about?
12
One has to remember that the digging of
the trenches around Madina in the battle
of Al-Ahzab, which was led by the prophet
Mohamed (Peace Be Upon Him) and his
companions as a defense system, and the
application of Omar Ibn Al-Khattab, the
second khalif after the prophet, of the filing
system in the Islamic State, for organizing
taxes on cultivated land and payroll of the
soldiers are examples of imitation as they
were borrowed from Non-Muslim Persians
(see more on this subject on www.islam1.
org/khutub/Imitation_of_None-Muslims.
htm, accessed April 13, 2004).

This work was previously published in Encyclopedia of Developing Regional Communities with Information and Communication Technology, edited by S. Marshall, W. Taylor, and X. Yu, pp. 21-27, copyright 2006 by Information Science Reference,
formerly known as Idea Group Reference (an imprint of IGI Global).

131

132

Chapter 1.11

Information Technology
Acceptance across Cultures
Amel Ben Zakour
FSJEGJ, Tunisia

Abstract

Introduction

This chapter introduces national culture as a

possible factor accounting for the differences
in information technology adoption and use between countries. Based upon culture theory and
the technology acceptance model (TAM), the
author offers a conceptual model aiming at better understanding IT acceptance across countries
of different cultures. It has been argued that six
value dimensionsindividualism/collectivism,
power distance, uncertainty avoidance, masculinity/femininity, high/low context, and polychronism/monochronismact as moderators of
the TAM relationships. Furthermore, the author
aims at helping IT designers and IT managers all
over the world to understand why certain national
cultural values may be congruent or not with the
IT to be designed or implemented.

With the globalization context, companies all

over the world (developed and less developed
ones) are using information technologies (IT)
that have become more and more sophisticated
and complex. These technologies provide huge
opportunities to gain a competitive advantage
since information could be obtained, processed
and transmitted at very low costs (Porter & Millar,
1985). Nevertheless, in order for companies to take
full advantage of IT, they have to understand the
new challenge provided by these IT (Tapscott &
Caston, 1994). Actually, even though organizations adopt IT that best fit their business activities,
they cannot guarantee performance leveraging
unless the organization members appropriately
use it. According to Agarwal (1999), Acquiring
appropriate IT is necessary but not sufficient
condition for utilizing it effectively (p. 85). With
the globalization incurring unlimited interconnection possibilities and an increasing number of
partnerships, firms belonging to less-developed
countries are investing massively in new infor-

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Information Technology Acceptance across Cultures

mation technologies that are expected to improve

their competitiveness. Nevertheless, we notice
that although these technologies are voluntarily
purchased by organizations, they are not fully used
or accepted in the work context at the individual
level. Differences in IT usage and adoption are
reported in many descriptive studies pertaining
to IT implementation in different countries and
contexts (Danowitz, Nassef, & Goodman, 1995;
Goodman & Green, 1992). The main causes of
these differences that have been identified are
technical, economic, or managerial.
Our main concern in this chapter is to shed
light on an issue that deserves a closer study
which is the cultural factor. Indeed, the latter
is expected to be an interesting explanation for
the differences in IT adoption and use between
countries. Furthermore, culture has been used
as an explanatory factor of several managerial
and organizational issues (Fisher & Smith, 2003;
Thomas & Pekerti, 2003; Laurent, 1983; Hofstede,
1985; Silvester, 1990; Hernandez, 2000). The
literature in the information systems (IS) field
provides few studies attempting to explore the
nature of the relationship between culture and
IT implementation. Furthermore, most of the
prior cross-cultural studies on IS hardly have
focused on the individual behavior toward IT;
they generally have focused on IT transfer (Hill,
Loch, Straub, & El-Sheshai, 1998), on organizational characteristics related to IT implementation
(Robey & Rodriguez-Diaz, 1989), or on end-user
computing characteristics (Igbaria & Zviran,
1996). Moreover, these studies have focused
on the concept of culture at the macro level of
analysis (i.e., they attempted to compare IS-related behaviors in different countries, supposing
that each country is characterized by a different
set of cultural values). For example, when studying IT diffusion in Japan and the USA, Straub
(1994), using Hofstedes cultural classification
of 51 countries and three regions, supposed that
the Japanese are high in uncertainty avoidance
(ranked 7th) and the Americans are low in this

dimension (ranked 43rd). Even though there is accumulated evidence pointing to national culture
as a factor influencing IT adoption and acceptance, few studies have handled this issue at the
individual level. Therefore, the main objective of
this chapter is to provide a conceptual framework
that examines the influence of national culture,
though a macro-level construct, on IT adoption
by integrating specific cultural value dimensions
to technology acceptance model (TAM) (Davis,
Bagozzi, & Warshaw, 1989). This framework is
aimed at being used at the individual level since
cultural dimensions characterizing national
culture (i.e., individualism/collectivism, power
distance, masculinity/femininity, uncertainty
avoidance, high/low context of communication
and polychronism/monochronism) are value
dimensions considered as individual psychological dispositions and TAM is designed to capture
acceptance at the individual level.
We have divided the chapter into four parts. The
first part is devoted to a review of comparative and
international studies in the IS field. The second
part presents an overview on culture theory, which
is found in the literature of various disciplines
such as anthropology or cross-cultural psychology. The third part focuses on the presentation of
TAM (Davis et al., 1989), judged to be the most
parsimonious model of information technology
usage, as well as TAM-relevant extensions. The
fourth part is aimed at presenting the rationale
sustaining the conceptual framework developed
to explain how national culture could influence
IT usage.

Studies Involving National

Culture and Information
Systems
Prior researches focusing on cultural issues in the
IS field were conducted mainly through two main
and different perspectives. According to Ronen
and Shenkar (1988), the emic perspective studies

133

Information Technology Acceptance across Cultures

behavior from within a single culture-system

(p. 72), whereas the etic perspective focuses on
universals and studies behavior from a position outside the system and in comparison with
various cultures (p. 72). Early and Singh (1995)
have provided more comprehensive classifications
of international and intercultural studies. Early
and Singh (1995) have proposed four research
approaches to study these issues; namely, unitary
form, gestalt form, reduced form, and hybrid form.
The difference between these approaches stems
from examining whole systems rather than their
components (Early & Singh, 1995, p. 330).
The unitary form emphasizes the understanding of a particular cultural group using the
specific constructs and concepts of this group.
Actually, this kind of approach is not so much
interested in cultural dimensions per se as much
as it focuses on knowing how a certain behavior
is manifested in a particular setting. There is no
attempt to establish universals based upon such
a perspective.
The gestalt form, unlike the unitary system
form, focuses on the comparison of the phenomenon with a similar one in different contexts, and
constructs and relationships are derived from
general principles rather than from the systems
themselves (Early & Singh, 1995, p. 330). This
approach is appropriate when the researcher is

concerned with the generalizability of a model

across nations.
The reduced form examines the phenomenon
of interest with regard to a specific feature of
culture, such as the case dealing with one cultural
dimension when hypothesizing about a certain
behavior. Generally, comparative studies that
explain the differences in the applicability of a
model by the country of origin actually are using the reduced form. The cultural explanation
is provided ad hoc.
The hybrid form goes beyond comparison in
search of similarities and differences, and it allows researchers to explain how certain aspects
of a culture can explain the phenomenon. The
reliance on nation or culture as a black box is
abandoned in favor of a more precise specification of theoretical relationships (Early & Singh,
1995, p. 334).
In order to have a better visibility of crosscultural researches in the IS field, we present
in Table 1 a summary of cross-cultural studies
on IS, grouped according to Early and Singhs
classification.
Our current research follows more the hybrid
form since we integrate to TAM a set of universal cultural dimensions deemed to characterize
national culture.

Table 1. Overview of comparative and national studies on IS

Gestalt Form Studies
Authors

IS Research Area

Countries

Key Results

Igbaria
and Zviran
(1996)

End-user computing
(EUC)

USA
Israel Taiwan

EUC characteristics (individual and organizational),

perceived usefulness, effectiveness of EUC, satisfaction and
acceptance vary across countries.

Straub, Keil,
and Brenner
(1997)

IT individual
acceptance

USA
Japan
Switzerland

TAM explains the use of e-mail in the USA and Switzerland

but not in Japan.
The cultural Japanese orientation may limit e-mail use and
dissociate usefulness from use.

continued on following page

134

Information Technology Acceptance across Cultures

Table 1. Continued
Leidner and
Carlsson
(1998)

Executive Support
System (ESS)
adoption

Sweden
USA
Mexico

National culture influences the perception of the benefits of

ESS on decision-making processes.

Hill et al.
(1998)

IT transfer

Jordan
Egypt
Saudi Arabia
Lebanon
Sudan

Cultural factors identified in the Arab world such as family

and kinship obligations, communal world view, religion,
valuing the past or the concept of time, are powerful for
predicting the outcomes of technology transfer.

Unitary Form Studies

Martinsons and
Westwood
(1997)

Management
Information System
(MIS) adoption and
use

Gamble and
Gibson
(1999)

Decision making and Hong-Kong

use of information

The collectivism characterizing the Chinese (paternalistic

and autocratic) leads to the distortion or restriction of the
information produced by the financial information system.

Rowe and
Struck
(1999)

Media choice

The use of asynchronous media is more associated with a

cultural value oriented toward reactivity.

China

France

The cultural Chinese dimensions of paternalism,

personalism, and high context communications shape the use
of MIS in China.

The media that provide more redundancy in the feedback are

more associated with relation than with task-oriented values.

Reduced Form Studies

Straub
(1994)

IT diffusion

Japan
USA

Since the Japanese are characterized by a high level

of uncertainty avoidance, they view e-mail as more
information-rich and socially present than the Americans.
The structural features of the Japanese language make e-mail
difficult to use.

Rice, DAmbra,
and More
(1998)

Media choice and

assessment

Hong-Kong,
Singapore,
Australia,
USA

The collectivists prefer face-to-face interaction more than the

individualists do.
The individualists consider telephone and business memo richer
than the collectivists do.
There is no difference in the perception of equivocal
situations between individualistic people and collectivistic
ones.

Tan, Wei,
Watson,
Clapper, and
McLean
(1998)

Computer-mediated
communica-tion

Singapore
USA

The impact of computer-mediated communication on

majority influence depends on national culture, especially on
individualism/collectivism dimension.

Hasan and Ditsa

(1999)

IT adoption

West-Africa
The MiddleEast Australia

Differences were reported in IT adoption along eight cultural

dimensions, which are Hofstedes four cultural dimensions,
polychronism/monochronism, time orientation, context, and
polymorphic/monomorphic.

Hofstede
(2000)

IT adoption

56 countries

The higher a country scores on uncertainty avoidance, the

slower it will be in adopting a new technology.
IT adoption also is influenced by masculinity/femininity and
individualism/collectivism dimensions.

30 countries

Hofstedes four cultural dimensions have moderating effects

on TAM causal relationships.

Hybrid Form Studies

Srite and
Karahanna
(2006)

IT adoption

135

Information Technology Acceptance across Cultures

Culture Theory
Even though originally rooted in anthropology,
a population-level discipline, culture has been
defined and researched by many other disciplines
such as cross-cultural psychology. Culture has
been defined according to several perspectives.
Definitions go from the most complex and the most
comprehensive (Kluckhohn, 1962) to the simplest
(Hofstede, 1997; Triandis, 1972). According to
Kluckhohn (1962), Culture consists of patterns,
explicit and implicit, of and for behavior acquired
and transmitted by symbols, constituting the distinctive achievement of human groups, including
their embodiments in artifacts; the essential core
of culture consists of traditional (i.e., historically
derived and selected) ideas and especially their
attached values; culture systems may, in one hand,
be considered as product of action, on the other
hand, as conditioning influences upon further
action (p. 73). Hofstede (1997) defines culture as
the collective programming of the mind which
distinguishes the members of one group or category of people from another (p. 5).
Researchers in comparative and intercultural
management most of the time use the concept of
organizational culture or national culture. Nevertheless, they omit the fact that individual behaviors
and attitudes in an organizational context could
be influenced by other kinds of cultures. Indeed,
culture is a multi-level phenomenon that could be
approached according to different levels such as
region, ethnic group, religion, language, nation,
profession, firm, gender, social class (Hofstede,
1997; Karahanna, Evaristo, & Srite, 2005; Schneider & Barsoux, 2003). Culture also could
be defined according to continental or political
belonging (Lvi-Strauss, 1985). Furthermore,
these different cultures interact with each other.
For example, several ethnic groups are found in
India.
In the present study, we are interested in
national culture since it has been shown to influence management and organizations (Hernandez,

136

2000; Hofstede, 1985; Laurent, 1983; Silvester,

1990; Zghal, 1994). In a comparative study
involving the USA, Indonesia, Japan, and nine
countries from Western Europe, Laurent (1983)
has found that managers from these countries
have different behavioral patterns. For example,
the Americans believe that the main goal of a
hierarchical structure is to organize tasks and
facilitate problem resolution, whereas for the
Japanese, the Indonesians, or the Italians, the goal
is to highlight the authority structure. In the same
vein, Hofstede (1985) has demonstrated that the
employees perceptions of the organization and
its management depend on the intensity of two
national values: power distance and uncertainty
avoidance. Four implicit patterns of organization
conception are defined: (1) the pyramid pattern
(both power distance and uncertainty avoidance)
are high, which corresponds to the hierarchical bureaucracy, quite common in France and
the Mediterranean countries; (2) the well-oiled
machine pattern (low power distance and high
uncertainty avoidance), which corresponds to
an impersonal bureaucracy found, for example,
in Germany; (3) the village market pattern (both
power distance and uncertainty avoidance are
low), which is found in Great Britain and Nordic
countries; and (4) the family pattern (high power
distance and low uncertainty avoidance), which is
found for example in African and Asian countries.
Hernandez (2000) also has shown that African
management systems are influenced by a national
cultural trait: paternalism. The latter corresponds
to high power distance and relationships based
upon mutual trust. Paternalism also has been
found to be perceived positively in Tunisian
culture contrarily to American culture, which
is focused on contractual relationships or Dutch
culture, which is focused on consensual relationships (DIribarne, 1989). In a comparative study
between France and Germany, Silvester (1990)
has confirmed the results obtained by Hofstede
(1985), showing that the French organization could
be assimilated to a pyramid and the German one

Information Technology Acceptance across Cultures

to a well-oiled machine. For example, they found

that French managers have a prescriptive kind of
work, whereas German managers have cooperative kind of work through which employees could
learn and build a collective identity.
Several sets of dimensions have been developed to characterize the concept of national
culture. Table 2 provides an overview of the most
known cultural dimensions found in several fields
of studies.
The analysis of the culture frameworks in
Table 2 reveals an overlap of cultural dimensions
identified by the various authors. Even though
the definitions of these dimensions are not fully

convergent in all the cases, still they show some

similarities. Table 3 gives an overview of the
similarities that exist between these dimensions.
The current research will use the following dimensions: individualism/collectivism, masculinity/
femininity, uncertainty avoidance, power distance
(Hofstede, 1997), high context/low context (Hall,
1989), and time perception (Hall, 1989; Trompenaars & Hampden-Turner, 1998).
Several reasons lead us to investigate national
culture through these six dimensions.
First, these dimensions rely on variables that
are linked more directly to social and organizational process: they focus on human values rather

Table 2. Overview of the most known cultural dimensions

Cultural Dimensions

Authors

Power Distance

Hofstede (1997)

Individualism/Collectivism

Hofstede (1997)

Masculinity/Femininity

Hofstede (1997)

Uncertainty Avoidance

Hofstede (1997)

Long-term Orientation

Hofstede (1997)

Confucian Work Dynamism

Chinese Culture Connection (1987)

Conservatism

Schwartz (1994)

Intellectual autonomy

Schwartz (1994)

Affective autonomy

Schwartz (1994)

Hierarchy

Schwartz (1994)

Egalitarianism

Schwartz (1994)

Mastery

Schwartz (1994)

Harmony

Schwartz (1994)

Universalism/Particularism

Trompenaars and Hampden-Turner (1998)

Individualism/Communitarianism

Trompenaars and Hampden-Turner (1998)

Neutral/Emotional

Trompenaars and Hampden-Turner (1998)

Specific/Diffuse

Trompenaars and Hampden-Turner (1998)

Achievement/Ascription

Trompenaars and Hampden-Turner (1998)

Attitudes to time

Trompenaars and Hampden-Turner (1998)

Attitudes to environment

Trompenaars and Hampden-Turner (1998)

Communication context

Hall (1989); Hall and Hall (1987)

Perception of space

Hall & Hall (1987); Hall (1989)

Monochronic/polychronic time

Hall (1989)

continued on following page

137

Information Technology Acceptance across Cultures

Table 2. Continued
Nature of people

Kluckhohn and Strodtbeck (1961)

Persons relationship to nature

Kluckhohn and Strodtbeck (1961)

Persons relationship to other people

Kluckhohn and Strodtbeck (1961)

Primary mode of activity

Kluckhohn and Strodtbeck (1961)

Conception of space

Kluckhohn and Strodtbeck (1961)

Persons temporal orientation

Kluckhohn and Strodtbeck (1961)

than on general beliefs about the way we see the

world. Indeed, culture is primarily a manifestation of core values (Straub, Loch, Evaristo,
Karahanna, & Srite, 2002); therefore, in order to
better capture the relationship between culture
and behavior, Triandis suggests using values
(Triandis, 1972). Second, the first four dimensions
(Hofstedes cultural dimensions) constitute the
most used and most recognized dimensions as
a whole or separately in studying cross-cultural
issues in management and organizations. Because

of its global coverage in terms of respondents, it

seems that Hofstedes study has been unrivaled
(Smith & Bond, 1999). In fact, the identification
of the cultural dimensions was based upon a field
study covering a sample of 40 countries in which
more than 116,000 questionnaires were collected.
Hofstedes work also has been validated directly
or indirectly by many other researchers in different settings (The Chinese Culture Connection,
1987; Hofstede & Bond, 1984; Shackleton &
Ali, 1990). The studies that have been conducted

Table 3. Similarities in cultural dimensions

138

Hofstede (1997)

Schwartz
(1994)

Trompenaars&
Hampden-Turner
(1998)

Chinese
Culture
Connection
(1987)

Hall (1989)

Kluckhohn
and
Strodtbeck
(1961)

Power distance

Hierarchy /
Egalitarianism

Individualism/
Collectivism

Autonomy

Individualism/
Communitarianism

Relational
Orientation

Masculinity/
Femininity

Mastery/
Harmony

Achievement/
Ascription
Inner-directed/
Outer-directed

Man-Nature
Orientation

Uncertainty
Avoidance

Long-term
Orientation

Conservatism

Attitudes to Time

Confucian
Work
Dynamism

Time
Perception

Time
Orientation

Specific/
Diffuse

Space (personal
space and
territory)

Space
Orientation
(public,
private)

High /Low
Context

Information Technology Acceptance across Cultures

since Hofstedes work (The Chinese Culture Connection, 1987; Schwartz, 1994; Trompenaars &
Hampden-Turner, 1998) exploring the national
culture through values have sustained and amplified his findings rather than having contradicted
them (Smith & Bond, 1999). The two dimensions
related to time orientation and communication
context are based upon Halls well-established
studies on intercultural communications. Indeed,
cross-cultural studies of styles of communication,
briefly reviewed by Smith and Bond (1999), reveal
a divergence between societies in several aspects
of communication and provide evidence sustaining Halls contention about high/low context
(Gudykunst & Ting-Toomey, 1988).
In the following paragraphs, we are going to
give more explanations about the origins of these
cultural dimensions.

Power Distance (PDI)

According to Hofstede (1997), individuals in high
PDI societies accept the unequal distribution
of power, which is what Schwartz (1994) calls
hierarchy. The latter is linked to the inequality
of the distribution of power, roles, and resources.
Conversely, in low PDI societies, individuals do
not accept this inequality. Indeed, in an organizational context, subordinates can criticize their
superiors, and superiors are supposed to take into
account this criticism, since everyone is free to
express his or her disagreement about any issue.
In a low PDI organization, the hierarchical system
depicts an inequality of roles that may change over
time (Hofstede, 1997). This kind of behavior is not
tolerable in a society that emphasizes authority.
What Schwartz identifies as egalitarianism (also
egalitarian commitment) (Schwartz, 1994) can be
assimilated to the concept of low PDI advocated
by Hofstede.
Egalitarianism dimension is driven by values
expressing the transcendence of selfish interests
and leading to a voluntary commitment to promote
welfare of other individuals (e.g., social justice

or loyalty). Even though Schwartz explains the

opposition hierarchy/egalitarianism by psychological concepts (i.e., self-enhancement/selftranscendence), Hofstede explains PDI dimension, relying more on social contexts in which
individuals have learned to accept or reject the
inequality of relationships between people (such
as country history, ideologies, language). These
two dimensions are fairly comparable and valid
to characterize national culture.

Individualism/Collectivism (I/C)
Individualism concept takes its roots in the
Western world. Indeed, in Britain, the ideas of
Hobbes and Adam Smith about the primacy of
the self-interested individual sustain this concept.
In contrast, Confucianism in the Eastern world,
emphasizing virtue, loyalty, reciprocity in human
relations, righteousness, and filial piety, underlies
a collectivistic view of the world. Even though
some societies are characterized as individualistic
and others as collectivistic, they have to deal with
both individualistic and collectivistic orientations.
These orientations co-exist, and what makes the
difference among societies is the extent to which
they emphasize individualistic and collectivistic
values.
Individualism/collectivism has been criticized
by several disciplines and especially psychology
(for a review, see Kagitibasi, 1997). Nevertheless, it is still the focal dimension in cross-cultural studies and has been used most often as an
explanatory variable (Schwartz, 1994).
Hui and Triandis (1986) define collectivism
as a cluster of attitudes, beliefs, and behaviors
toward a wide variety of people (p. 240). Seven
aspects of collectivism has shown to be relevant
in characterizing it, which is the consideration of
the implications of our decisions for other people,
the sharing of material resources, the sharing of
nonmaterial resources (e.g., affection or fun), the
susceptibility to social influence, self-presentation and facework, the sharing of outcomes and

139

Information Technology Acceptance across Cultures

the feeling of involvement in others lives (Hui

& Triandis, 1986).
All the dimensions pertaining to this concept
of I/C (Hofstede, 1980), developed by Schwartz
(autonomy/conservatism), Trompenaars and
Hampden-Turner (individualism/communitarianism) and Kluckhohn and Strodtbeck (relational
orientation), deal with one unique issue: the
persons relationship to other people. Schwartz
defines autonomy/conservatism according to the
persons embeddedness in the group; meanwhile,
Hofstede, Trompenaars, and Hampden-Turner
define I/C, emphasizing individual vs. group
interest.
Two underlying ideas support the autonomy/
conservatism dimension. The first idea explains
the autonomy pole: an individual is considered
autonomous when he or she is holding independent
rights and desires and the relationship he or she
has with the others is dictated by self-interest and
negotiated agreements. The second idea explains
the conservatism pole: an individual is part of a
group and his or her significance pertains to his
or her participation in and identification with the
group (Schwartz, 1994). At the individual level,
Schwartz argues that there are two kinds of values: individualistic values, which are hedonism,
achievement, self-direction, social power, and
stimulation; and collectivistic values, which are
pro-social, restrictive conformity, security, and
tradition.
According to Hofstede, in individualist cultures, the interests of the individuals prevail
over the interests of the group (Hofstede, 1997,
p. 50). People are supposed to decide and achieve
alone. In collectivist cultures, people prefer to
decide and act in a group so that responsibility
and reward will be shared. Collectivism is associated with childrearing patterns that emphasize
conformity, obedience and reliability (Triandis,
1989, p. 16).
Individualism/Collectivism construct is
considered to be the most important aspect of
cultural differences because of its parsimony and

140

its potential to explain variations in economic

development (Kagitibasi, 1997). Indeed, according to Hofstede and Bond (1988), there is a high
correlation between individualism and the level
of economic growth at the national level.

Masculinity/Femininity (MAS)
The MAS dimension was derived empirically
from Hofstedes work on work-related values
across countries. Hofstede distinguishes between
cultures according to their emphasis on achievement or on interpersonal harmony. In labeling this
dimension according to gender, Hofstede refers to
the social and culturally determined roles associated with men vs. women and not to the biological
distinction. Several other studies have identified
almost the same dimension labeled as mastery/harmony (Schwartz, 1994) or achievement/ascription
(Trompenaars & Hampden-Turner, 1998). All the
dimensions show an obvious overlapping, since
they are driven by almost the same set of opposing
values. Empirically, a cross-cultural study conducted by Schwartz (1994) shows a positive correlation between MAS and mastery. In one pole,
the underlying values are assertiveness, material
success, control or mastery, and competition. This
is the pole of masculinity (mastery, achievement).
In the pole of femininity, the dominant values are
modesty, caring for others, warm relationships,
solidarity, and the quality of work life.

Uncertainty Avoidance
This dimension has been inferred from Hofstedes
survey pertaining to the theme of work stress
when he addressed his questionnaire to IBM
employees. According to Hofstede, the extent to
which individuals tend to avoid uncertainty can
differentiate among countries. Indeed, the feeling
of uncertainty is something that could be acquired
and learned in the diverse institutions of a society such as family, school, or state. Each society
will have its proper behavioral model toward this

Information Technology Acceptance across Cultures

feeling of uncertainty. Hofstede argues that uncertainty is strongly linked to anxiety. The latter
could be overcome through technology, laws, and
religion (Hofstede, 1997). In strong uncertainty
avoidance societies like South America, people
have rule-oriented behaviors. On the contrary, in
societies where uncertainty avoidance is weak,
people do not need formal rules to adopt a specific
behavior. In the same vein, Triandis (1989) makes
the difference between loose and tight cultures.
He sustains that loose cultures encourage freedom and deviation from norms, whereas in tight
cultures, norms are promoted, and deviation from
those norms is punished. Shuper, Sorrentino,
Otsubo, Hodson, and Walker (2004) also have
confirmed that countries do differ in uncertainty
orientation. Indeed, they have found that Canada
is an uncertainty-oriented society that copes with
uncertainty by attaining clarity and finding out
new information about the self and the environment, and Japan is a certainty-oriented society
that copes with uncertainty by maintaining clarity
and adhering to what is already known (Shuper
et al., 2004)

Time Orientation
The most overlapping dimensions are attitudes
to time (Trompenaars & Hampden-Turner, 1998),
time perception (Hall, 1989), and time orientation
(Kluckhohn & Strodtbeck, 1961), which highlight
two main dimensions describing the concept of
time: the structure of time (discrete vs. continuous) and the horizon of time reference (reference
to the past, present, or future). According to Hall
and Hall (1987), a monochronic person runs one
activity at a time and associates to each activity
a precise time, while a polychronic person can
perform many activities simultaneously without
preparing an exact schedule, or if it exists, it
can be ignored. Studies conducted in this sense
(Schramm-Nielsen, 2000) have shown that polychronic time is very specific to the Mediterranean

countries, the Middle East, the Arab world, and

Latin America. Indeed, in these countries, people
assume that time is elastic, and thus, the appointment time and planning are not always respected.
Actually, this behavior is considered very normal,
because individuals emphasize interpersonal
relationships. On the contrary, North-European
and Anglo-Saxon countries usually adopt the
monochronic time. In these countries, time is
divided into segments and rigorously scheduled so
that each activity will be assigned a specific and
unique time within which it will be performed.
Priorities also are classified according to time.
Consequently, planning and schedules are sacred and cannot be changed. Trompenaars and
Hampden-Turner (1998) called these two types of
time systems the sequential and the synchronic.
Time can be legitimately conceived of as a
line of sequential events passing us at regular
intervals. It can also be conceived as cyclical and
repetitive (Trompenaars & Hampden-Turner,
1998, p. 126).
The time assumption underlying the remaining dimensions (long-term orientation, Confucian
work dynamism, and conservatism) pertains to
the temporal horizons to which persons refer to in
their activities: past, present, and future. Confucian work dynamism is based upon values derived
from Confucian teachings. In one pole, called by
Hofstede long-term orientation, are found values
such as perseverance and thriftiness, and in the
short-term orientation pole are found values such
as personal steadiness and stability and respect
for tradition.
Schwartz identifies a value called conservatism
that is similar in many aspects to the Confucian
work dynamism, since it emphasizes the cohesiveness of the group and the respect for the traditional
order of the group. This emphasis is depicted in
at least three values characterizing Confucian
work dynamism: ordering relationships, personal
steadiness, and respect for tradition.

141

Information Technology Acceptance across Cultures

High Context/Low Context of

Communication
The nature of the communication context has
been highlighted only by Hall (1989) as a relevant
dimension capable of discriminating between
cultures. In order to better understand the concept
of high context/low context of communication, it
is crucial to define context and to shed light on the
relationship between meaning, context, and communication. According to Hall and Hall (1987),
context is the information that surrounds an
event and inextricably bound up with the meaning
of that event (p. 7). A distinction can be made
between internal and external contexting. The
former depends on past experience internalized
into a person or the structure of the nervous system
or both (Hall, 1989). The latter is based on the
situation and/or setting pertaining to a particular
event. Meaning is the interpretation attributed to
a particular message or communication. Words
and sentences have different meanings depending on the context in which they are embedded
(Hall, 1983, p. 56). According to Hall (1989),
meaning is made up of the message and the two
kinds of context, which are the background and
preprogrammed responses of the recipient on
the one hand and the situation on the other hand.
Since information, context, and meaning are the
key components in defining high and low context
cultures, Hall contends that the more information that is shared, the higher the context
(Hall, 1983, p. 56). Therefore, high-context communications are those that rely more on context
(shared or stored information) than on transmitted
information. In this case, little of the meaning
is conveyed in the transmitted part of the message. In high-context cultures, simple messages
with deep meaning flow freely (Hall, 1989, p.
39). People do not need to elaborate their speech
codes because of the largely shared assumptions
about the rules of communication. Indeed, the
emphasis is more on the manner of transmitting a
message than on the message per se. For example,

142

gestures and voice intonation can make the difference between messages of the same content. On
the contrary, low-context communications rely
more on the transmitted part of the message and
less on the context. Americans, Germans, Swiss,
and Scandinavians are found to be low-context.
In such cultures, characterized by Hall (1989) as
fragmented cultures, people have more elaborate
codes in communication because of their lack of
shared assumptions about communication rules.
All the meaning they try to transmit is reflected
in the information contained in the literal message. In this case, it seems obvious that there is no
need to use different kinds of signals as a guide
to interpretation.

Technology Acceptance
Model
The technology acceptance model (TAM) is considered a tool designed to understand and measure
the IT individual determinants of use. The most
comprehensive TAM is designed by Davis et al.
(1989). It is based on two key concepts: perceived
usefulness (PU), defined as the prospective
users subjective probability that using a specific
application system will increase his or her job
performance within an organizational context
(p. 985); and perceived ease of use (PEOU), defined as the degree to which the prospective user
expects the target system to be free of effort (p.
985). According to this model, an individual who
intends to use a system will have two types of
beliefs (PU and PEOU), which influence behavioral intention through attitudes. In addition, PU
will have a direct effect on behavioral intention.
Finally, external variables will have a direct effect
on the two beliefs.
In order to study individual adoption of IT,
the present research will be using the TAM.
The use of the latter is motivated by theoretical,
empirical, and practical considerations. From a
theoretical point of view, TAM takes its roots

Information Technology Acceptance across Cultures

from the theory of reasoned action (TRA) (Ajzen

& Fishbein, 1980). The latter provides a simple
model to understand and predict a persons behavior. It goes from the principle that an individual
behaves most of the time in a rational manner,
takes into account available information, and assesses consequences of his or her actions implicitly
or explicitly (Ajzen, 1988). According to TRA,
individual behavior is influenced by behavioral
intentions. These intentions at the same time are
determined by attitudes toward this behavior and
subjective norms. In turn, attitudes and subjective
norms are dependent on behavioral and normative
beliefs, respectively.
TRA is one of the most prominent theories
of social psychology and the most powerful to
predict a specific behavior (Agarwal, 1999; Liska,
1984). Two meta-analyses of empirical studies
testing the TRA conducted by Sheppard, Hartwick, and Warshaw (1988) show that the theory
has a powerful predictive utility. Moreover, it is
proved to be broadly applicable to a large array
of behaviors (Agarwal, 1999; Sheppard et al.,
1988). A study of a large variety of behaviors,
such as voting for a presidential candidate, having an abortion, or smoking marijuana, confirms
the causal relationships of the model, considering
attitudes and subjective norms as independent
variables and intention as the dependent variable
(Ajzen, 1988).
From an empirical point of view, several
researches applied TAM to different kinds of
IT, such as text processing application (Davis
et al., 1989), computers (Igbaria, 1995), e-mail
(Karahanna & Straub, 1999), the World Wide
Web (Moon & Kim, 2001), with different kinds
of end user populations, such as students (Davis
et al., 1989; Mathieson, 1991; Szajna, 1996), employees (Karahanna, Straub, & Chervany, 1999;
Lucas & Spitler, 2000), or even physicians (Hu,
Chau, Sheng, & Tam, 1999), and in different
organizational contexts. Other studies test TAM
and compare it to competing theories such as
TRA (Davis et al., 1989) or Theory of Planned

Behavior (Mathieson, 1991). All of these studies

show that TAM has a strong predictive power.
Moreover, TAM extensions (Agarwal & Karahanna, 2000; Agarwal & Prasad, 1999; Dishaw &
Strong, 1999; Lucas & Spitler, 1999; Venkatesh &
Davis, 2000; Venkatesh, Speir, & Morris, 2002),
the main objective of which is to extend TAM
by integrating other variables hypothesized to
influence directly TAM constructs or indirectly
by moderating the relationships among them,
support its causal relationships and confirm its
predictive utility. A nonexhaustive summary of
TAM extensions is presented in Table 4.
From a practical point of view, TAM gives the
possibility to practitioners, especially IS designers, to identify the areas in which we can act to
improve IT use. Indeed, the two key concepts of
TAM (i.e., perceived usefulness and perceived
ease of use) could be controlled by IT designers.
Consequently, developers could make it easy to
use IT and to identify more concretely managerial
benefits driven by IT that they are developing.

The Influence of National

Culture on IT Adoption:
Conceptual Model and
Research Propositions
We draw upon an extension of the TAM that integrates subjective norms (Venkatesh & Morris,
2000) as a determinant of the intention to use IT.
Subjective norms reflect a persons perception
of the social pressure to perform the behavior
in question. The importance attributed to social
norms in determining and predicting behavior
varies across cultures (Triandis, 1977); therefore,
we expect that integrating subjective norms will
strengthen our understanding of differences in
behavioral intentions and will allow a better
capturing of the cultural effect on IT use.
In order to explore the cultural factor, we rely
upon Hofstedes four cultural dimensions:

143

Information Technology Acceptance across Cultures

Table 4. Summary of TAM extensions

Authors
Taylor
and Todd (1995)

Igbaria (1995)

Jackson, Chow,
and Leitch (1997)

IT Tested

Variables Added to TAM

Research Methodology

Hardware and
software in a
computing resource
center

- Subjective norms
- Perceived behavioral control

Field study
Questionnaire addressed to 1,000
students from a business school

Computers

- Subjective Norms
- Normative beliefs and motivation to comply
with the referent group
- Computer anxiety
- Computer knowledge
- Direction support
- Info center support
- Organizational politics
- Organizational use of the system (colleagues,
subordinates, and CEOs)

Field study
Questionnaire addressed to
471 managers and professionals
working in 54 North-American firms

Different kinds of IS

- Situational involvement
- Intrinsic involvement
- Prior use

Filed study
Questionnaire addressed to 585
employees from organizations
developing or revising their IS
Field study
Questionnaire addressed to 100 users
working in a worldwide American
firm in the transportation sector

Karahanna and
Straub (1999)

Electronic Mail

-
-
-
-

Venkatesh (1999)

Virtual Workplace
System

- Game-based training

-
-
-
-

Social Influence
Social Presence
Perceived Accessibility
Availability of User Training and Support

Technology-Task Fit (TTF)

Task Characteristics
Tool Experience
Tool Functionality

Dishaw and
Strong (1999)

Maintenance support
software tools

Lucas and Spitler

(2000)

Broker workstations.

- Broker strategy (work approach)

- Perceived System Quality

Venkatesh and
Morris (2000)

System for data and

information retrieval

- Gender
- Subjective Norms
- Experience

Experiment with business

professionals: 69 in the first study and
146 in the second
Field study
Questionnaire addressed to
programmer analysts working in
three American firms, leaders in their
respective fields (financial services,
aerospace manufacturing, and
insurance)
Field study
Questionnaire addressed to 41 brokers
Longitudinal field study
Questionnaire addressed to 445
individuals from five organizations

World Wide Web

-
-
-
-

Personal Innovativeness
Playfulness
Self-efficacy
Cognitive Absorption

Field study
Questionnaire addressed to 250
students

Venkatesh and
Davis (2000)

- Scheduling
information system
- Windows-based
environment
- Windows-based
customer account
management
system

-
-
-
-
-
-
-

Subjective norms
Image
Job relevance
Output quality
Result demonstrability
Experience
Voluntariness

Three longitudinal field studies

Questionnaires addressed to 48
floor supervisors in a medium-sized
manufacturing firm in the first study,
to 50 members of the financial
department of a large financial firm in
the second study, and to 51 employees
from a small accounting services firm

Moon and Kim

(2001)

World Wide Web

- Perceived Playfulness

Agarwal and
Karahanna (2000)

144

Field study
Questionnaire addressed to 152
students

Information Technology Acceptance across Cultures

Individualism/collectivism: Defined as
the degree to which people in a society are
integrated into strong cohesive ingroups.
Masculinity/femininity: Defined as the
extent to which a society attributes qualities
such as assertiveness, material success to
men. and modesty and concern about the
quality of life to women.
Uncertainty avoidance: Defined as the
extent to which the members of a culture
feel threatened by uncertain and unknown
situations.
Power distance: Defined as the extent to
which the less powerful members of institutions and organization within a country
expect and accept that power is distributed
unequally.

Moreover, we rely upon Halls work on intercultural communication, which is based on two
key cultural dimensions:

High context/low context: Defined as the

extent to which people are aware of and pay
attention to all types of situational and contextual cues when interpreting messages.
Polychronism (vs. monochronism): Defined as the preference to perform many
activities at the same time.

These dimensions are hypothesized to have direct effects or moderator ones on TAM constructs
and relationships. Integrating cultural dimensions to TAM is an attempt to better understand
the genuine influence of national culture on IT
adoption at the individual level. Indeed, crosscultural studies on IS (see Table 1) focused on
IS phenomena at the organizational level, such
as IT transfer. Moreover, all the authors (except
Srite & Karahanna, 2006) do not conceptualize
national culture at the individual level. They
have just studied IT-related behaviors in different
countries, supposing that each country is characterized by a different set of cultural values. Their

findings can be criticized in a sense that cultural

values that characterize a country could not be
espoused by all the members of this country.
For example, Hofstede has found that Japan is a
collectivist country, but this does not mean that
each Japanese person will have the same level of
collectivism. Actually, Hofstede has studied workrelated values at the macro level by developing
national scores for each cultural value. Therefore,
it is worthwhile to analyze and measure culturally espoused values when dealing with national
culture at the individual level. In this study, since
we are attempting to deepen our understanding
of IT adoption at the individual level, we have
chosen to conceptualize cultural dimensions at
the individual level so that they can be integrated
in TAM. The conceptual model driven by this
integration is presented in Figure 1.
In the next paragraphs, we present the rationale
underlying model propositions.
Proposition 1. Individuals espousing high-context
values will have a less strong perception of IT usefulness than those holding low-context values.
Since high-context oriented individuals prefer
less coded messages, the electronic communication, implying very explicit content, will not
be very useful as a means of communication.
It cannot inform the receiver of the context of
the message. The receiver probably will seek to
know more about the message by other traditional means, implying more physical cues (face
expression, intonation of the voice). Martinsons
and Westwood (1997) have demonstrated that the
use of management information systems (MIS)
in Chinese culture depends on high-context
communications. Computers do not convey the
necessary richness of meaning in a high context
communication environment (Martinsons &
Westwood, 1997, p. 220) Indeed, since in China, a
high-context country, coding is very restricted and
so messages are terse in words, using computerbased MIS will not allow a full understanding of

145

Information Technology Acceptance across Cultures

Figure 1. Conceptual model

Masculinity/
Femininity

Power
Distance
High Context/
Low Context
Polychronism/
Monochronism

P3 (-)

P1 (-)
Perceived
Usefulness
P2 (-)

P4 (+)

System
Usage

Perceived
Ease of Use
Subjective
Norms

P5 (-)

Individualism/
Collectivism

P6 (-)

Masculinity/
Femininity

P7 (+)
P8 (+)

messages. Actually, in order to better interpret

the words, the Chinese rely upon other cues such
as tone, dynamics, facial expression, and body
language, which are not permitted when using
computer-based communication.
Proposition 2. Individuals adopting monochronic
time values will be using IT more than those
adopting polychronic time values.
Individuals adopting the monochronic time are
worried about the monitoring of time and prefer
doing one thing at a time; therefore, they can view
in IT a way to have control over time. Rowe and
Struck (1999) argued that people who are biased
towards action see in asynchronous media the
way to coordinate their tasks without interrupting
them (p. 166). In this case, we could expect that
e-mail or voicemail would be appreciated highly
by individuals espousing monochronic cultural
values. On the contrary, individuals holding more
fluid attitudes toward time are able to engage
in many things at once, pay more attention to
interpersonal relationships than to schedules,
and accept disturbances. Therefore, they will be
expected to view IT as less useful compared to
individuals holding monochronic values.

146

Power
Distance
Uncertainty
Avoidance

Proposition 3. Individuals in high power distance

cultures accept less IT than individuals in low
power distance cultures.
In high power distance cultures, IT could
threaten the hierarchy, which reflects the
existential inequality between higher-ups and
lower-downs (Hofstede, 1997, p. 37), because
it suggests decentralization. Conversely, in low
power distance cultures, individuals are more
interdependent, whatever their ranks in the hierarchy. Therefore, they will be more favorable to
IT, which does not contradict their perception of
power distribution. For example, in an interpretive study conducted by Hasan and Ditsa (1999)
comparing IT adoption in Australia, West Africa,
and the Middle East, they have found that in the
Middle East, high in power distance, modern
IT, like the Internet, was seen as a threat to the
authoritarian structure of their society. In the
same vein, Martinsons and Westwood (1997) have
demonstrated through an explanatory theory that
paternalism in Chinese culture is an influential
factor regarding the use of computer-based MIS,
compared to American culture.
Straub, Keil, and Brenner (1997), testing TAM
in three countries (the U.S., Switzerland, and Ja-

Information Technology Acceptance across Cultures

pan), also have found that in high power distance

cultures, such as Japan , the leveling effect of
computer-based media is not well appreciated. In
the same sense, Shore and Venkatachalam (1996)
have developed a general framework, contending that the introduction of an IT application in
a high power distance culture will be difficult
to accommodate for superiors having authority
roles and subordinates used to being controlled by
superiors. Actually, a new information technology
may alter lines of authority and responsibility
that are clearly defined in a high power distance
culture. For example, a new system may provide
end users with access to previously guarded and
inaccessible information, or it may establish
new lines of authority through new channels of
electronic communication, such as e-mail (Shore
& Venkatachalam, 1996).
Proposition 4. Masculinity/femininity moderates
the relationship between perceived usefulness and
system usage.
Perceived usefulness will be more determinant for system usage for individuals espousing
masculine values, since the perceived usefulness
of IT is related intimately to achievement values
that predominate in masculine cultures. In feminine cultures, work environment and the quality
of life are much more important than attaining
goals; this is why we expect much less impact
of perceived usefulness of IT on usage for individuals adopting feminine values. Leidner and
Carlsson (1998) have found that the Mexicans,
belonging to a masculine culture, show more
interest in executive information systems (EIS)
than the Swedes. More precisely, the Mexicans
perceive EIS as more useful than the Swedes,
because such a system allows increasing the
speed of decision making, and therefore, work
goals could be reached more easily. In addition,
in masculine cultures in which individuals are
highly assertive, media lacking social presence,
such as e-mail, would not be favored (Straub et

al., 1997), because it does not allow end users to

be highly aggressive and assertive.
Proposition 5. Individualism/collectivism moderates the relationship between subjective norms
and system usage.
People in individualist cultures are more
concerned with themselves than with the group.
Therefore, for individuals holding individualistic
cultural values, the opinions of the members of
the group will not have weight in their decision to
adopt IT. Conversely, people holding collectivist
values will be more concerned with the maintenance of the groups cohesiveness. As a result,
they will be expected to show more interest in
others opinions about IT because they want to
conform to the groups behavior. In a comparative
study on the applicability of the TRA across two
different cultural settings, Korea and the U.S., Lee
and Green (1991) have found that social norms
are more important for collectivistic cultures
(Korea) than for individualistic ones (U.S.) in
influencing behavior.
Proposition 6. Masculinity/femininity moderates
the relationship between subjective norms, and
system usage, such as subjective norms, will have
much less influence on system usage for masculinity-oriented individuals.
In feminine cultures, individuals are expected
to pay more attention to the opinions of others in
behaving since they are more people-oriented than
in masculine cultures in which the most important
thing is goal achievement.
Proposition 7. Power distance moderates the
relationship between subjective norms and system
usage.
In high power distance cultures, since individuals are not supposed to disagree with their
superiors, their reliance upon the opinions of

147

Information Technology Acceptance across Cultures

superiors will be more marked when assessing

IT than for individuals from low power distance
cultures.
Martinsons and Westwood (1997) have demonstrated through their explanatory theory that
paternalism in Chinese culture is an influential
factor regarding the use of management information systems compared to American culture.
Paternalism is a concept very close to high power
distance, since in paternalistic societies, key decisions are made by the patriarch, and management
systems are very centralized.
According to Kowner and Wiseman (2003),
people in tight (vs. loose) cultures in which deviation from behavior is not permitted are more
likely to obey the behavior ascribed by their social
position (p. 182). Consequently, social influence
will play a greater role in shaping IT adoption for
individuals belonging to tight culture than for
those belonging to loose cultures.
Proposition 8. Uncertainty avoidance acts also as
a moderator of the relationship between subjective
norms and system usage.
The social influence exerted by important persons will be much more important in determining
IT use in cultures seeking to avoid uncertainty
than in cultures comfortable with uncertainty.
Actually, in order to deal with uncertainty and
ambiguity, individuals in strong uncertainty
avoidance cultures are very concerned by the
establishment and respect for rules, and therefore, subjective norms will be more important as
guidance to behavior than for individuals in weak
uncertainty avoidance cultures that rely more on
their proper competence to evaluate a situation.
The process of compliance (Kelman, 1958), one
of the basic processes underlying the relationship
between subjective norms and behavior, will be
more salient when individuals are holding high
uncertainty avoidance values. Indeed, compliance supposes that the individual will conform
because a social actor that is important to him or

148

her (and having the power of reward and punishment) wants him or her to behave in a specific
manner. Therefore, conforming to referent others by adopting IT will weaken the uncertainty
pertaining to the nonadoption of IT.

Practical Implications
The framework we provide is an attempt to highlight the importance of culture when dealing with
IT adoption in different national cultural settings.
It should improve IT designers awareness of the
importance of cultural values for IT acceptance
in foreign countries. Actually, the framework
we offer does not suggest prescriptions to be
adopted by IT designers. Rather, it encourages
the latter to take into account cultural values
when designing IS and/or preparing IT implementation procedures for companies around the
world. Actually, IT designers should be aware that
implementation failure could be caused by cultural
incompatibility. As a result, they should adapt
their implementation tactics so that they will be
more congruent with end users cultural systems.
For example, in high power distance cultures, end
users may be reluctant to use computer-mediated
communication tools, because this use may lead
them to contradict their superiors. Therefore, IT
designers should adjust IT features to the social
needs of end users.
Furthermore, the conceptual model we have
proposed suggests that the key belief in TAM,
which is the perceived usefulness of IT, has cultural antecedents. Indeed, the perceived usefulness of IT depends on the type of the context of
communication (high or low) and on the type of
time adopted (monochronic or polychronic). In
addition, the well-established positive relationship between perceived usefulness and system
usage is shown to be moderated by two cultural
values; namely, power distance and masculinity/
femininity. Finally, the intensity of the influence
of subjective norms on system usage has been

Information Technology Acceptance across Cultures

shown to be different, depending on the cultural

values espoused by the IT users. Indeed, social
influence would be most salient when end users
accept and expect inequality in power, do not tolerate uncertainty, are more concerned by the group
than by their selves, and are people-oriented.

Conclusion
In this chapter, we offer a theoretical perspective
on the effect of national culture on IT adoption
and use. Through the conceptual model we have
presented, we wanted to move beyond models
considering culture as a black box and offering an ad hoc cultural explanation of IT-related
behaviors. First, we have identified pertinent
cultural values that could offer a fruitful explanation of differences in IT- related behaviors and
beliefs across countries. Then, we have provided
propositions linking these cultural values to TAM
constructs and relationships. We contend that
high context/low context of communication and
polychronism/monochronism have a direct effect
on perceived usefulness of IT; power distance and
masculinity/femininity have moderator effects
on the relationship between perceived usefulness
and system usage and individualism/collectivism;
uncertainty avoidance, masculinity/femininity,
and power distance have moderator effects on
the relationship between subjective norms and
system usage.
In the current context of internationalization,
taking into account the societal environment in
the understanding of IT acceptance behavior becomes more relevant in comparison with the past.
Indeed, information technologies are so ubiquitous
and pervasive that one should explore more their
relationship with different societal contexts varying across countries. Actually, IT that is accepted
in a specific cultural context may not be accepted
in the same way in another culture.
The framework presented here is a first step
in better understanding the relationship between

national culture and IT adoption and use. It goes

without saying that this conceptual model needs
to be empirically tested. A field study research
involving samples from different countries should
be conducted.

References
Agarwal, R. (1999). Individual acceptance of
information technologies. Retrieved October
2003, from http://www.pinnaflex.com/pdf/framing/CH06.pdf
Agarwal, R., & Karahanna, E. (2000). Time flies
when youre having fun: Cognitive absorption and
beliefs about information technology usage. MIS
Quarterly, 24(4), 665-694.
Agarwal, R., & Prasad, J. (1999). Are individual
differences germane to the acceptance of new
information technologies? Decision Sciences,
30(2), 361-391.
Ajzen, I. (1988). Attitudes, personality,and behavior. Chicago: Dorsey Press.
Ajzen, I., & Fishbein, M. (1980). Understanding
attitudes and predicting social behavior. Englewood-Cliffs, NJ: Prentice-Hall.
Chinese Culture Connection. (1987). Chinese
values and the search for culture free dimensions
of culture. Journal of Cross-Cultural Psychology,
18(2), 143-164.
Danowitz, A. K., Nassef, Y., & Goodman, S. E.
(1995). Cyberspace across the Sahara: Computing
in North Africa. Communications of the ACM,
38(12), 23-28.
Davis, F. D., Bagozzi, R. P., & Warshaw, P. R.
(1989). User acceptance of computer technology:
A comparison of two theoretical models. Management Science, 35(8), 982-1003.

149

Information Technology Acceptance across Cultures

DIribarne, P. (1989). La logique de lhonneur:

Gestion des entreprises et traditions nationales.
Paris: Seuil.
Dishaw, M. T., & Strong, D. M. (1999). Extending
the technology acceptance model with task-technology fit construct. Information and Management, 36, 9-21.
Earley, P. C., & Singh, H. (1995). International
and intercultural management research: Whats
next? Academy of Management Journal, 38(2),
327-340.
Fisher, R., & Smith, P. B. (2003). Reward allocation
and culture: A meta-analysis. Journal of CrossCultural Psychology, 34(3), 251-268.
Gamble, P. R., & Gibson, D. A. (1999). Executive values and decision making: Relationship
of culture and information flows. Journal of
Management Studies, 32(2), 217-240.

Hill, C. E., Loch, K. D., Straub, D. W., & ElSheshai, K. (1998). A qualitative assessment of
Arab culture and information technology transfer.
Journal of Global Information Management,
6(3), 29-38.
Hofstede, G. (1980). Cultures consequences:
International differences in work-related values.
Beverly Hills, CA: Sage.
Hofstede, G. (1985). The interaction between national and organizational value systems. Journal
of Management Studies, 22(4), 347-357.
Hofstede, G. (1997). Cultures and organizations:
Software of the mind. London: McGraw-Hill.
Hofstede, G. J. (2000). The information age across
countries. Proceedings of the 5me Colloque de
lAIM: Systmes dInformation et Changement
Organisationnel, Montpellier, France.

Goodman, S. E., & Green, J. D. (1992). Computing in the Middle East. Communications of the
ACM, 35(8), 21-25.

Hofstede, G., & Bond, M. H. (1984). Hofstedes

culture dimensions. An independent validation
using Rockeachs value survey. Journal of CrossCultural Psychology, 15(4), 417-433.

Gudykunst, W. B., & Ting-Toomey, S. (1988).

Culture and interpersonal communication. Newbury Park, CA: Sage.

Hofstede, G., & Bond, M. H. (1988). The Confucius connection: From cultural roots to economic
growth. Organizational Dynamics, 16(4), 4-21.

Hall, E. T. (1983). The dance of life: The other

dimension of time. New York: Anchor Press
Doubleday.

Hu, P. J., Chau, P. Y. K., Sheng,O. R. L., & Tam, K.

Y. (1999). Examining the technology acceptance
model using physician acceptance of telemedecine
technology. Journal of Management Information
Systems, 16(2), 91-112.

Hall, E. T. (1989). Beyond culture. New York:

Anchor Books Editions.
Hall, E. T., & Hall, R. M. (1987). Hidden differences. Doing business with the Japanese. New
York: Anchor Press Doubleday.
Hasan, H., & Ditsa, G. (1999). The impact of
culture on the adoption of IT: An interpretive
study. Journal of Global Information Management, 7(1), 5-15.
Hernandez, E. (2000, March-April-May). Afrique:
Actualit du modle paternaliste. Revue Franaise
de Gestion, 128, 98-105.
150

Hui, C. H., & Triandis, H. C. (1986). Individualism-collectivismA study of cross-cultural

researchers. Journal of Cross-Cultural Psychology, 17(2), 225-248.
Igbaria, M. (1995). An examination of the factors
contributing to microcomputer technology acceptance. Accounting, Management and Information
Technologies, 4(4), 205-224.
Igbaria, M., & Zviran, M. (1996). Comparison of
end-user computing characteristics in the U.S.,

Information Technology Acceptance across Cultures

Israel and Taiwan. Information and Management,

30(1), 1-13.
Jackson, C. M., Chow, S., & Leitch, R. A. (1997).
Toward an understanding of the behavioral intention to use an information system. Decision
Sciences, 28(2), 357-389.
Kagitibasi, C. (1997). Individualism and collectivism. In J. Berry, W. Poortinga, Y. H. Pandey, J.
Dasen P. R. Saraswathi, T. S. Segall, et al. (Eds.),
Handbook of cross-cultural psychology (Vol. 3).
Boston: Allyn and Bacon.
Karahanna, E., Evaristo, R. J., & Srite, M. (2005).
Levels of culture and individual behavior: An
integrative perspective. Journal of Global Information Management, 13(2), 1-20.
Karahanna, E., & Straub, D. W. (1999). The psychological origins of perceived usefulness and
perceived ease of use. Information and Management, 35, 237-250.
Karahanna, E., Straub, D. W., & Chervany, N. L.
(1999). Information technology adoption across
time: A cross-sectional comparison of pre-adoption and post-adoption beliefs. MIS Quarterly,
23(2), 183-213.
Kelman, H. C. (1958). Compliance, identification,
and internalization three processes of attitude
change. The Journal of Conflict Resolution, 2(1),
51-60.
Kluckhohn, C. (1962). Culture and behavior. New
York: The Free Press of Glencoe.
Kluckhohn, F. R., & Strodtbeck, F. L. (1961).
Variations in value orientations: Evanston, IL:
Row, Peterson.
Kowner, R., & Wiseman, R. (2003). Culture and
status-related behavior: Japanese and American
perceptions of interaction in asymmetric dyads.
Cross-Cultural Research, 37(2), 178-210.

Laurent, A. (1983). The cultural diversity of

Western conceptions of management. International Studies of Management and Organization,
13(1-2), 75-96.
Lee, C., & Green, R. T. (1991, second quarter).
Cross-cultural examination of the Fishbein behavioral intentions model. Journal of International
Business Studies, 22(2), 289-305.
Leidner, D. E., & Carlsson, S. A. (1998).
Les
bnefices des systmes dinformation pour dirigeants dans trois pays. Systmes dInformation
-
27.
et Management, 3(3), 5
Levi-Strauss, C. (1985). Anthropologie structurale. Paris: Librairie Plon.
Liska, A. L. (1984). A critical examination of the
causal structure of the Fishbein/Ajzen attitudebehavior model. Social Psychology Quarterly,
47(1), 61-74.
Lucas, H. C., & Spitler, V. (2000). Implementation
in a world of workstations and networks. Information and Management, 38, 119-128.
Martinsons, M. G., & Westwood, R. I. (1997).
Management information systems in the Chinese
business culture: An explanatory theory. Information & Management, 32(5), 215-228.
Mathieson, K. (1991). Predicting user intentions:
Comparing the technology acceptance model
with the theory of planned behavior. Information
Systems Research, 2(3), 173-191.
Moon, J-W., & Kim, Y-G. (2001). Extending the
TAM for a World-Wide-Web context. Information
and Management, 38, 217-230.
Porter, M. E., & Millar, V. E. (1985, July-August).
How information gives you competitive advantage. Harvard Business Review, 63(4), 149-160.
Rice, R. E., DAmbra, J., & More, E. (1998).
Cross-culture comparison of organizational media
evaluation and choice. Journal of Communication, 48(3), 3-26.

151

Information Technology Acceptance across Cultures

Robey, D., & Rodriguez-Diaz, A. (1989). The

organizational and cultural context of systems
implementation: Case experience from Latin
America. Information and Management, 17,
229-239.
Ronen, S., & Shenkar, O. (1988). Clustering variables: The application of nonmetric multivariate
analysis techniques in comparative management
research. International Studies of Management
and Organization, 18(3), 72-87.
Rowe, F., & Struck, D. (1999). Cultural values,
media richness and telecommunication use in
an organization. Accounting, Management and
Information Technologies, 9, 161-192.
Schneider, S. C., & Barsoux, J. L. (2003). Management interculture (trans. V. Lavoyer). Paris:
Pearson Education France.
Schramm-Nielsen, J. (2000, March-April-May).
Dimensions culturelles des prises de dcision
Une comparaison France-Danemark. Revue
Franaise de Gestion, 76-86.
Schwartz, S. H. (1994). Beyond

individualism-collectivism. In U. Kim, H. C. Triandis, C.

Kagitibasi, S. -C. Choi, & G. Yoon (Eds.), Individualism and collectivism: Theory, method and
applications. Newbury Park, CA: Sage.
Shackleton, V. J., & Ali, A. H. (1990). Work- related values of managersA test of the Hofstede
model. Journal of Cross-Cultural Psychology,
21(1), 109-118.
Sheppard, B. H., Hartwick, J., & Warshaw, P.R.
(1988). The theory of reasoned action: A metaanalysis of past research with recommendations
for modifications and future research. Journal of
Consumer Research, 15, 325-343.
Shore, B., & Venkatachalam, A. R. (1996). Role
of national culture in the transfer of information
technology. Strategic Information Systems, 5(1),
19-35.

152

Shuper, P. A., Sorrentino, R. M., Otsubo,Y.,

Hodson,G., & Walker, A. M. (2004). Theory
of uncertainty orientation: Implications for the
study of individual differences within and across
cultures. Journal of Cross-Cultural Psychology,
-
480.
35(4), 460
Silvester, J. (1990). Systme hirarchique et
analyse socitale: Comparaison France-Allemagne-Japon. Revue Franaise de Gestion (77),
107
-
114.
Smith, P. B., & Bond, M. H. (1999). Social psychology across cultures. Boston: Allyn and Bacon.
Srite, M., & Karahanna, E. (n.d.). The influence of
national culture on the acceptance of information
technologies: An empirical study. MIS Quarterly,
30(3), 679-704.
Straub, D., Keil, M., & Brenner, W. (1997). Testing

the technology acceptance model across cultures:

A three country study. Information and Management, 33, 1-11.
Straub, D., Loch, K., Evaristo, R., Karahanna,
E., & Srite, M. (2002). Towards a theory based
definition of culture. Journal of Global Information Management, 10(1), 13-23.
Straub, D. W. (1994). The effect of culture on IT
diffusion: E-mail and fax in Japan and the U.S.
Information Systems Research, 5(1).
Szajna, B. (1996). Empirical evaluation of the
revised technology acceptance model. Management Science, 42(1), 85-92.
Tan, B. C. Y., Wei, K., Watson, R. T., Clapper,
D. L., & McLean, E. R. (1998). Computer-mediated communication and majority influence:
Assessing the impact in an individualistic and
collectivistic culture. Management Science, 44(9),
1263-1278.
Tapscott, D., & Caston, A. (1994). Lentreprise de
la deuxime re. La rvolution des technologies
de linformation. Paris: Dunod.

Information Technology Acceptance across Cultures

Taylor, S., & Todd, P. A. (1995).

Understanding
information technology usage: A test of competing models. Information Systems Research, 6(2),
144-176.
Thomas, D. C., & Pekerti, A. A. (2003). Effect of
culture on situational determinants of exchange
behavior in organizations: A comparison of New
Zealand and Indonesia. Journal of Cross-Cultural
Psychology, 34(3), 269-281.
Triandis, H. C. (1972). The analysis of subjective
culture. New York: John Wiley and Sons.
Triandis, H. C. (1977). Interpersonal behavior.
Monterey, CA: Brooks/Cole.
Triandis, H. C. (1989). The self and social behavior in differing cultural contexts. Psychological
Review, 96(3), 506-520.
Trompenaars, F., & Hampden-Turner, C. (1998).
Riding the waves of culture (2nd ed.). New York:
MacGraw-Hill.

Venkatesh, V. (1999). Creation of favorable user

perceptions: Exploring the role of intrinsic motivation. MIS Quarterly, 23(2), 239-260.
Venkatesh, V., & Davis, F. (2000). A theoretical
extension of technology acceptance model: Four
longitudinal field studies. Management Science,
46(2), 186-204.
Venkatesh, V., & Morris, M. G. (2000). Why
dont men ever stop to ask for directions? Gender, social influence, and their role in technology
acceptance and usage behavior. MIS Quarterly,
24(1), 115-139.
Venkatesh, V., Speir, C., & Morris, M. G. (2002).
User acceptance enablers in individual decision
making about technology: Toward an integrated
model. Decision Sciences, 33(2), 297-316.
Zghal, R. (1994). La culture de la dignit et le
flou de lorganisation: Culture et comportement
organisationnelSchma thorique et applicaTunis: Centre dEtudes, de
tion au cas tunisien.
Recherches et de Publications.

This work was previously published in Information Resources Management: Global Challenges, edited by W. Law, pp. 25-53,
copyright 2007 by IGI Publishing, formerly known as Idea Group Publishing (an imprint of IGI Global).

153

154

Chapter 1.12

Is Information Ethics
Culture-Relative?
Philip Brey
University of Twente, The Netherlands

Abstract

Introduction

In this article, I examine whether information

ethics is culture relative. If it is, different approaches to information ethics are required in
different cultures and societies. This would have
major implications for the current, predominantly
Western approach to information ethics. If it is
not, there must be concepts and principles of
information ethics that have universal validity.
What would they be? The descriptive evidence
is for the cultural relativity of information ethics
will be studied by examining cultural differences
between ethical attitudes towards privacy, freedom of information, and intellectual property
rights in Western and non-Western cultures. I then
analyze what the implications of these findings
are for the metaethical question of whether moral
claims must be justified differently in different
cultures. Finally, I evaluate what the implications
are for the practice of information ethics in a
cross-cultural context.

Information ethics1 has so far mainly been a

topic of research and debate in Western countries, and has mainly been studied by Western
scholars. There is, however, increasing interest
in information ethics in non-Western countries
like Japan, China, and India, and there have been
recent attempts to raise cross-cultural issues in
information ethics (e.g., Ess, 2002; Gorniak-Kocikowska, 1996; Mizutani, Dorsey & Moor, 2004).
Interactions between scholars of Western and
non-Western countries have brought significant
differences to light between the way in which they
approach issues in information ethics. This raises
the question whether different cultures require a
different information ethics and whether concepts
and approaches in Western information ethics can
be validly applied to the moral dilemmas of nonWestern cultures. In other words, is information
ethics culturally relative or are there concepts and
principles of information ethics that have universal validity? The aim of this essay is to arrive at
preliminary answers to this question.

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Is Information Ethics Culture-Relative?

Moral Relativism and

Information Ethics
In discussions of moral relativism, a distinction
is commonly made between descriptive and
metaethical moral relativism. Descriptive moral
relativism is the position that as a matter of empirical fact, there is extensive diversity between the
values and moral principles of societies, groups,
cultures, historical periods, or individuals. Existing differences in moral values, it is claimed,
are not superficial but profound, and extend to
core moral values and principles. Descriptive
moral relativism is an empirical thesis that can
in principle be supported or refuted through
psychological, sociological, and anthropological
investigations. The opposite of descriptive moral
relativism is descriptive moral absolutism, the
thesis that there are no profound moral disagreements exist between societies, groups, cultures,
or individuals. At issue in this essay will be a
specific version of descriptive moral relativism,
descriptive cultural relativism, according to which
there are major differences between the moral
principles of different cultures.
Much more controversial than the thesis of descriptive moral relativism is the thesis of metaethical
moral relativism, according to which the truth or

justification of moral judgments is not absolute

or objective, but relative to societies, groups, cultures, historical periods, or individuals.2 Whereas
a descriptive relativist could make the empirical
observation that one society, polygamy, is considered moral whereas in another it is considered
immoral, a metaethical relativist could make the
more far-reaching claim that the statement polygamy is morally wrong is true or justified in
some societies while false or unjustified in others.
Descriptive relativism therefore makes claims
about the values that different people or societies actually have, whereas metaethical relativism
makes claims about the values that they are justified in having. Metaethical moral relativism is
antithetical to metaethical moral absolutism, the

thesis that regardless of any existing differences

between moral values in different cultures, societies, or individuals, there are moral principles that
are absolute or objective, and that are universally
true across cultures, societies, or individuals. Metaethical moral absolutism would therefore hold
that the statement polygamy is morally wrong
is either universally true or universally false;
it cannot be true for some cultures or societies
but false for others. If the statement is true, then
societies that hold that polygamy is moral are in
error, and if it is false, then the mistake lies with
societies that condemn it.
The question being investigated in this essay is
whether information ethics is culturally relative.
In answering this question, it has to be kept in
mind that the principal aims of information ethics
are not descriptive, but normative and evaluative.
That is, its principal aim is not to describe existing morality regarding information but rather to
morally evaluate information practices and to prescribe and justify moral standards and principles
for practices involving the production, consumption, or processing of information. A claim that
information ethics is culturally relative therefore a
claim that metaethical moral relativism is true for
information ethics. It is to claim that the ethical
values, principles, and judgments of information
ethics are valid only relative to a particular culture,
presumably the culture in which they have been
developed. Since information ethics is largely a
product of the West, an affirmation of the cultural
relativity of information ethics means that its
values and principles do not straightforwardly
apply to non-Western cultures.
But if the cultural relativity of information
ethics depends on the truth of metaethical relativism, does any consideration need to be given
to descriptive relativism for information ethics?
This question should be answered affirmatively.
Defenses of metaethical relativism usually depend
on previous observations that descriptive relativism is true. If descriptive relativism is false, it
follows that people across the world share a moral

155

Is Information Ethics Culture-Relative?

framework of basic values and principles. But if

this is the case, then it seems pointless to argue
for metaethical moral relativism: why claim that
the truth of moral judgments is different for different groups if these groups already agree on
basic moral values? On the other hand, if descriptive relativism is true, then attempts to declare
particular moral principles of judgments to be
universally valid come under scrutiny. Extensive
justification would be required for any attempt to
adopt a particular moral framework (say, Western information ethics) as one that is universally
valid. In the next section, I will therefore focus
on the question whether there are good reasons to
believe that there are deep and widespread moral
disagreements about central values and principles
in information ethics across cultures, and whether
therefore descriptive cultural relativism is true for
information ethics.

The Descriptive
Cultural Relativity of
Information-Related
Values
In this section, I will investigate the descriptive
cultural relativity of three values that are the topic
of many studies in information ethics: privacy,
intellectual property, and freedom of information.
Arguments have been made that these values
are distinctly Western, and are not universally
accepted across different cultures. In what follows I will investigate whether these claims seem
warranted by empirical evidence. I will also relate
the outcome of my investigations to discussions
of more general differences between Western and
non-Western systems of morality.
How can it be determined that cultures have
fundamentally different value systems regarding
notions like privacy and intellectual property?
I propose that three kinds of evidence are relevant:

156

1.

2.

3.

Conceptual: the extent to which there are

moral concepts across cultures with similar
meanings. For example, does Chinese culture have a concept of privacy that is similar
to the American concept of privacy?
Institutional: the extent to which there
is similarity between codified rules that
express moral principles and codified statements that express moral judgments about
particular (types of) situations. For example,
are the moral principles exhibited in the
laws and written rules employed in Latin
cultures on the topic of privacy sufficiently
similar to American laws and rules that it
can be claimed that they embody similar
moral principles?
Behavioral: the similarity between customs
and behaviors that appear to be guided by
moral principles. This would include tendencies to avoid behaviors that are immoral
regarding a moral principle, tendencies to
show disapproval to those who engage in
such behaviors, and to show disapproval to
those who do not, and tendencies to show
remorse or guilt when engaging in such
behaviors. For instance, if a culture has
a shared privacy principle that states that
peeking inside someones purse is wrong,
then it can be expected that most people try
not to do this, disapprove of those who do,
and feel ashamed or remorseful when they
are caught doing it.

It is conceivable that in a particular culture a

value or moral principle is widely upheld at the
behavioral level, but has not (yet) been codified
at the institutional and conceptual level. But this
is perhaps unlikely in cultures with institutions
that include extensive systems of codified rules,
which would include any culture with a modern
legal system. It is also conceivable that a moral
value or principle is embodied in both behavioral
customs and codified rules, but no good match
can be found at the conceptual level. In that case,

Is Information Ethics Culture-Relative?

it seems reasonable to assume that the value or

principle at issue is embodied in the culture, but
different concepts are used to express it, making
it difficult to find direct translations.
A full consideration of the evidence for descriptive moral relativism along these three lines
is beyond the scope of this article. I only intend
to consider enough evidence to arrive at a preliminary assessment of the cultural relativity of
values in contemporary information ethics.

Privacy
It has been claimed that in Asian cultures like
China and Japan, no genuine concept or value of
privacy exists. These cultures have been held to
value the collective over the individual. Privacy
is an individual right, and such a right may not
be recognized in a culture where collective interest tend to take priority over individual interests. Using the three criteria outline above, and
drawing from studies of privacy in Japan, China
and Thailand, I will now consider whether this
conclusion is warranted.
At the conceptual level, there are words in
Japanese, Chinese, and Thai that refer to a private
sphere, but these words seem to have substantially
different meanings than the English word for privacy. Mizutani et al. (2004) have argued that there
is no word for privacy in traditional Japanese.
Modern Japanese, they claim, sometimes adopt
a Japanese translation for the Western word for
privacy, which sounds like puraibashii, and
written in katakana. Katakana is the Japanese
phonetic syllabary that is mostly used for words of
foreign origin. According to Nakada and Tamura
(2005), Japanese does include a word for private,
Watakusi, which means partial, secret and
selfish. It is opposed to Ohyake, which means
public. Things that are Watakusi are considered
less worthy than things that are Ohyake. Mizutani
et al. (2004) point out, in addition, that there are
certainly behavioral customs in Japan that amount

to a respect for privacy. There are conventions that

restrict access to information, places, or objects.
For example, one is not supposed to look under
clothes on public streets.
In China, the word closest to the English privacy is Yinsi, which means shameful secret
and is usually associated with negative, shameful
things. L (2005) claims that only recently that
Yinsi has also come to take broader meanings
to include personal information, shameful or not,
that people do not want others to know (see also
Jingchun, 2005; McDougall & Hansson, 2002).
This shift in meaning has occurred under Western
influences. As for institutional encoding of privacy
principles, L maintains that there currently are
no laws in China that protect an individual right
to privacy, and the legal protection of privacy
has been weak and is still limited, though there
have been improvements in privacy protection
since the 1980s.
Kitiyadisai (2005), finally, holds that the concept of privacy does not exist in Thailand. She
claims that the Western word privacy was adopted
in the late nineteenth or early twentieth century
in Thailand, being transliterated as privade,
but this word gained a distinctly Thai meaning,
being understood as a collectivist rather than an
individual notion. It referred to a private sphere
in which casual dress could be worn, as opposed
to a public sphere in which respectable dress had
to be worn. In the Thai legal system, Kitiyadisai
claims there has not been any right to privacy
since the introduction of privacy legislation in
1997 and a Thai constitution, also in 1997, that
for the first time guarantees basic human rights.
Kitiyadisai argues, however, that Thai privacy
laws are hardly enacted in practice, and many
Thais remain unaware of the notion of privacy.
It can be tentatively concluded that the introduction of a concept of privacy similar to the
Western notion has only taken place recently in
Japan, China, and Thailand, and that privacy legislation has only taken place recently. In traditional

157

Is Information Ethics Culture-Relative?

Japanese, Chinese, and Thai culture, which still

has a strong presence today, distinctions are made
that resemble the Western distinction between
public and private, and customs exist that may be
interpreted as respective of privacy, but there is
no recognized individual right to privacy.

Intellectual Property Rights

In discussing the cultural relativity of intellectual
property rights (IPR), I will limit myself to one
example: China. China is known for not having a developed notion of private or individual
property. Under communist rule, the dominant
notion of property was collective. All means of
production, such as farms and factories, were to
be collectively owned and operated. Moreover,
the state exercised strict control over the means
of production and over both the public and private
sphere. A modern notion of private property was
only introduced since the late 1980s. Milestones
were a 1988 constitutional revision that allowed
for private ownership of means of production and
a 2004 constitutional amendment that protects
citizens from encroachment of private property.
The notion of intellectual property has only
recently been introduced in China, in the wake of
Chinas recent economic reforms and increased
economic interaction with the West. China is
currently passing IPR laws and cracking down
on violations of IPR in order to harmonize the
Chinese economic system with the rest of the
world. But as journalist Ben Worthen observes,
the average citizen in China has no need and
little regard for intellectual property. IPR is not
something that people grew up with and the
percent of citizens who learn about it by engaging in international commerce is tiny. Worthen
also points out that Chinese companies have no
incentive to respect IPR unless they are doing
work for Western companies that demand it
and that since most of the intellectual property
royalties are headed out of China there isnt a lot
of incentive for the government to crack down
on companies that choose to ignore IPR.3 All in
all, it can be concluded that Chinas value system

158

traditionally has not included a recognition of

intellectual property rights, and it is currently
struggling with this concept.

Freedom of Information
Freedom of information is often held to comprise
two principles: freedom of speech (the freedom
to express ones opinions or ideas, in speech or
in writing) and freedom of access to information.
Sometimes, freedom of the press (the freedom to
express oneself through publication and dissemination) is distinguished as a third principle. In
Western countries, freedom of information is often
defined as a constitutional and inalienable right.
Laws protective of freedom of information are often especially designed to ensure that individuals
can exercise this freedom without governmental
interference or constraint. Government censorship
or interference is only permitted in extreme situations, pertaining to such things as hate speech,
libel, copyright violations, and information that
could undermine national security.
In many non-Western countries, freedom of
information is not a guiding principle. There
are few institutionalized protections of freedom
of information; there are many practices that
interfere with freedom of information, and a
concept of freedom of information is not part
of the established discourse in society. In such
societies, the national interest takes precedence,
and an independent right to freedom information
either is not recognized or is made so subordinate
to national interests that it hardly resembles the
Western right to freedom of information. These
are countries in which practices of state censorship
are widespread; mass media are largely or wholly
government-controlled, the Internet, databases,
and libraries are censored, and messages that do
not conform to the party line are cracked down
upon.
Let us, as an example, consider the extent to
which freedom of information can be said to be
a value in Chinese society. Until the 1980s, the

Is Information Ethics Culture-Relative?

idea of individual rights or civil rights was not a

well-known concept in China. Government was
thought to exist to ensure a stable society and a
prosperous economy. It was not believed to have
a function to protect individual rights against collective and state interests. As a consequence of
this general orientation, the idea of an individual
right to freedom of information was virtually
unknown. Only recently has China introduced
comprehensive civil rights legislation. In its 1982
constitution, China introduced constitutional
principles of freedom of speech and of the press.
And in 1997, it signed the International Convention on Economic, Social, and Cultural Rights,
and in 1998 the International Convention on Civil
and Political Rights (the latter of which it has not
yet ratified).
Even though the Chinese government has
recently come to recognize a right to freedom of
information, as well as individual human rights
in general, and has introduced legislation to this
effect, state censorship is still rampant, and the
principle of upholding state interest still tends to
dominate the principle of protecting individual
human rights. Internet censorship presents a good
example of this. Internet traffic in China is controlled through what the Chinese call the Golden
Shield, and what is known outside mainland China
as the Great Firewall of China. This is a system
of control in which Internet content is blocked by
routers, as well as at the backbone and ISP level,
through the filtering of undesirable URLs and
keywords. A long list of such forbidden URLs
and keywords has been composed by the Chinese
State Council Information Office, in collaboration
with the Communist Partys Propaganda Department. This system is especially geared towards
censorship of content coming from outside mainland China (Human Rights Watch, 2006).

Rights-Centered and
Virtue-Centered Morality
A recurring theme in the above three discussions
has been the absence of a strong tradition of individual rights in the cultures that were discussed
those of China, Japan, and Thailand and the
priority that is given to collective and state interests. Only very recently have China, Japan, and
Thailand introduced comprehensive human rights
legislation, which has occurred mainly through
Western influence, and there is still considerable
tension in these societies, especially in China
and Thailand, between values that prioritize the
collective and the state and values that prioritize
the individual.
Various authors have attempted to explain the
worldview that underlies the value system of these
countries. In Japan and Thailand, and to a lesser
extent China, Buddhism is key to an understanding
of attitudes towards individual rights. Buddhism
holds a conception of the self that is antithetical
to the Western conception of an autonomous
self which aspires to self-realization. Buddhism
holds that the self does not exist and that human
desires are delusional. The highest state that humans can reach is Nirvana, a state of peace and
contentment in which all suffering has ended. To
reach Nirvana, humans have to become detached
from their desires, and realize that the notion of
an integrated and permanent self is an illusion. In
Buddhism, the self is defined as fluid, situationdependent, and ever-changing. As Mizutani et al.
and Kitiyadisai have noted, such a notion of the
self is at odds with a Western notion of privacy
and of human rights in general, notions which
presuppose a situation-independent, autonomous
self which pursues its own self-interests and which
has inalienable rights that have to be defended
against external threats.
In part through Buddhism, but also through
the influence of other systems of belief such as
Confucianism, Taoism, and Maoism, societies

159

Is Information Ethics Culture-Relative?

like those of China and Thailand have developed

a value system in which the rights or interests
of the individual are subordinate to those of the
collective and the state. To do good is to further
the interests of the collective. Such furtherances
of collective interests will generally also benefit
the individual. The task of government, then, is to
ensure that society as a whole functions well, in a
harmonious and orderly way, and that social ills are
cured, rather than the ills of single individuals. In
other words, government works for the common
good, and not for the individual good.
Only recently have countries like China and
Thailand come to recognize individual human
rights and individual interests next to collective
interests. But according to L (2005), the collectivist ethic still prevails:
Adapting to the demands of social diversity, the
predominant ethics now express a new viewpoint
that argues against the simple denial of individual
interests and emphasizes instead the dialectical
unification of collective interests and individual
interests: in doing so, however, this ethics points
out that this kind of unification must take collective
interests as the foundation. That is to say, in the
light of the collectivism principle of the prevailing
ethics, collective interests and individual interests
are both important, but comparatively speaking,
the collective interests are more important than
individual interests. (L, 2005, p. 12)
If this observation is correct, then the introduction of human rights legislation and property rights
in countries like China is perhaps not motivated
by a genuine recognition of inalienable individual
human rights, but rather a recognition that in the
current international climate, it is better to introduce human rights and property rights, because
such principles will lead to greater economic
prosperity, which is ultimately to the benefit of
the collective.
The dominant value systems prevalent in
China, Thailand, and Japan are examples of

160

what philosopher David Wong (1984) has called

virtue-centered moralities. According to Wong,
at least two different approaches to morality can
be found in the world: a virtue-centered morality
that emphasizes the good of the community, and
a rights-centered morality that stresses the value
of individual freedom. Rights-centered morality
is the province of the modern West, although it
is also establishing footholds in other parts of the
world. Virtue-centered morality can be found
in traditional cultures such as can be found in
southern and eastern Asia and in Africa. Wongs
distinction corresponds with the frequently made
distinction between individualist and collectivist
culture, that is found, amongst others, in Geert
Hofstedes (1991) well-known five-dimensional
model of cultural difference. However, this latter
distinction focuses on social systems and cultural
practices, whereas Wong makes a distinction
based in differences in moral systems.
In Wongs conception of virtue-centered moralities, individuals have duties and responsibilities that stem from the central value of a common
good. The common good is conceived of in terms
of an ideal conception of community life, which
is based on a well-balanced social order in which
every member of the community has different duties and different virtues to promote the common
good. Some duties and virtues may be shared by
all members. The idea that human beings have
individual rights is difficult to maintain in this
kind of value system, because recognition of such
rights would have to find its basis in the higher
ideal of the common good. But it seems clear that
attributing rights to individuals is not always to
the benefit of the common good. The recognition
of individual property rights, for example, could
result in individual property owners not sharing
valuable resources that would benefit the whole
community. In virtue-centered moralities, the
ideal is for individuals to be virtuous, and virtuous
individuals are those individuals whose individual
good coincides with their contribution to the
common good. Individual goods may be recog-

Is Information Ethics Culture-Relative?

nized in such communities, but they are always

subordinate to the common good. Individuals
deserve respect only because of their perceived
contribution to the common good, not because
they possess inalienable individual rights.

Conclusion
The discussion of privacy, intellectual property
rights, and freedom of information has shown
that a good case can be made for the descriptive
cultural relativity of these values. These values
are central in information ethics, as it has been developed in the West. Moreover, it was argued that
the uncovered cultural differences in the appraisal
of these values can be placed in the context of a
dichotomy between two fundamentally different
kinds of value systems that exist in different societies: rights-centered and virtue-centered systems
of value. Information ethics, as it has developed
in the West, has a strong emphasis on rights, and
little attention is paid to the kinds of moral concerns that may exist in virtue-centered systems
of morality. In sum, it seems that the values that
are of central concern in Western information
ethics are not the values that are central in many
non-Western systems of morality. The conclusion
therefore seems warranted that descriptive moral
relativism is true for information ethics.

Metaethical Moral
Relativism and
Information Ethics
In the first section, it was argued that descriptive moral relativism is a necessary condition for
metaethical moral relativism, but is not sufficient
to prove this doctrine. However, several moral
arguments exist that use the truth of descriptive
relativism, together with additional premises, to
argue for metaethical relativism. I will start with
a consideration of two standard arguments of this

form, which are found wanting, after which I will

consider a more sophisticated argument.

Two Standard Arguments for

Metaethical Relativism
There are two traditional arguments for metaethical moral relativism that rely on the truth of
descriptive moral relativism (Wong, 1993). The
one most frequently alluded to is the argument
from diversity. This argument starts with the
observation that different cultures employ widely
different moral standards. Without introducing
additional premises, the argument goes on to
conclude that therefore, there are no universal
moral standards. This argument rests on what is
known in philosophy as a naturalistic fallacy, an
attempt to derive a norm from a fact, or an ought
from an is. The premise of the argument is
descriptive: there are different moral standards.
The conclusion is normative: no moral standard
has universal validity. No evidence has been
presented that the truth of the premise has any
bearing on the truth of the conclusion.
A second, stronger argument for moral relativism is the argument from functional necessity,
according to which certain ethical beliefs in a
society may be so central to its functioning that
they cannot be given up without destroying the
society. Consequently, the argument runs, these
ethical beliefs are true for that society, but not
necessarily in another. However, this argument
is also problematic because it grounds the truth
of ethical statements in their practical value for
maintaining social order in a particular society.
Such a standard of justification for ethical statements is clearly too narrow, as it could be used to
justify the moral beliefs of societies whose beliefs
and practices are clearly unethical, for instance,
fascist societies. If a society operates in a fundamentally unethical way, then the transformation
of some of its social structures and cultural forms
would seem acceptable if more ethical practices
are the result.

161

Is Information Ethics Culture-Relative?

Wongs and Harmans Argument for

Metaethical Relativism
More convincing arguments for moral relativism
have been presented by David Wong (1984, 2006)
and Gilbert Harman (1996, 2000). Their argument runs, in broad outline, as follows. There are
deep-seated differences in moral belief between
different cultures. Careful consideration of the
reasons for these moral beliefs they have shows
that they are elements of different strategies to
realize related but different conceptions of the
Good. No good arguments can be given why one
of these conceptions of the Good is significantly
better than all the others. Therefore, these moral
beliefs are best explained as different but (roughly)
equally valid strategies for attaining the Good.
This is a much better argument than the previous two, since it puts the ball in the metaethical
absolutists court: he will have to come up with
proof that it is possible to provide good arguments
for the superiority of one particular conception of
the Good over all other conceptions. Metaethical
absolutists can respond to this challenge in two
ways. First, they may choose to bite the bullet
and claim that a rational comparison of different
conceptions of the Good is indeed possible. Different conceptions of the Good, they may argue,
rely on factual or logical presuppositions that
may be shown to be false. Alternatively, they
may argue that there are universally shared moral
intuitions about what is good, and these intuitions
can be appealed to in defending or discrediting
particular conceptions of the Good. For instance
an individual who believes that physical pleasure is
the highest good could conceivably be persuaded
to abandon this belief through exposure to arguments that purport to demonstrate that there are
other goods overlooked by this individual that
are at least as valuable. Such an argument could
conceivably rely on someones moral intuitions
about the Good that could be shown to deviate
from someones explicit concept of the Good.
Second, a mixed position could be proposed,

162

according to which it is conceded that individuals

or cultures may hold different conceptions of the
Good that cannot be rationally criticized (pace
metaethical relativism) but that rational criticism of individual moral beliefs is nevertheless
possible (pace metaethical absolutism) because
these beliefs can be evaluated for their effectiveness in realizing the Good in which service they
stand. After all, if moral beliefs are strategies to
realize a particular conception of the Good, as
Wong and Harman have argued, then they can
be suboptimal in doing so. A belief that Internet
censorship is justified because it contributes to
a more stable and orderly society can be wrong
because it may not in fact contribute to a more
stable and orderly society. Empirical arguments
may be made that Internet censorship is not necessary for the maintenance of social order, or even
that Internet censorship may ultimately work to
undermine social order, for example, because it
creates discontentment and resistance.
In the existing dialogue between proponents
of rights-centered and virtue-centered systems
of morality, it appears that both these approaches
are already being taken. Western scholars have
criticized the organicist conception of society
that underlies conceptions of the Good in many
Asian cultures, while Western definitions of the
Good in terms of individual well-being have
been criticized for their atomistic conception of
individuals. Rights-based systems of morality
have been criticized for undervaluing the common
good, whereas virtue-based systems have been
criticized for overlooking the importance of the
individual good. In addition, both rights-centered
and virtue-centered systems of morality have
been criticized for not being successful by their
own standards. Western individualism has been
claimed to promote selfishness and strife, which
results in many unhappy individuals plagued
by avarice, proverty, depression, and loneliness.
Western societies have therefore been claimed to
be unsuccessful in attaining their own notion of
the Good, defined in terms of individual well-be-

Is Information Ethics Culture-Relative?

ing. Virtue-centered cultures have been claimed

to be have difficulty in developing strong economies that serve the common good, because good
economies have been argued to require private
enterprise and a more individualist culture. In
addition, strong state control, which is a feature
of many virtue-centered cultures, has been argued
to lead to corruption and totalitarianism, which
also do not serve the common good.
In light of the preceding observations, it seems
warranted to conclude, pace metaethical absolutism, that rational criticism between different moral
systems is possible. It does not follow, however,
that conclusive arguments for universal moral
truths or the superiority of one particular moral
system over others are going to be possible. Critics of a particular moral system may succeed in
convincing its adherents that the system has its
flaws and needs to be modified, but it could well
be that no amount of criticism ever succeeds in
convincing its adherents to abandon core moral
beliefs within that system, however rational and
open-minded these adherents are in listening to
such criticism.

Conclusion
I have argued, pace metaethical relativism, that it
is difficult if not impossible to provide compelling
arguments for the superiority of different notions
of the Good that are central in different moral
systems, and by implication, that it is difficult to
present conclusive arguments for the universal
truth of particular moral principles and beliefs. I
have also argued, pace metaethical absolutism,
that is nevertheless possible to develop rational
arguments for and against particular moral values
and overarching conceptions of the Good across
moral systems, even if such arguments do not
result in proofs of the superiority of one particular
moral system or moral principle over another.
From these two metaethical claims, a normative position can be derived concerning the
way in which cross-cultural ethics ought to take

place. It follows, first of all, that it is only justified for proponents of a particular moral value
or principle to claim that it ought to be accepted
in another culture if they make this claim on the
basis of a thorough understanding of the moral
system operative in this other culture. The proponent would have to understand how this moral
system functions and what notion of the Good it
services, and would have to have strong arguments
that either the exogenous value would be a good
addition to the moral system in helping to bring
about the Good serviced in that moral system,
or that the notion of the Good serviced in that
culture is flawed and requires revisions. In the
next section, I will consider implications of this
position for the practice of information ethics in
cross-cultural settings.

Information Ethics in a
Cross-Cultural Context
It is an outcome of the preceding sections that
significant differences exist between moral systems of different cultures, that these differences
have important implications for moral attitudes
towards uses of information and information
technology, and that there are good reasons to
take such differences seriously in normative studies in information ethics. In this section, I will
argue, following Rafael Capurro, that we need an
intercultural information ethics that studies and
evaluates cultural differences in moral attitudes
towards information and information technology.
I will also critically evaluate the claim that the Internet will enable a new global ethic that provides
a unified moral framework for all cultures.

Intercultural Information Ethics

The notion of an intercultural information ethics
(IIE) was first introduced by Rafael Capurro (2005,
in press), who defined it as a field of research in
which moral questions regarding information

163

Is Information Ethics Culture-Relative?

technology and the use of information are reflected

on in a comparative manner on the basis of different cultural traditions. I will adopt Capurros
definition, but differ with him on what the central
tasks of an IIE should be. Capurro defines the tasks
of IIE very broadly. For him, they not only the
comparative study of value systems in different
cultures in relation to their use of information
and information technology, but also
studies of
the effect of information technology on customs,
languages, and everyday problems, the changes

produced by the Internet on traditional media, and

the economic impact of the Internet to the extent
that it can become an instrument of cultural oppression and colonialism.
I hold, in contrast, that studies of the effects of
information technology in non-Western cultures
are more appropriately delegated to the social
sciences (including communication studies,
cultural studies, anthropology and science, and
technology studies). An intercultural information
ethics should primarily focus on the comparative
study of moral systems. Its overall aim would be
to interpret, compare, and critically evaluate moral
systems in different cultures regarding their moral
attitudes towards and behavior towards information and information technology.
This task for IIE can be broken down into
four subtasks, the first two of which are exercises
in descriptive ethics and the latter two of which
belong to normative ethics. First,

IIE should engage in interpretive studies of moral systems in

particular cultures, including the systems of value
contained in the religious and political ideologies
that are dominant in these cultures. The primary
focus in such interpretive studies within the context of IIE should be on resulting moral attitudes
towards the use and implications of information
technology and on the moral problems generated
by uses of information technology within the
context of the prevailing moral system. Second,
IIE should engage in comparative studies of moral
systems from different cultures, and arrive at
analyses of both similarities and differences in

164

the way that these moral systems are organized

and operate, with a specific focus on the way in
which they have different moral attitudes towards
implications of information technology and on
differences in moral problems generated by the
use of information technology.
Third, IIE should engage in critical studies
in which the moral systems of particular cultures are criticized based on the insights gained
through the interpretive and comparative studies
alluded to above, particularly in their dealings
with information technology. Critical studies
may be directed towards critizing moral values
and beliefs in cultures other than ones own, and
proposing modifications in the cultures moral
system and ways in which it should solve moral
problems, but may also involve self-criticism,
in which ones own moral values and the moral
system of ones own culture is criticized based
on insights gained from the study of alternative
moral systems. Fourth, IIE should engage in
interrelational studies that focus on the construction of normative models for interaction between
cultures in their dealings with information and
information technology that respect their different moral systems. Interrelational studies hence
investigate what moral compromises cultures can
make and ought to make in their interactions and
what shared moral principles can be constructed
to govern their interactions.

Global Ethics and the Information

Revolution
Some authors have argued that globalization and
the emergence of the Internet have created a global
community, and that this community requires its
own moral system that transcends and unifies the
moral systems of all cultures and nations that
participate in this global community. The ethics needed for the construction of such a moral
system has been called global ethics. The idea of
a global ethics or ethic was first introduced by
German theologian Hans Kng in 1990 and later

Is Information Ethics Culture-Relative?

elaborated by him in a book (Kng, 2001). His

aim was to work towards a shared moral framework for humanity that would contain a minimal
consensus concerning binding values and moral
principles that could be invoked by members of
a global community in order to overcome differences and avoid conflict.
Krystyna Grniak-Kocikowska (1996) has
argued that the computer revolution that has taken
place has made it clear that a future global ethic
will have to be a computer ethic or information
ethic. As she explains, actions in cyberspace are
not local, and therefore the ethical rules governing such actions cannot be rooted in a particular
local culture. Therefore, unifying ethical rules
have to be constructed in cyberspace that can
serve as a new global ethic. Similar arguments
have been presented by Bao and Xiang (2006)
and De George (2006).
No one would deny that a global ethic, as
proposed by Kng, would be desirable. The construction of an explicit, shared moral framework
that would bind all nations and cultures would
evidently be immensely valuable. It should be
obvious, however, that such a framework could
only develop as an addition to existing local
moral systems, not as a replacement of them.
It would be a framework designed to help solve
global problems, and would exist next to the local moral systems that people use to solve their
local problems. In addition, it remains to be seen
if cross-cultural interactions over the Internet
yield more than a mere set of rules for conduct
online, a global netiquette, and will result in a
global ethic that can serve as a common moral
framework for intercultural dialogue and joint
action. Hongladarom (2001) has concluded, based
on empirical studies, that the Internet does not
create a worldwide monolithic culture but rather
reduplicates existing cultural boundaries. It does
create an umbrella cosmopolitan culture to some
extent, but only for those Internet users who engage

in cross-cultural dialogue, which is a minority,

and this umbrella culture is rather superficial.
Claims that the Internet will enable a new global
ethic may therefore be somewhat premature. In
any case, such intercultural dialogue online will
have to be supplemented with serious academic
work in intercultural information ethics, as well
as intercultural ethics at large.

Conclusion
It was found in this essay that very different moral
attitudes exist in Western and non-Western countries regarding three key issues in information ethics: privacy, intellectual property, and freedom of
information. In non-Western countries like China,
Japan, and Thailand, there is no strong recognition
of individual rights in relation to these three issues.
These differences were analyzed in the context
of a difference, proposed by philosopher David
Wong, between rights-centered moralities that
dominate in the West and virtue-centered moralities that prevail in traditional cultures, including
those in South and East Asia. It was then argued
that cross-cultural normative ethics cannot be
practiced without a thorough understanding of
the prevailing moral system in the culture that is
being addressed. When such an understanding
has been attained, scholars can proceed to engage
in moral criticism of practices in the culture and
propose standards and solutions to moral problems. It was argued, following Rafael Capurro,
that we need an intercultural information ethics
that engages in interpretive, comparative, and
normative studies of moral problems and issues
in information ethics in different cultures. It is
to be hoped that researchers in both Western and
non-Western countries will take up this challenge
and engage in collaborative studies and dialogue
on an issue that may be of key importance to
future international relations.

165

Is Information Ethics Culture-Relative?

References
Bao, X., & Xiang, Y. (2006). Digitalization and
global ethics. Ethics and Information Technology, 8, 41-47.
Capurro, R. (2005). Privacy: An intercultural
perspective. Ethics and Information Technology,
7(1), 37-47.
Capurro, R. (in press). Intercultural information
ethics. In R. Capurro, J. Frhbaure, & T. Hausmanningers (Eds.), Localizing the Internet. Ethical issues in intercultural perspective. Munich:
Fink Verlag. Retrieved January 25, 2007, from
http://www.capurro.de/iie.html
De George, R. (2006). Information technology,
globalization and ethics. Ethics and Information
Technology, 8, 2940.
Ess, C. (2002).
Computer-mediated colonization,
the renaissance, and educational imperatives for
an intercultural global village. Ethics and Information Technology, 4(1), 11-22.
Gorniak-Kocikowska, K. (1996). The computer
revolution and the problem of global ethics. Science and Engineering Ethics, 2, 177190.
Harman, G. (1996). Moral relativism. In G. Harman & J. J. Thompson (Eds.), Moral relativism
and moral objectivity (pp. 3-64). Cambridge, MA:
Blackwell Publishers.
Harman, G. (2000). Is there a single true morality?
In G. Harman (Ed.), Explaining value: And other
essays in moral philosophy (pp. 77-99). Oxford:
Clarendon Press.
Hofstede, G. (2001). Cultures consequences.
Thousand Oaks, CA: Sage.
Hongladarom, S. (2001). Global culture, local
cultures and the Internet: The Thai example. In
C. Ess (Ed.), Culture, technology, communication: Towards an intercultural global village (pp.

166

307-324). Albany, NY: State University of New

York Press.
Human Rights Watch. (2006). Race to the bottom. Corporate complicity in Chinese Internet
censorship. Human Rights Watch Report, 18(8C).
Retrieved January 25, 2007, from http://www.
hrw.org
Johnson, D. (2000). Computer ethics (3rd ed.).
Upper Saddle River, NJ: Prentice Hall.
Jingchun, C. (2005). Protecting the right to privacy
in China. Victoria University of Wellington Law
Review, 38(3). Retrieved January 25, 2007, from
http://www.austlii.edu.au/nz/journals/VUWLRev/2005/25.html
Kitiyadisai, K. (2005). Privacy rights and protection: Foreign values in modern Thai context. Ethics
and Information Technology, 7, 17-26.
Kng, H. (2001). A global ethic for global politics
and economics. Hong Kong: Logos and Pneuma
Press.
L, Y.-H. (2005). Privacy and data privacy issues
in contemporary China. Ethics and Information
Technology, 7, 7-15.
McDougall, B., & Hansson, A. (Eds.). (2002).
Chinese concepts of privacy. Brill Academic
Publishers.
Mizutani, M., Dorsey, J., & Moor, J. (2004). The
Internet and Japanese conception of privacy. Ethics and Information Technology, 6(2), 121-128.
Nakada, M., & Tamura, T. (2005). Japanese conceptions of privacy: An intercultural perspective.
Ethics and Information Technology, 7, 27-36.
Wong, D. (1984). Moral relativity. Berkeley, CA:
University of California Press.
Wong, D. (1993). Relativism. In P. Singer (Ed.), A
companion to ethics (pp. 442-450). Blackwell.

Is Information Ethics Culture-Relative?

Wong, D. (2006). Natural moralities: A defense

of pluralistic relativism. Oxford: Oxford University Press.

Endnotes
By information ethics I mean the study of ethical
issues in the use of information and information
technology. Contemporary information ethics is
a result of the digital revolution (or information
revolution) and focuses mainly on ethical issues
in the production, use, and dissemination of
digital information and information technologies.
It encloses the field of computer ethics (Johnson,
2000) as well as concerns that belong to classical

information ethics (which was a branch of library

and information science), media ethics, and
journalism ethics.
2
This doctrine is called metaethical rather than
normative because it does not make any normative
claims, but rather makes claims about the nature
of moral judgments. Normative moral relativism
would be the thesis that it is morally wrong to
judge or interfere with the moral practices of
societies, groups, cultures, or individuals who have
moral values different from ones own. This is a
normative thesis because it makes prescriptions
for behavior.
3
Worthen, B. (2006). Intellectual property: Chinas
three realities. CIO Blogs. Online at http://blogs.
cio.com/intellectual_property_chinas_three_
realities. Accessed October 2006.

This work was previously published in International Journal of Technology and Human Interaction, Vol. 3, Issue 3, edited by B.
C. Stahl, pp. 12-24, copyright 2007 by IGI Publishing, formerly known as Idea Group Publishing (an imprint of IGI Global).

167

168

Chapter 1.13

Personalization Techniques and

Their Application
Juergen Anke
Dresden University of Technology, Germany
David Sundaram
University of Auckland, New Zealand

Introduction
Personalization is an approach to increase the
usability of complex information systems and
present the user with a comprehensible interface
that is tailored to his or her needs and interests.
In this article, we examine general techniques
that are employed to achieve the personalization
of Web sites. This is followed by a presentation
of real-world examples. It will be shown how different levels of personalization can be achieved
by employing the discussed techniques. This
leads finally to a summary of the current state in
personalization technologies and the issues connected with them. The article closes with some
ideas on further research and development, and
a conclusion.
In general, the concept of personalization refers
to the ability of tailoring standardized items to the
needs of individual people. It is originally derived
from the ideas of Pine (1993) who proposed that
companies should move from the paradigms of

standardized products and homogeneous markets

to customizable products that meet the requirements of many different customers. The principle
of mass customization applies to a certain degree
to most car manufacturers and some computer
manufacturers, for example, Dell.
In the digital world of the World Wide Web,
the degree of customization can be much higher
than in the physical world. Currently, a number
of online portals and e-commerce shops make
use of personalization to provide a better user
experience. Although Web sites may be the most
popular examples of personalization, the concept
is not limited to the Web. Every information system
that deals with large amounts of data and/or has
a heterogeneous group of users can benefit from
it. Examples include e-learning environments,
electronic books, computer-operated voice and
telephony services, and tourist guides.
Personalization is also very useful for mobile
devices like personal digital assistants (PDAs) or
mobile phones (cf, Mulvenna, Anand, & Buchner,

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Personalization Techniques and Their Application

2000). Technologies like mobile Internet access,

WAP (Wireless Application Protocol), and future
multimedia applications based on high-capacity
wireless technologies require the designers of
services for these devices to deal with limited
input capabilities and small display sizes. For
that reason, every method that assists the user in
navigating and finding information easily adds
real value to applications for such devices.

Personalization Techniques
The core idea of personalization is to customize
the presentation of information specifically to the
user to make user interfaces more intuitive and
easier to understand, and to reduce information
overload.
The main areas of tailoring presentation to
individual users are content and navigation. Content refers to the information being displayed, and
navigation refers to the structure of the links that
allow the user to move from one page to another.
Personalized navigation can help the user to easily
find what he or she is looking for or to discover
new information. For example, a system discussed
by Belkin (2000) assists users in refining search
queries by giving recommendations on related
or similar terms.

Adaptable vs. Adaptive

There are two approaches to achieve personalization: adaptable and adaptive methods. The former
is a term for systems that can be customized by
the user in an explicit manner; that is, the user
can change the content, layout, appearance, and
so forth to his or her needs. This data is called
a user profile, and all personalized presentation
is based on data the user provided for configuration purposes. It is important to note that the
customized appearance does not change over
time until the user decides to change his or her
preferences.

In contrast, adaptive methods change the presentation implicitly by using secondary data. This
data can be obtained from a variety of sources,
for example, from the users actions, from the
behaviour of other users on that site, or based on
the currently displayed content. Methods that use
this data as input are discussed in detail below.
The most distinctive characteristic of adaptive
methods is that they are constantly monitoring
the users activities to adjust the arrangement and
selection of relevant information.
Adaptive methods or machine-learning algorithms are huge steps toward automated customization. Current static interfaces suffer from the
fact that the designer has to anticipate the needs,
interests, and previous knowledge of the users in
advance. As these preferences change over time,
customization that requires human interaction
for collecting and identifying preferences leads
quickly to outdated user profiles.
Table 1 shows how adaptive and adaptable
methods can be applied to customize content and
navigation. The examples given are intended to
be generic; more concrete examples are examined
in the case studies below.

Degree of Personalization
Another important criterion for classification is
the degree of personalization. Systems can have
transient or persistent personalization, or be nonpersonalized. With transient personalization, the
customization remains temporary and is largely
based on a combination of the users navigation
and an item-to-item correlation. For example, if an
item is selected, the system attaches similar items
as recommendations to it whereby the content of
the shopping cart is taken into consideration.
Persistent personalization systems maintain a
permanent user account for every user to preserve
his or her settings and preferences across separate
sessions. Although this raises privacy issues and
is the most difficult to implement, it offers the
greatest benefit. These systems can make use

169

Personalization Techniques and Their Application

Table 1. Application of adaptable and adaptive methods to content and navigation

of user-to-user correlation algorithms and thus

provide higher accuracy.
Another technology that belongs to the broad
area of personalization is recommender systems
(Mulvenna et al., 2000). Whereas straight personalization tailors just the presentation of information, recommender systems support the user in
discovering new information. As recommendation
relies on user preferences and interests, it is often
part of personalized systems. From another perspective, one can say that recommender systems
provide a selection of the most suitable content
for the user. The application of recommender
systems to e-commerce is discussed by Schafer,
Konstan, and Riedl (2001).
In the following sections, we look at two
cases that highlight the adaptive and adaptable
approaches to personalization.

Application and Impact of

Personalization: Amazon.COM
Amazon.com is one of the pioneers of e-commerce. Originally set up as a bookstore, it has
grown to a general retailer for a wide range of

170

products. It also provides an auction platform and a

marketplace where customers can sell used goods.
The marketplace is seamlessly integrated into
the main product catalogue, therefore customers
can decide whether they want to buy a particular
product as a new or a used one.

Goal
As the Amazon.com product catalogue contains
more than 2 million products, users can easily get
frustrated if they do not find what they are looking for. Thus, one of the main goals is to tailor
the product catalogue as much as possible to the
needs and interests of the user.
Aside from easy navigation, the site offers a
seamlessly integrated recommendation system. It
is intended to offer customers products that are
either related to their interests or to the product
that is currently displayed to exploit cross-selling
potentials.

Personalization Techniques
Amazon.com is a highly developed online shopping site and incorporates a combination of

Personalization Techniques and Their Application

numerous adaptive and adaptable methods. The

generated user profile is stored in a database on
the server; that is, the degree of personalization
is persistent.
The prevalent recommendation method is
based on the purchases of other customers. It
appears as a list beginning with the phrase Customers who bought this book also bought on
each product detail page (Figure 1a). A second
list contains up to five authors whose books
were bought by customers who also bought the
currently selected book. Both of these features
are collaborative-filtering methods of implicit
collected data. Another feature called purchase
circles is also a collaborative-filtering mechanism
that displays the top-10 products for a particular
region, institution, or company.
Other recommendation methods are based on
the activities of the customer. For example, there is
always one product recommendation in a sidebar
that is apparently directly derived from the list
of recently viewed items (Figure 1c). Moreover,
there is also a recommendation page for each of
the main categories (books, software, etc.) that
is completely based on the customers purchase
history.
Amazon.com also encourages customers to
rate their purchased products (Figure 1e). Ratings
belong to the collaborative-filtering methods.
However, in contrast to the list of products other
customers bought, they require explicit interaction
for the sake of personalization. The ratings can
also be part of customer reviews and comments.
Interestingly, it is even possible to rate the ratings;
that is, customers can indicate whether they found
the rating helpful or not (e.g., 3 of 5 people found
the following review helpful). This mechanism
is used to display the most helpful comments
first and let the least helpful comments move to
secondary pages.
As mentioned before, one of the goals is to
exploit cross-selling potentials. One of the recent
additions to Amazon.com is the Great Buy
feature. It offers the customer the current product

together with another product in a bundle at a

special price (Figure 1b). The two products must
have a complementary relationship to be valuable
to the customer.

Outcome and Lessons Learned

Amazon.com has clearly utilized a very broad
range of different personalization methods. As
this site has developed constantly in product
variety, functionality, and comfort, it is nearly at
the state of the art in this area. The goals stated
above are nearly completely fulfilled. The site is
easy to navigate, the products are easy to find,
and the accuracy of the recommendations seems
to be very high, which animates the customer to
buy further products.
The Great Buy feature is certainly one of the
best ways to take advantage of the cross-selling
potential, whereas the list of products other customers bought is more useful for the discovery of
new, interesting items and navigation. The latter
seems to be less accurate compared to the earlier
days of the site. This might have something to do
with the enormous number of customer profiles
that do not provide enough distinctive attributes
to form useful clusters.
Ratings and reviews can be considered as a
helpful feature; however, there are a number of
relatively unqualified comments. To improve
the situation, the rate-the-rating feature was
introduced (far after the review function itself).
While this highlights the more valuable reviews,
there is still room for improvement.

Application and Impact of

Personalization: Yahoo!
Yahoo! was one of the first search portals on the
Web and one of the first Web sites that applied
personalization on a larger scale (Manber, Patel, &
Robinson, 2000). In 1996, the My Yahoo! service
was introduced. It allows setting up a personal

171

Personalization Techniques and Their Application

Figure 1. Amazon.com detail page for a book

(a) Similar products

(c) Recommendation based on recently

viewed items

(d) Aggregated rating of

other customers
(e) Own rating

version of Yahoo! for every user. Not only the

content, but also the layout and the appearance of
the page can be modified. Yahoo! is a completely
adaptable system; therefore, all personalization
is based on the data the user entered beforehand.
The ZIP code is especially central as a lot of
personalized features rely on it. The intelligence
of the system lies in the ability to use this data
in different situations to tailor the presentation
specifically to the user.

172

Goal
The goal of Yahoo! is to bind its users by differentiating from other Web catalogues and search
engines, and to provide a fully customizable and
integrated portal. As the customer structure of
Yahoo! is very heterogeneous, it is a good idea
to offer personalization and let users construct an
individual start page. Yahoo!s service is free for

Personalization Techniques and Their Application

its users; money is mainly earned with advertising and revenue provisions of shopping partners.
Thus, the second goal is to make advertising as
effective as possible. This can be achieved by
selecting banner ads that are likely to be of interest for the user.

Personalization Techniques
Yahoo! offers an adaptable system that requires
the user to explicitly provide information for
personalization. The user profile is kept on a
server between different visits, thus Yahoo! offers
persistent personalization.
My Yahoo! enables registered users to build
their own Yahoo! pages. The content is selected as
so-called modules. Among the available modules
are ones for weather, news, sports results, stock
quotes, horoscopes, movie reviews, personal news
filters, and many more. Further configuration can

be done within these modules. In the example

shown in Figure 2, the headline module (a) is set
up to show world news from British news agencies and German Formula 1 news. The weather
module (b) displays the current weather situation
of selected cities only. Modules can be edited
or deleted (c) directly on the page. Some of the
modules offer an individual default setting that is
based on the user profile. For example, when the
sports-results module is added, it already contains
the results of teams in the users area.
Not only the content, but also the layout is
customizable. The chosen modules can be distributed on multiple pages that in turn consist
of two or three columns where modules can be
ordered arbitrarily. There are also options like
colour sets and themes to change the appearance
of My Yahoo!.

Figure 2. A customized version of the My Yahoo! start page

(d) Yahoo! Companion toolbar

(b) Weather module

(c) Delete
or configure
module

(a) Headline
module

173

Personalization Techniques and Their Application

Outcome and Lessons Learned

Conclusion

The goal of flexible customization to provide an

integrated portal can be considered to be fulfilled. Whether the effectiveness of advertising
is increased by personalization cannot be decided
with certainty since most of the literature does
not mention it. However, as it is vital for Yahoo!
to have effective advertising, it can be assumed
that it incorporates a relatively advanced system
for selecting banner advertisements on an adaptive basis.
Apart from these sophisticated personalization
features, it is also essential to design high-quality default pages for people who do not want to
customize at all. It turned out that only a small
number of users actually customize Yahoo!; most
of them take what is offered. The reasons for this
may be either that the personalization tools are
too complicated, or the users do not need complex
personalization as the default page is satisfactory
for them. Addressing all types of users also includes not requiring users to enter any personal
data and not forcing them to use personalization
features.
Yahoo! has decided not to use adaptive methods. They believe that these methods are still not
good enough for such a complex site as a users
behaviour is not sufficiently predictable. People
must not be unsure of how the systems work;
otherwise, it prevents them from experimenting
as they fear breaking something. The people from
Yahoo! reckon that any kind of personalization
should encourage the users to experiment.
While it is beyond the scope of this article to
look at the numerous other cases of personalization, it is worthwhile to mention the large-scale
personalization project at ibm.com. Karat, Brodie,
Karat, Vergo, and Alpert (2003) describe this in
detail, and they provide an overview of the many
different personalization techniques used and the
reasons for selecting them for the project.

Current State of Development

174

A number of different approaches to personalization are currently used in various applications.

Highly sophisticated e-commerce sites usually
employ a combination of multiple methods and
contain adaptive as well as adaptable elements
in a coherent user interface. As the case study
of Amazon.com has shown, organisations can
add real value for the users by making it more
convenient to tailor the presentation to individual
needs.
Adaptive methods for personalization are
very powerful means to manage information
overload and simplify user interfaces. Despite
their sophisticated algorithms, they can produce
unpredictable and unwanted results. This is an
area where further research is necessary as adaptable methods have the serious disadvantage of
their configuration remaining static until the user
changes it explicitly. This can be a tedious task
for most users, which lowers the value even of
the most flexible customization services. Hence,
there are only a small number of users who actually want to customize their pages.

Directions for Future Research and

Development
Improve Group-Based Personalization
The personalization based on a users preferences
is increasingly being extended by collaborative
methods. Instead of recommending items based
on the profile of a single user, the system should
try to take advantage of other users ratings and
preferences as well. While this might in fact increase the accuracy of predictions, it raises the
issue of proper user selection.

Personalization Techniques and Their Application

New Input Methods

Methods that allow formulating queries in a natural language rather than in special inhuman query
syntax could make the interaction between the
user and the system even more personalized and
individualistic (Zadrozny, Budzikowska, Chai,
Kambhatla, Levesque, & Nicolov, 2000).

Combination of Explicit and Implicit

Data
Both adaptive and adaptable methods have their
strengths and weaknesses, and may be more or less
applicable for a particular situation. However, it
seems to turn out that the combination of explicit
and implicit user data provides the best results.
On one hand, the effort of manual customization
is minimized, and on the other hand, an adaptive
method will not cause much unpredicted results
when it is limited by explicit statements. Wells
and Wolfers (2000) explain how customers who
use online banking services need to make some
basic statements about their financial goals and
situations. After that, adaptive methods are used
to offer financial products and services that are
tailored to the customer and suitable for his or her
particular needs. However, not all techniques applicable to a certain scenario will be successful in
practice. Alpert, Karat, Karat, Brodie, and Vergo
(2003) show in their study about users attitudes
toward adaptive systems that users have a strong
desire to always be in full control of all interaction. It is therefore important to carefully analyze
the potential acceptance barriers of a designed
solution before finally deploying it.

Future Outlook
Personalization technologies have found their way
out of experimental systems of researchers into
commercial applications. They are a powerful
means to handle information overload, to make
complex information systems more usable for a

heterogeneous group of people, and to help online

businesses establish personal relations with their
customers (one-to-one marketing). Although we
have focused on applications of personalization
to Web sites, they can be used in a wide range of
human-computer interactions. Personalization
techniques are the key to mass customization
and provide people with a much more individual
experience instead of standardized services.

REFERENCES
Alpert, S. R., Karat, J., Karat, C., Brodie, C., &
Vergo, J. G. (2003). User attitudes regarding a
user-adaptive ecommerce Web site. User Modeling and User-Adapted Interaction, 13(4), 1-2.
Belkin, N. J. (2000). Helping people find what
they dont know. Communications of the ACM,
43(8), 58-61.
Hirsh, H., Basu, C., & Davison, B. D. (2000).
Learning to personalize. Communications of the
ACM, 43(8), 102-106.
Kantor, P. B., Boros, E., Melamed, B., Mekov,
V., Shapira, B., & Neu, D. J. (2000). Capturing
human intelligence in the Net. Communications
of the ACM, 43(8), 112-115.
Karat, C. M., Brodie, C., Karat, J., Vergo, J. G.,
& Alpert, S. R. (2003). Personalizing the user
experience on ibm.com. IBM Systems Journal,
42(4), 686-701.
Manber, U., Patel, A., & Robinson, J. (2000).
Experience with personalization on Yahoo! Communications of the ACM, 43(8), 35-39.
McCarthy, J. (2000). Phenomenal data mining.
Communications of the ACM, 43(8), 75-79.
Mobasher, B., Cooley, R., & Srivastava, J. (2000).
Automatic personalization based on Web usage
mining. Communications of the ACM, 43(8),
142-151.

175

Personalization Techniques and Their Application

Mulvenna, M. D., Anand, S. S., & Buchner, A.

G. (2000). Personalization on the Net using Web
mining. Communications of the ACM, 43(8),
122-125.
OConnor, M., Cosley, D., Konstan, J. A., & Riedl,
J. (2001). PolyLens: A recommender system for
groups of users. Proceedings of ECSCW 2001,
Bonn, Germany.
Pine, B. J., II. (1993). Mass customization. Boston:
Harvard Business School Press.
Riecken, D. (2000). Personalized views of personalization. Communications of the ACM, 43(8),
26-28.
Schafer, J. B., Konstan, J. A., & Riedl, J. (2001).
E-commerce recommendation applications. Journal of Data Mining and Knowledge Discovery,
5(1/2), 115-152.
Wells, N., & Wolfers, J. (2000). Finance with a
personalized touch. Communications of the ACM,
43(8), 30-34.
Zadrozny, W., Budzikowska, M., Chai, J., Kambhatla, N., Levesque, S., & Nicolov, N. (2000).
Natural language dialogue for personalized
interaction. Communications of the ACM, 43(8),
116-120.

Adaptive Personalization Systems: These

change the presentation implicitly by using secondary data. This data can be obtained from a
variety of sources, for example, from the users
actions, from the behaviour of other users on
that site, or based on the currently displayed
content.
Decision-Support Systems and Tools: In a
wider sense, it can be defined as systems and tools
that affect the way people make decisions.
Mass Customization: The customization of
products and services for individual customers,
but at a mass-production price.
Personalization: An approach to increase the
usability of complex information systems and
present the user with a comprehensible interface
that is tailored to his or her needs and interests.
Recommender Systems: A special type of
decision-support system that gives recommendations for further actions or related items.
Tailoring: In the context of personalization,
it can be with respect to content or navigation.
Content refers to the information being displayed,
and navigation refers to the structure of the links
that allow the user to move from one page to
another.

KEY TERMS
Adaptable Personalization Systems: Systems that can be customized by the user in an
explicit manner; that is, the user can change the
content, layout, appearance, and so forth to his
or her needs.

This work was previously published in Encyclopedia of E-Commerce, E-Government, and Mobile Commerce, edited by M.
Khosrow-Pour, pp. 919-925, copyright 2006 by Information Science Reference, formerly known as Idea Group Reference (an
imprint of IGI Global).

176

177

Chapter 1.14

User-Centered Evaluation of
Personalized Web Sites:
Whats Unique?

Sherman R. Alpert
IBM T. J. Watson Research Center, USA
John G. Vergo
IBM T. J. Watson Research Center, USA

Abstract
In addition to traditional usability issues, evaluation studies for personalized Web sites and
applications must consider concerns specific
to these systems. In the general case, usability
studies for computer-based applications attempt
to determine whether the software, in actual use,
meets users needs; whether users can accomplish
their goals in using the software; whether users
can understand and use the application (whether
they comprehend what they can do and how); the
rate, frequency, and severity of user errors; the
rate of and time duration for task completion; and
so on. But in the case of user-centered evaluations
of personalized Web sites, there are additional
questions and issues that must be addressed. In
this paper, we present some of these, based on our

experience in usability studies of a personalized

e-commerce site.

INTRODUCTION
Personalized Web sites attempt to adapt and
tailor the user experience to a particular users
preferences, needs, goals, interests, knowledge,
or interaction history. A personalized site adapts
its content, content structure, the presentation of
information, the inclusion of hyperlinks, or the
availability of functionality to each individual
users characteristics and/or usage behavior. Such
a site may place specific information, which
it thinks you will be interested in, at a distinguished or obvious location on a Web page.
Another personalized site may choose to add or

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

User-Centered Evaluation of Personalized Web Sites

elide specific content or hyperlinks to additional

information based on what it knows about the
current users knowledge or interests. An e-commerce site that knows what model laptop you own
may only show accessories compatible with that
model. A site that displays information about
movies and theater schedules may use knowledge
of the users postal code to display only theaters
within n miles of the users location. A personalized news site may elect to show (or not) todays
baseball scores, depending on whether the user
has viewed this sort of information in previous
site visits. A book seller may use knowledge of the
books you have ordered in the past to recommend
new works by the same author or other authors of
the same genre, or may suggest additional books
purchased by other users that have bought the same
book as you are now ordering. Data about the user,
used to drive the sites personalizations, may be
obtained by information explicitly provided by
the user and by inferences made by the system
based on previous user interactions.
The personalization approach begs many
questions, Do personalized Web sites actually
improve the users experience when using such
sites? Do specific personalization features improve
and others detract from user experience? Does
personalization actually add value to users? Is
the site not only usable but acceptable, attractive,
and desirable to users?
Personalized Web sites are a specific example
of the more general field of adaptive systems. The
literature of the evaluation of adaptive systems
is replete with evaluative studies of how well the
system works. These evaluations have focused
on algorithms and user model representations
for programmatically implementing the systems adaptive behavior, including determining
how well the detection and gathering of implicit
information about users functions, how appropriately are inferences drawn about users,
and how robust are the systems techniques for
using such information to provide some type of
adaptive functionality. For example, evaluations

178

of adaptive systems might consider whether the

systems inferences about the user indeed coincide with the users prior behavior (Weibelzahl
& Weber, 2003). As another example, evaluators
need to check if [the systems] inferences or the
conclusions drawn by the system concerning the
user-computer interaction are correct since it is
not necessary that there will be a direct one to
one mapping between raw data and their semantically meaningful counterparts (Gupta & Grover,
2004). Thus many adaptive system evaluations
focus on how well the system functions in an
(objective) application-centered sense. Many
such studies focus on an individual personalization technique, such as recommender systems or
collaborative filtering (e.g., Mobasher, Dai, Luo,
& Nakagawa, 2001; Zhu & Greiner, 2005). Still
others have focused on success of a personalized
site as measured by the number of site visits and
return visits, number of purchases on an e-commerce site, click-throughs to suggested content,
and so forth.
Of course, many of these measures are useful and must be considered in the evaluation of a
personalized site. However, evaluations of personalized Web sites must also consider the more
subjective user-centered perspective, and the literature is considerably sparser in this regard. User
satisfaction is only partially determined by the
accuracy of the algorithmic implementation and,
further, user satisfaction may not be achieved even
in systems that do provide accurate information
(Swearingen & Sinha, 2001). In a user-centered
design approach, design decisions are based on
the experimentally validated value to users of
a systems features and facilities (Vredenburg,
Isensee, & Righi, 2001). Thus user-centered evaluations must involve the testing of the system by
(and for) users. Some of the existing evaluation
literature has suggested using an evaluator who
attempts to take the role of a typical user of
the system; but we have learned from experience
(Alpert, Karat, Karat, Brodie, & Vergo, 2003;
Karat, Brodie, Karat, Vergo, & Alpert, 2003)

User-Centered Evaluation of Personalized Web Sites

that testing must involve actual intended users

of the system because doing so may elicit results
that are unexpected based solely on the system
developers analysis, and reveals more accurately
the real users perspective.
In the general case, usability studies for computer-based applications attempt to determine
whether the software, in actual use, meets users
requirements; whether the software performs as
it should in supporting users as they attempt to
accomplish specific tasks; whether typical users
are successful in achieving their goals; whether
users can understand and use the application
(whether they comprehend what they can do and
how); the rate, frequency, and severity of user
errors; the rate of and time duration for task
completion; and so on. The goal is to test how
well the functionality fits user needs, how well
the flow through the application fits user tasks,
and how well the response of the application fits
user expectations (Usability First, n.d.). Usability
testing may involve such abstract issues as the appropriateness of an overall system metaphor and
such low level details as the placement of widgets
on the screen, the use of color, and the wording
of textual content. Usability testing informs the
design of interactive software by obtaining the
users perspective regarding the design, rather than
simply relying on the intuitions of the designers
and implementers of the software.
Even when researchers have spoken directly
to the idea of evaluating the usability of adaptive or personalized systems, they have not
addressed the entire problem: Such discussions
often revolve around traditional usability issues
or around Boolean comparisons of a whole system
with and without its entire adaptive functionality (Weibelzahl, Lippitsch, & Weber, 2002). Of
course, standard usability issues must be addressed when evaluating adaptive systems: We
do care whether users can accomplish their goals
using the system, and further, whether they can
accomplish their goals faster or more accurately
using a system that adapts to their needs than

when using a system that does not. And usability

evaluations and system evaluations will have interdependencies: If the systems personalization
mechanisms simply do not function well, the user
will lose confidence in the system and all of its
personalized recommendations (McLaughlin &
Herlocke, 2004). But there are other factors that
must be considered, issues that go to the heart of
the user experience.
As in the design of any software, developers
of personalized solutions begin with what they
consider to be good ideas regarding the sorts
of adaptations that would serve users, using intuitions about users and the softwares functionality. And as in case of other interactive software,
these intuitions must again be verified by contact
with actual users at some time before application
deployment. In the case of personalized applications, however, additional questions that go beyond
traditional usability must be addressed by usercentered evaluative studies. The functionality
offered by personalized applications must not
only match users needs and, perhaps even more
importantly than before, their expectations, but
also their desires and level of trust in computational systems. User studies in this domain must
not only assess ostensibly quantitative measures
such as time-to-task completion, but qualitative
issues such as users confidence and belief in the
systems recommendations, personalizations, and
other adaptive behaviors. We must determine not
only whether the system is usable for performing
tasks, but also whether it is acceptable to users
and enhances, rather than degrades, the user experience. In this paper, we discuss some of these
issues that go beyond the purview of traditional
usability evaluations.
The issues discussed in this paper are derived
from user studies of personalized Web sites in a
particular e-commerce context, specifically the
Web site of a large computer hardware, software,
and services provider (Alpert et al., 2003; Karat
et al., 2003). As such, the issues themselves are
derived from empirical evaluations. They may be

179

User-Centered Evaluation of Personalized Web Sites

considered a step moving toward the full realization of the issues and factors that must be addressed
in personalized Web site evaluations.

USER-CENTERED
PERSONALIZATION MEASURES
As mentioned previously, there are many proposals in the adaptive systems literature aimed at
evaluating whether the application or Web site
in question performs its adaptations correctly or
accurately. And usability studies of (nonpersonalized) Web sites have shown that ease of use can
increase the number of revisits and purchases
(e.g., Nielsen, 2003). We touch only gently on
general usability issues here. Instead, this discussion focuses on the users views and opinions of
personalized adaptations, not only whether they
work as intended, but even if they do so, whether
users want the Web site to be making and using
inferences and data about the user to influence
or direct an adaptive presentation to the user.
Evaluations of adaptive and personalized applications must ultimately address the question, do
the adaptive features actually improve the users
experience when using the site? (see also Chin,
2001). Designers, researchers, and developers
may have many ideas for personalized functionality for a Web site that they think would provide
users with some benefit. But actual users when
confronted with such features may find them
useless or, worse, objectionable. The intuitions of
the builders of (personalized and all) interactive
software must be confirmed by actual potential
users of that software.
Here we introduce some of the questions and
issues that must be addressed when performing
user-centered evaluations of personalized Web
sites. These must be addressed in addition to
traditional usability concerns, which we will not
discuss but that, of course, must be incorporated
into the user-centered evaluation. For example,
fundamental to any evaluation is whether the site

180

(and its adaptive behaviors) supports users in accomplishing their goals. Or, even more specifically
related to personalization, does the inclusion of
personalized features make the user more efficient and decrease time-to-goal completion.
These are important, and ignoring such issues
would be foolish. In this paper, we only touch
on these more traditional concerns but go further
in discussing issues that are of a more subjective
nature and relate to the overall user experience,
not simply quantitative efficiency measures such
as time-to-task completion.

Do Users Want the Personalized

Behavior?
Beyond whether site adaptations support the user
in accomplishing tasks and goals, the next question we must ask in a user-centered analysis of a
personalized site is whether users actually desire
the personalizations the site intends or purports to
provide. For example, Alpert et al. (2003) found
many users were not pleased with a sites attempts
to infer their needs, goals, or interests to thereby
provide user-specific adaptive content. A number
of study participants declared they would prefer
that the site not attempt to infer their intentions,
declaring their opinion that computers are not
smart enough to do a good job of inferring users goals and plans, and therefore personalized
interactions may be based on specious inferences.
Users say things like dont stereotype me -- just
give me the options because I prefer choosing for
myself rather than having the computer tell me
whats good for me (Nielsen, 1998).
One type of adaptive system behavior is to
intervene while a user is working. The intervention
might be to offer advice relating to the users plan
for achieving particular goals; to offer just-in-time
instruction related to information currently in
view; to offer additional, related information or
an alternative or associated product; and so on. In
a user-centered view of the system, the question
is, is an intervention wanted or timed correctly?

User-Centered Evaluation of Personalized Web Sites

For example, Clippy is the assistive agent in

Microsoft Office applications (who by default
appears in the form of an interactive paper clip),
who offers advice on how to accomplish goals
while using the application. The question is,
should Clippy intervene while users are working,
and when? Overwhelmingly, at least anecdotally,
users dislike the assistive agent and its interventions, and many users disable the advice-giving
assistance. Problems here include not wishing
to be interrupted while focusing on actual work.
This can be considered the annoyance factor
and may be based on the cognitive load of having to interrupt a plan to focus on advice for an
alternative plan. Perhaps more important is the
fact that the assistant often does a poor job of
inferring the users goals and plans. Thus, not
only is the interruption unwanted, but the advice
is not useful anyway! Due to the extensive use of
these application products, many users are wary
of computer applications attempts to infer their
needs and simply do not trust them to be smart
enough to do a good job at it.

Questions Regarding the Use of

Prior User Behavior for Adaptive
Site Behavior
In personalized Web sites, a user may attempt to
find or explicitly ask for information on a particular topic, and the site uses information explicitly
provided by the user and implicitly inferred about
the user to find and (perhaps adaptively) display
or recommend pointers to the appropriate information. The first question this raises is, what
implicit information does the site use in making
personalization decisions? Where (sites, pages,
topics, products, etc.) has the user browsed before?
Which search hitlist items has the user clicked on
before? Does the site use information about the
users apparent current information needs, for
example, current task and inferred goals; recent
needs; needs some time in the past; general longterm interests; where the user navigated and the

topical or product information the user viewed;

or information gathered about other users? How
does a personalized site adapt to individual users
that is, on what information are the systems
adaptations based?
First, personalized sites typically use information explicitly provided by users about themselves, in either questionnaires or in the course of
performing actual tasks (e.g., the users mail zip
code is obtained when the user makes a purchase;
this can later be used to provide information
personalized by location, such as nearby movie
theater listings). In the case of multi-user systems
such as Web sites, personalized systems may
also use information about other users, including
their behavior on the site (e.g., for collaborative
filtering, what did other users purchase?). In addition, intelligent personalized sites use implicit
information gathered about the user, based on the
users previous behavior on the site. This includes
information obtained by observing the users
previous navigations (e.g., what topics, products,
genres did the user look at before?) and clickstream
data (e.g., what links did the user click?).
An application-centered evaluation of such
a site might ask questions about how well the
systems algorithms perform, for example, does the
collaborative filtering engine accurately display
the most popular related items? (McLaughlin &
Herlocke, 2004). A user-centered evaluation of
such an application must ask at least two questions. First, does it make sense in the context
of this site to use prior user behavior to decide
what to present to that user? And, do users want
such personalized site features and have it based
on their own previous performance and/or the
previous behavior and decisions of other users?
Personalized content, based on implicit information, such as previous navigation paths, also
met with mixed reactions in our previous studies.
The notion of basing content on previous navigation was rated positively by participants overall,
but some participants were so adamant in their
disapproval of this feature, and their opinions were

181

User-Centered Evaluation of Personalized Web Sites

so clear and passionately articulated, that they

must be considered by developers of personalized
sites. One participant stated the explicit-implicit
problem succinctly: I like to have more control
and less assumptions made. Other participants
expressed skepticism that this feature could be
implemented in a way that would make it generally valuable to them. Specifically, participants
expressed some level of concern about a sites
ability to do this well enough to be useful. The
overarching message was, adapting content
based on past navigation would be a nice thing,
but we dont believe you can do it well, so dont
do it at all.
Clearly, using a particular users previous
behavior is not always useful: I viewed pages
about x before, but I have very different goals and
needs now; dont bias what I can view based on
what I did before. Participants mentioned shifting goals as a problem that they believe would
not be handled well by systems that use inferred
goals to guide current and future presentations.
They asserted that their needs, even on a single
site, change over time, and they do not want
adaptive behavior of the site to be based on their
obsolete goals. As Nielsen (1998) puts it, Having
the computer personalize the website to the user
assumes that the computer can guess the users
needs. This is difficult to do and even more difficult when you consider that the same person may
have different desires at different times.
Refining this result further, adapting content
based on immediate context was met with favor
by our participants. For example, personalized
recommendations can be based on the content of
the Web page that is currently visible. It appears
that using the immediate or current context to
influence concurrent or immediately ensuing
content is seen to be useful, whereas attempting
to infer current goals based on navigation or other
information from the past was not universally
welcome. Overall, participants agreed that there
ought to be a logical limit regarding how far
back into ones past history the site should look

182

for implicit information about goals and interests,

and further that this limit should be the current
session. That is, the consensus view was that a
users past history should have a limited life span:
It is acceptable for the site to adapt content based
on the users current task and navigation context
and even the users context and history since
he/she logged on today, but it should not look at
past behavior beyond this point to a disjoint time
or session in the past.
A further complicating issue relating to the
implicit gathering of information about a users
past interaction history is: Is all navigation
behavior and clickstream data even relevant or
meaningful, shifting goals notwithstanding?
Therefore exactly when should such information
not be collected by the site? Participants in our
studies did not want data collected implicitly
about them and remembered in their profile
during exploratory sessions on a site unless
they specifically authorize it. As one participant
explained, Hey, I might just be knocking around
the site for a while...it doesnt mean anything, and
youll fill up my profile with a lot of junk if you
implicitly collect that information. Wait until I
know that Im really after something before you
start collecting data about what Im doing, and
let me tell you when that is.

Consideration of Context
Some personalization features are based on the
behavior, attitudes, likes and dislikes, navigation,
or purchases of other users. For example, many
existing e-commerce sites recommend additional
related products to the user when a purchase
transaction is about to be completed: People
who bought this product also bought products
X, Y, and Z. This is a simple form of collaborative filtering or recommender technologies (e.g.,
Burke, 1999; Resnick & Varian, 1997; Schafer,
Konstan, & Riedl, 2001). When applied in the
context of e-commerce, these technologies use the
buying behavior of prior customers to attempt to

User-Centered Evaluation of Personalized Web Sites

predict which products a new customer may be

interested in purchasing. But in our user studies,
some participants expressed objections such as, I
am not like other people, I have different needs.
Additionally, this feature might be perceived
simply as an intrusive marketing ploy: One of
our participants found it to be an obnoxious
attempt at marketing more products.
Looking a bit deeper, it appears that user attitudes about collaborative filtering may in fact
be influenced by a variety of factors including the
type of e-commerce site involved, the particular
product being purchased, the type of user, and the
reason an item is being purchased. Our previous
evaluation studies involved a specific class of
user, a particular type of e-commerce site, for
a particular category of products and services:
Users were business managers buying computer
equipment for their enterprises, driven by specific
business requirements in their actual jobs. What
other customers purchased is not interesting or
important to these users. The upshot is that our
users had equivocal reactions to a feature that is
nonetheless in extensive use: Collaborative filtering is clearly viewed by users as a benefit in other
e-commerce settings, such as book and music
sellers like Amazon.com. The clear conclusion
is that individual personalization strategies, such
as recommender systems, are not necessarily effective or beneficial across the spectrum of users
and activities. Instead, user opinions and desires
regarding such technologies may depend on
multiple dimensions of the e-commerce scenario
in question, such as the type of product being
purchased, whether the purchase is for ones self
or for ones company, and whether subjective
parameters such as taste or genre are relevant to
the item being purchased.

Confidence in and Understanding of

the Sites Personalizations
As we can already see, usability measures for
personalized systems are significantly intercon-

nected; For example, if a user does not trust that

the site is capable of making reasonable inferences
regarding his/her current goals, the he/she will
not want the sites (perhaps specious) adaptations.
From the users perspective, then, an important
question is, do users believe computer systems can
do the proposed adaptations well (enough to be
useful)? Are users confident that a site can perform
the personalized adaptation without annoying
them or wasting their time? As we have noted,
in our previous studies we found that the confidence issue is so important that some participants
disclosed that they do not want any information
that has been collected implicitly by the site to
become a part of their persistent personal profile.
Instead, they wanted only information explicitly
provided by them to be remembered by the site: I
want the stored information to be based on what
I told you, not what you think Ive said.
Especially if the site is making decisions regarding the customizing of information shown
to individual users, are users confident the site
is giving them good or the best information
available? When the user knows that a screen
has been adapted to them, it is natural for them
to wonder just what has been adapted to them.
What might a different user see? Are they missing
out on something? (Kay, 2001). These thoughts
may intensify a users lack of confidence in an
individually customized site. In a survey of Web
users, 94% said that being able to trust the information on a site is very important or somewhat
important to them as they decide to (re)visit a Web
site (80% said very important; 80% also said that it
is very important that the site be easy to navigate)
(Consumer Reports WebWatch, 2002).
Confidence in the site also relies heavily on
being able to make sense of the site and its behaviors. In a user-centered analysis of a personalized
site, one ought to ask, does the user understand
what is being adapted due to personalization,
and why? The issue is one of determinism and
comprehension. To navigate through web-sites
(e.g., presentations, online-shops, or learning

183

User-Centered Evaluation of Personalized Web Sites

courses) the user requires a mental model of the

sites structure (Weibelzahl & Weber, 2001, p.
74). It may be difficult to form a coherent model
of the structure, content, and behavior of a site
when these change over time and circumstance
(although Weibelzahl & Weber argue that an
adaptive site may require building a less complex
mental model than a nonadaptive site). A site may
be more difficult to understand and learn to use
when multiple instances of the same user-system
interaction result in different behavior by the system. Does the user understand what is changing
and why it is changing? These are questions that
must be addressed in a user-centered evaluation of
a personalized site. Does the user understand why
the site is adapting its behavior and output? Is
the system predictable: Will the user know, with
a reasonable degree of certainty, what will happen
when (s)he takes an action? (DHertefelt, 2000).
Can the user understand how the site behaves and
is he/she disconcerted by the fact that the content
shown in a specific context is not always the
same? Is the user experience degraded because
the ability of the user to understand and predict
what the site will do is diminished? Paramythis,
Totter, and Stephanidis (2001) refer to this issue
as transparency. Users want to readily be able
to make sense of site behavior, to understand a
sites rationale for displaying particular content.
When past behavior or other nonobvious implicit
information is used to generate content, the users confused reaction Where did this come
from? conveys to users the notion that they are
not in control. On the other hand, content whose
origin is obvious or readily inferable is met with
favor, while content that is based on something
the user, or other users, did at some temporally
distant time is often met with disapproval.
Users seem to have similar expectations of a
dialog with a software application or Web site as
with other people. That is, it appears that users
desire responses from a Web site to follow logical
conversational conventions (Grice, 1975; Searle,
1969); for example, they expect responses to be

184

relevant to the ongoing dialog and to be as informative as required but not more so. This result
coincides with the experimental work of Reeves
and Nass (1999). After conducting numerous
studies, Reeves and Nass concluded that users
expectations regarding their interactions with
computers and other media are based in large part
on social interactions in real life, that is, users
expect computers to obey rules that come from
the world of interpersonal interaction. When a site
provides more information or, in some manner,
other information than what is expected, the
user may not immediately know what to make
of the extra information. In discussing many
features, we heard from our users that they want
to understand, without difficulty, why and how
the computer side of the ongoing bilateral conversation, represented by the content the Web site
displays, chose to say what it has displayed.
Further, does a users need to understand what
is happening vary with the type of personalized
site? If personalized adaptation is based on, say,
previous navigation, the resultant application
behavior may not always be clear to the user.
For example, the simple collaborative filtering
mechanism implemented on the Amazon site is
explained as Customers who bought [this book]
also bought Thus the attempt is made to have
users understand this simple personalization. On
the other hand, when personalized results for a
particular search query result in differing output
at different times, does the user understand why
the results differ? During usability evaluation, we
must also be sure that when a site does explain
its behavior, the explanation is satisfactorily informative from the users perspective.
Andersen, Andersen, and Hansen (2001) assert
that users of adaptive e-commerce sites wish to
be surprised by the site, but in a very different
manner than what is being discussed here. They
mean pleasantly surprised in terms of value-added
services as a reward for using the site. First, they
state that users wish to see recommended products
related to what they are already purchasing. We

User-Centered Evaluation of Personalized Web Sites

saw in our user study that this is dependent on the

e-commerce context: Users of book seller sites
(as in Andersen et al.) may wish to have this feature, whereas in many other product-purchasing
contexts this is not a welcome feature. Andersen
et al. also state that users wish to be surprised
by special offers associated with use of the site
(e.g., lower prices than in brick-and-mortar stores
and flexible delivery options). On the other hand,
users do not wish to be surprised in the conventional sense; if they ask, What color is the sky?
they do not expect a response about shoe sizes in
North America.
Confidence and trust in a site of course also
touch on the broader issue of privacy. The importance of this issue is brought into relief by a
survey in which Web users were asked to rate their
confidence in various types of organizations and
enterprises (businesses, newspapers, banks, charities, government, etc.). Web sites rated among the
lowest, and the overwhelming reason was lack of
trust regarding the handling of private information (Consumer Reports WebWatch, 2002). For
a given site then, usability evaluations must also
determine whether users are willing to divulge
information about themselves to the site. Do users of the site under evaluation believe that the
value added by the sites personalization features
is worth the user-perceived risk in disclosing
private information? Are they confident that the
site will use the personal information in ethical
ways? Is an official privacy statement in which
the site spells out a legally binding description of
how it will use and disseminate (or not) personal
information necessary?

Who is in Control?
In our previous studies regarding personalized
Web sites, the most important issue to users was
their fervent desire to be or at least feel like
they are in control. The feeling of security
experienced by a user of an interactive system is
determined by the users feeling of control of the

interactive system (DHertefelt, 2000). The issue

is one of trust, and again has interdependencies
with other issues, such as confidence and comprehensibility. An interactive system that allows
the user to feel in control should in the first place
be comprehensible (DHertefelt, 2000). But also,
does the user trust that the inferences the personalized site has made about him are correct? People
usually do not know the reasons behind some of the
personal recommendations, which often results in
their distrust of the personalization mechanism
(Wang & Lin, 2003). Is there a way for the user
to understand the kinds of inferences the site is
making? Should there be? Can the user correct
the inferences and other profile information? How
granular is the control that is, is it simply on/off
for all adaptation, for subsets of the adaptation
features? What is the accuracy rate required for
the adaptation to become a net benefit to the user?
If content or features are elided as a result of inferences made about the user, can the user easily
see what they are missing: While users may feel
that at times personalization may be useful, they
also believe the application must provide ways to
view the full, nonpersonalized or nonadapted
content or information (Nielsen, 1998). Can the
user view, and even modify, his/her personalization profile? These are the sorts of questions that
must be addressed in a user-centered evaluation
of a personalized Web site.
In our studies, this control issue emerged most
strikingly in terms of being in control of ones
profile, the persistently stored personal information collected by the site about the user. As noted
previously with regard to confidence and trust in
a site, users ask, What are you going to do with
this data; can I trust you to maintain intimate
information about me? While this result is certainly no surprise, it is prevalent among all users
and is a strongly emotional issue. As such, it can
be considered a deal breaker users will not
provide personal information and will not perform
transactions on a Web site that in their opinion
cannot be fully trusted to use profile information

185

User-Centered Evaluation of Personalized Web Sites

in an ethical and private manner. Andersen et al.

(2001) also discuss this issue. Users want to be
able to review, modify, delete, or add personal
information to their profile at any time. This
should be another focus of usability evaluations
of personalized Web sites, to determine how the
users of the specific site feel regarding these issues and what might be done to ameliorate their
concerns (for example, Kay and her colleagues
have attempted to address these sorts of feeling
in control issues in personalized systems with
scrutable user models and adaptations; see, e.g.,
Kay, 2001; Czarkowski & Kay, 2005).
Of course, there is a delicate balance that
must be maintained between adaptability of the
system and the users sense of control: Increasing
the flexibility can also increase complexity and diminish comprehensibility (DHertefelt, 2000). Thus
the need for actually testing personalized sites with
users is further intensified personalized systems
attempt to balance adaptability versus users sense
of control, and usability testing must be performed
to determine whether the site has succeeded in this
balancing act.

How Much Work is Involved on the

Part of the User?
Web site users are loath to spend extra time
and effort to enter personal information, even
to enable a site to know more about them for
personalization purposes. Nielsen (1998) invokes
the notion of the paradox of the active user to
explain this user behavior and preference. When
encountering a new application, users wish to
immediately engage in actual tasks, tasks that
further their own goals. This is especially true
when visiting Web sites; users have a specific
goal in mind (e.g., purchasing a CD) when they
visit an e-commerce site, and have little desire
to first enter into an extended session of entering
information about themselves before being able to
accomplish that goal. Web users are extremely

186

impatient and want to get something useful out

of a site immediately (Nielsen, 1998).
An alternative to lengthy data-entering sessions is a technique known as permission marketing (Godin, 1999), in which a users profile
or user model is incrementally constructed over
time as the user interacts with a service or product provider, such as an e-commerce site. In this
scenario, customers are asked only for information
sufficient to complete a specific transaction or
obtain a particular service, but over time a more
complete user profile is obtained. An important
aspect of permission marketing is that users
receive an observable and immediate benefit
in return for the small amount of information
explicitly provided to the site and are therefore
motivated to comply.
Notably, in our e-commerce studies in which
permission marketing was evaluated, users rated
the feature very highly. Users also expect that
information explicitly entered once will persist
that is, users do not want to be asked by the
site to enter the same information at a later time
for a subsequent transaction or interaction.
The clear implications for usability evaluations are to determine whether users consider the
explicit entering of personal information onerous
or annoying, and whether there is consensus
among users regarding the implied cost-benefit
relationship. Do users ultimately find benefit in
the sites personalization features in spite of the
amount or level of work involved in enabling the
site to personalize or adapt its interactions?

WHAT IS THE SAME?

The aforementioned discussion focuses on what
is different about evaluations of personalized
Web sites and systems, that is, what distinguishes
evaluations of such sites from other Web sites.
It enumerates questions and issues that must be
addressed in the assessment of personalized sites

User-Centered Evaluation of Personalized Web Sites

from a user-centered perspective, including asking not only is the site usable, but does it provide
functionality and affordances users actually desire, that provide obvious value to users, and that
users will choose to use. The evaluation process
must address these many questions posed previously. The pragmatics of evaluating these issues
is where things are the same as other thorough
and comprehensive user-centered assessments.
The techniques and approach are the same, the
focus and issues addressed are expanded.
As for other types of Web sites, user-centered
evaluations of personalized sites certainly ought
to involve a system prototype that users can
view and perhaps minimally interact with. Upon
deciding what personalized features you wish to
potentially incorporate into the site, a prototype
that reifies those features should be constructed.
Evaluation studies should be performed early in
the design process using site prototypes; this can
be a money- and effort-saving technique if designers initial intuitions regarding adaptive features
and functionality do not match users opinions of
them. This savings may be greater in the case of
adaptive functionality than traditional software
because the former may involve techniques that
are difficult and time consuming to implement.
As Paramythis et al. (2001) suggest, Eliciting
user feedback regarding the modeling process
requires that at least a prototype of the system exists. Taking this a step further, initial user testing
should begin when only a prototype exists, that
is, before the site is fully implemented, so that
many design and implementation decisions are
made prior to the expense and time of building
the full site.
A reasonable method of putting the site and its
personalization behaviors in front of users without
the expense of actually building the system (including its adaptive functionality, which is often
complex to implement) is to use paper prototypes,
a technique often used in formative usability
evaluations. Here, drawings of potential screens

are shown to potential users, providing a sense of

what the fully implemented system would look
like. The images may be skeletal wire frames of
the proposed screens or fully rendered realistic
images of actual Web pages. Here, drawings of
potential screens are shown to potential users who
may express their opinions of particular screens,
including their content and affordances. Usability
study experimenters may be in control, demonstrating the proposed site design to focus groups or
individual users by walking the study participants
through scenarios of interaction, an approach
much like the Design Walk-through technique
for usability evaluation (Vredenburg et al., 2001).
Or participants may discuss what they would
choose to do at the interface, what actions they
would take to accomplish some scenario-based
goal. Experimenters can then show the appropriate image of the sites response to the user action
for example, a new Web page or the same page
with updated information and the evaluation
process iterates over several scenarios.
Evaluations of design ideas might be also
performed with prototype Web pages built using
presentation software, such as Microsoft PowerPoint, thus using digital images on a screen
rather than paper. Realistic screen shots can be
created using a graphical drawing and editing
tool. Taking this notion a step further, in addition
to being able to show users static screen images,
PowerPoint possesses the ability to easily add programmatic behaviors to provide true interactivity,
without the cost of a full, real implementation
of the actual system. Users cannot only see what
screens might look like in the implemented site,
but can push buttons, enter text, and in general
use the interactive prototype, whose behaviors
mimic the ultimate implementation, in hands-on,
scenario-based interactions. This technique offers a prototype for user testing of greater fidelity
than merely showing static screen images and
gives users a truer sense of the look-and-feel of
the site.

187

User-Centered Evaluation of Personalized Web Sites

Of course, this implies that driving the evaluation should be realistic scenarios of use of the
Web site. Each scenario should involve a story
in which the user must accomplish a particular
goal by using the Web site. The full corpus of
scenarios must include ones that exercise the personalization features the site offers. For example,
a scenario of use for a computer-purchasing site
might involve the goal of purchasing a memory
card that is compatible with a particular computer
(Alpert et al., 2003; Karat et al., 2003).
One approach might be to contrast performing
a particular scenario using two different prototype
sites, with and without specific personalization
features that might or might not support the user
in accomplishing specific goals. Experimenters
can walk users through interaction with mock-site
prototypes, or users can use, hands on, a minimally interactive prototype, to accomplish this
same scenario goal with the two prototypes. Or
the walk-through or interaction may occur with
only a single prototype, one that incorporates and
demonstrates a particular personalization feature.
Then, experimenter-led discussions with focus
groups and with individual study participants,
written and oral questionnaires, time-to-task
completion, number of clicks, and keyboard
interactions those materials and procedures
typically incorporated into user evaluations of
interactive systems would also be employed (as
in the personalization evaluation studies reported
in Alpert et al. (2003) and Karat et al. (2003).
For example, questionnaires might list features such as The site will conduct constrained
searches for accessories and upgrades, searching
only among those that are compatible with the
products you already own, and study participants
would be asked to rate each feature in terms of
its value to the participant using a Likert scale
ranging from Highly valuable to Not at all
valuable. Evaluations must interrogate whether
the personalization features actually helped the
user accomplish his/her goals. Questionnaires
might include assertions, including such general

188

statements as, I believe the system helped me in

accomplishing my goals and The site did not
understand very well what I wanted to do and did
not offer information helpful to accomplishing
my goals, and more specific statements such as,
When I wanted to purchase memory compatible
with my laptop, I found the site helped me in
accomplishing my goal and When I wanted to
purchase memory compatible with my laptop, it
was helpful that the site showed only those accessories that are compatible with my particular
computer model. Assertions might also probe
the affective issues that have been discussed; for
instance, I liked when the site tried to understand
what I was trying to accomplish, Im comfortable with the Web site trying to figure out what
my goals are, and The site should not try to
infer what Im trying to accomplish and should
let me be completely in control. These too would
require Likert ratings from I strongly agree to
I strongly disagree. Group and individual discussions should be sure to focus as well on issues
such as the desirability or acceptability of the sites
attempts to infer users needs and goals, and how
well the systems intervention indeed facilitated
the success of goal achievement.

CONCLUSION
Design of personalization or user-adaptive systems (or any software technology) cannot occur in
a vacuum, specifically, it cannot usefully proceed
without assessing the value and usefulness to
users of the concepts proposed and implemented
by researchers and developers. Is the overall user
experience enhanced due to the inclusion of personalized features and functionality?
We can see that many of the presented usercentered measures interact and overlap: User
trust and confidence in a Web site will plainly be
affected by the quality of the systems inferences
regarding what the user wants to see; if a user
cannot understand why particular information is

User-Centered Evaluation of Personalized Web Sites

being displayed, he/she will have less confidence

in the site; if the user is confused by the sites
actions, they will feel less in control of the usersystem interaction, and so on. The measures may
be evaluated with and by users as individual items
or joined to form more complex evaluation issues.
In either case, the evaluation of personalized Web
sites must consider several features, issues, and
questions, including:

personalization features will of course vary across

Web site domains, user types, and individual
users. Researchers, usability professionals, and
developers need to have a broad sensitivity to
the issues we have listed, and will be well-served
by investing in appropriate user-centered studies
before implementing personalization features on
their Web sites.

REFERENCES

Do users desire the particular personalization adaptations provided by the site?

Do adaptations that apparently succeed in
one application context (e.g., collaborative
filtering on Amazon.com) necessarily enhance the user experience in the context of
the site being evaluated?
Is the site using appropriate information for
its adaptations (e.g., previous navigation,
explicitly entered information, previous
click-throughs, etc.)?
Do users find the sites adaptations predictable and acceptable, or unintelligible and
even unnerving?
Are users confused by seeing information
they did not explicitly request?
Do users have confidence and trust in the
site based on experiencing its personalized
adaptations?
Are users satisfied with the amount of work
they must perform to allow the site to personalize user interactions?
Do users have an adequate sense of being
in control (even though the site may behave
differently at different times and circumstances)?
Are users willing to relinquish control to the
extent required to allow the site to personalize interactions?

Many of the questions, issues, and measures

we have raised and studied are qualitative (and
at times, emotional) in nature. Assessments of

Alpert, S. R., Karat, J., Karat, C.-M., Brodie,

C., & Vergo, J. G. (2003).
User attitudes regarding a user-adaptive ecommerce Web site. User
Modeling and User-Adapted Interaction, 13(4),
373-396.
Andersen, H. H. K., Andersen, V., & Hansen, C.
B. (2001). Establishing

criteria for evaluating intelligent agents in e-commerce. In S. Weibelzahl, D.

Chin, & G. Weber (Eds.), Empirical evaluation
of adaptive systems.
Burke, R. (1999). Integrating knowledge-based
and collaborative-filtering. In Proceedings of
the AAAI 1999 Workshop on AI and Electronic
Commerce (pp. 14-20).
Carroll, J. M., & Rosson, M. B. (1987). The paradox
of the active user. In J. M. Carroll (Ed.), Interfacing
thought: Cognitive aspects of human-computer
interaction. Cambridge, MA: MIT Press.
Chin, D. N. (2001). Empirical evaluation of user
models and user-adapted systems. User Modeling
and User-Adapted Interaction, 11, 181-194.
Consumer Reports WebWatch. (2002, April 16). A
matter of trust: What users want from Web sites.
Retrieved from http://www.consumerwebwatch.
org/dynamic/web-credibility-report-a-matter-oftrust.cfm
Czarkowski, M., & Kay, J. (2005). Scrutability as a
core interface element. In C.-K. Looi et al. (Eds.),

189

User-Centered Evaluation of Personalized Web Sites

Artificial Intelligence in Education, Proceedings

of AIED-2005 (pp. 783-785). Amsterdam: IOS
Press.
DHertefelt, S. (2000, January 3). Trust and the
perception of security. Retrieved from http://
www.interactionarchitect.com/research/report20000103shd.htm
Godin, S. (1999). Permission marketing: Turning
strangers into friends, and friends into customers.
New York: Simon & Schuster.
Grice, H. P. (1975). Logic and conversation. In P.
Cole & J. Morgan (Eds.), Syntax and semantics 3:
Speech Acts (pp. 41-58). New York: Academic.
Gupta, A., & Grover, P. S. (2003). Proposed
evaluation framework for adaptive hypermedia
systems. In Proceedings of the Third Workshop
on Empirical Evaluation of Adaptive Systems.
Retrieved from http://www.easy-hub.org/hub/
workshops/ah2004/proceedings.html
Karat, C. M., Brodie, C., Karat, J., Vergo, J. G.,
& Alpert, S. R. (2003).
Personalizing the user
experience on IBM.com. IBM Systems Journal,
42(4), 686-701.
Kay, J. (2001, July). Scrutability for personalised
interfaces. ERCIM News (No. 46). Retrieved from
http://www.ercim.org/publication/Ercim_News/
enw46/kay.html
McLaughlin, M. R., & Herlocke, J. L. (2004). A
collaborative filtering algorithm and evaluation
metric that accurately model the user experience.
In Proceedings of the 27th Annual International
Conference on Research and Development in
Information Retrieval (pp. 329-336). New York:
ACM Press.
Mobasher, B., Dai, H., Luo, T., & Nakagawa, M.
(2001). Improving the effectiveness of collaborative filtering on anonymous Web usage data. In
Proceedings of the IJCAI 2001 Workshop on
Intelligent Techniques for Web Personalization
(ITWP01).

190

Nielsen, J. (1998, October 4). Personalization is

over-rated. Jakob Nielsens Alertbox. Retrieved
from http://www.useit.com/alertbox/981004.
html
Paramythis, A., Totter, A., & Stephanidis, C.
(2001). A modular approach to the evaluation
of adaptive user interfaces. In S. Weibelzahl, D.
Chin, & G. Weber (Eds.), Empirical Evaluation of
Adaptive Systems, Proceedings of Workshop Held
at the Eighth International Conference on User
Modeling. Retrieved from http://art.ph-freiburg.
de/um2001/documents/paramythis.pdf
Reeves, B., & Nass, C. (1999). The media equation:
How people treat computers, television, and new
media like real people and places. Stanford, CA:
CSLI Publications/Cambridge University Press.
Resnick, P., & Varian, H. R. (1997). Recommender systems. Communications of the ACM,
40(3), 56-58.
Schafer, J. B., Konstan, J., & Riedl, J. (2001).
Ecommerce recommendation applications. Journal
of Data Mining and Knowledge Discovery, 5(1/2),
115-152.
Searle, J. R. (1969). Speech acts: An essay in the
philosophy of language. London: Cambridge
University Press.
Swearingen, K., & Sinha, R. (2001). Beyond algorithms: An HCI perspective on recommender
systems. Paper presented at the ACM SIGIR 2001
Workshop on Recommender Systems.
Usability First. (2004). Introduction to usability.
Retrieved from http://www.softwareevaluation.
de
Ven katesh, V., & Agar wal, R. (2005).
Turning visitors into customers: A usability-centric perspective on purchase behavior in electronic channels. Retrieved from
http://disc-nt.cba.uh.edu/chin/speakerseries/Venkatesh-Agarwal.pdf

User-Centered Evaluation of Personalized Web Sites

Vredenburg, K., Isensee, S., & Righi, C. (2001).

User centered design: An integrated approach.
Upper Saddle River, NJ: Prentice Hall.
Wang, J.-C., & Lin, J.-P. (2000). Are personalization systems really personal Effects of
conformity in reducing information overload.
In Proceedings of the 36th Hawaii International
Conference on System Sciences. Retrieved from
http://computing.breinestorm.net/conformity+in
formation+personalization+study+edu/.
Weibelzahl, S., Lippitsch, S., & Weber, G. (2002).
Advantages, opportunities, and limits of empirical
evaluations: Evaluating adaptive systems. Knstliche Intelligenz, 3(02), 17-20.
Weibelzahl, S., & Weber, G. (2001). Mental
models for navigation in adaptive Web-sites and
behavioral complexity. In T. Arnold & C. S. Her-

rmann (Eds.), Cognitive Systems & Mechanisms.

Abstracts of the Fifth Annual Meeting of the
German Cognitive Science Society in Leipzig,
KogWis 2001 (p. 74). Leipzig: Leipziger Universittsverlag.
Weibelzahl, S., & Weber, G. (2003). Evaluating
the inference mechanism of adaptive learning
systems. In P. Brusilovsky, A. Corbett, & F. de
Rosis (Eds.), User Modeling: Proceedings of the
Ninth International Conference (LNAI 2702, pp.
154-168). Berlin: Springer.
Zhu, T., & Greiner, R. (2005). Evaluating an
adaptive music-clip recommender system. In S.
Weibelzahl, A. Paramythis, & J. Masthoff (Eds.),
Proceedings of the Fourth Workshop on the Evaluation of Adaptive Systems, at the 10th International
Conference on User Modeling (pp. 69-73).

This work was previously published in Human Computer Interaction Research in Web Design and Evaluation, edited by P.
Zaphiris, pp. 257-272, copyright 2007 by Information Science Publishing (an imprint of IGI Global).

191

192

Chapter 1.15

The Usability Engineering

Behind User-Centered
Processes for Web Site
Development Lifecycles
Theresa A. OConnell
Humans & Computers, Inc., USA
Elizabeth D. Murphy
U.S. Census Bureau, USA1

Abstract
Usability is integral to software quality. Software
developers increasingly acknowledge the importance of user-centered, Web site development. The
value of usability engineering and the role of the
usability engineer (UE) are less understood. A
common assumption is that the UEs role is only
to be a user advocate. To this role, we add the
responsibility of addressing concerns of other
stakeholders in Web site design and development.
We discuss usability engineering and the processes
that it encompasses, such as project planning,
requirements definition, user-centered design
(UCD) and evaluation/testing within the context
of traditional software engineering lifecycles. We

define the UEs role throughout a user-centered,

Web site development lifecycle. This lifecycle integrates compatible usability engineering processes
into software engineering processes, drawing
examples from research and experience.

Introduction
People use the Web in a variety of ways. Their
interaction with the Web can be self-motivated
or externally motivated; their proficiency novice
or expert; their needs and expectations simple or
complex. To engineer a successful and satisfactory user experience with a Web site, we need
to understand issues such as why people go to

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

The Usability Engineering Behind User-Centered Processes for Web Site Development Lifecycles

a Web site; what they expect and intend to accomplish at the site; and everything impacting
on their experience.
A Web site is the result of a set of processes,
usually iterative, beginning with conceptualization, planning and requirements definition, then
going on to design, version production, and testing/
evaluation, before culminating in the site launch.
For usability engineering to be fully integrated into
Web site development, its practices must be fully
integrated into software development lifecycles
(Addelston & OConnell, 2004, 2005). Lifecycles
are structured frameworks for software development activities. For example, Figure 1 incorporates
elements that iterative lifecycles typically include.
In practice, the sequence and frequency of activities can vary. Research and experience show
that including usability in software engineering
lifecycles is critical (Mayhew, 1992).
Developing a Web site is a team effort. Each
team member has roles and responsibilities. The
roles of the usability engineer (UE) are integral
to these processes and to the team implement-

ing them, primarily because the UE promotes a

user-centered perspective. Software engineering
of a Web site addresses a variety of purposes:
building a new site; upgrading, refurbishing,
maintaining or introducing new information or
functionality to an existing site; and replacing
a legacy site. These purposes track to the goals
of the Web site providers. Software engineering
is not inherently user centered. It becomes user
centered by incorporating usability engineering.
User-centered processes for Web site development
are compatible and simultaneous with software
engineering lifecycle processes.
Building usability into a Web site requires
user-centered processes. Such processes require
defined roles and activities, which, in turn, depend on common definitions of concepts, inputs,
outputs, and tools. From the start, team members
must share an understanding of users attributes
and needs. This understanding underpins the collaboration necessary to incorporate user-centered
processes into Web site development.

Figure 1. A generic, variable sequence, iterative Web site development lifecycle illustrates points where
usability engineering is most beneficial
Product
Conceptualization
Engineering

Users

Project
Planning
Usability

Evaluation/
Testing

Software

Launch &
Site Maintenance

Requirements
Definition

Engineering
Version
Production

Design

Note: With the exception of version production, each of the activities in the outer ovals includes both usability
engineering and software engineering processes. In practice, the sequence and frequency of activities can vary.

193

The Usability Engineering Behind User-Centered Processes for Web Site Development Lifecycles

Guided by Figure 1, this chapter addresses the

following topics:

Usability
Users
User interface
Usability engineering
Software engineering
Integrating usability engineering into software engineering lifecycles
Lifecycle activities

The chapter ends with a summary and references.

Usability
People outside the field of usability engineering
sometimes consider usability to be obvious, but
vague and unstructured something common
sense can recognize and accomplish. Sometimes
they are surprised to learn that the field has its own
definitions and established processes. Although
those people are happy to define usability as I
know it when I see it, for UEs, a strict definition underlies our focus on users needs and our
goal of meeting those needs through usability
engineering processes. This chapter discusses
users needs in the context of human-computer
interaction (HCI), specifically as users interact
with Web sites.
The International Organization for Standardization (ISO) defines usability through the
attributes of users interactions with software
products in specific contexts of use: efficiency,
effectiveness, and user satisfaction. We boil
these attributes down to two outcomes: (1) user
success and (2) user satisfaction (1998). The ISO
definition implies that usable software must be
accessible.
Throughout this chapter, we draw illustrations
from our work in a specialized branch of usability
called accessibility. Accessibility enables people

194

with disabilities to experience success and satisfaction with software to a degree comparable to
that enjoyed by people without disabilities. Although some authors treat accessibility as distinct
from usability, we consider it to be a subdomain
of usability in which the users are people with
physical and/or cognitive disabilities (Hix &
OConnell, 2005).

Users
In the context of this chapter, users are people who
interact with Web sites. In the sense in which we
use the term, users are also known as end users,
the people who visit Web sites and interact with
their contents. The term user excludes people
employed in a Web site project, for example, the
UEs. It excludes the sites providers and others
who have any stake in the Web site. People close to
the project can be too technical or too expert in a
domain to represent a user who does not have the
same training, goals or background. Those close
to the project run a high risk of unintentionally
clouding their view of users needs with their own
commitment to achieving the projects goals.
Many variables are inherent to users (Bias &
Karat, 2005). Some are intrinsic, for example, age;
gender; experience with technology; intellectual
or aesthetic preferences; interaction styles; and the
presence or absence of physical or cognitive disabilities. Other variables, such as employer goals
and working environment are extrinsic, but affect
the user experience. Many user attributes can
decline with age, for example, memory and perception of color contrast (O'Connell, in press).
Each user brings a cluster of capabilities and
limitations to any interaction with a Web site.
These are the well-documented human capabilities
and limitations in perception, manual dexterity,
memory, problem solving, and decision making
(e.g., Baddeley, 1990; Brown & Deffenbacher,
1979; Mayer, 1992). For example, the limitations
of working memory are well known: seven plus

The Usability Engineering Behind User-Centered Processes for Web Site Development Lifecycles

or minus two chunks (sets of items), with the

size of a chunk varying depending on the users
experience (Miller, 1956). Research has found
that working memory can be futher limited by
environmental characteristics, for example, noise,
fatigue, perceived time pressure, and other sources
of stress (Bradley, 1990). Correctly applying such
research findings to Web site design is the UEs
responsibility.
Some user attributes are physical. The term
user includes people with disabilities as well as
people without disabilities. Users with physical
disabilities require a Web site that accommodates
those disabilities. For example, a user with low
manual dexterity may be a skilled problem solver,
but have difficulty clicking on small screen elements with a mouse. As content density increases
on Web pages, the need for small controls rises.
Consider a page with a long list of expandable
or contractible menu options. By offering the
expanded state as the default, the site accommodates users with low manual dexterity, relieving
them of the need to click on the tiny controls.
Tradeoffs are inevitable. The user who cannot expand or contract the list must contend with content
density. Accommodating users capabilities and
limitations is rarely simple. Sometimes we must
accommodate several factors at once. The user
with the manual dexterity disability may also have
a visual disability brought about by age.
The World Wide Web Consortium (W3C)
considers the needs of users with a wide variety
of disabilities, the W3C gives guidelines and
success criteria for assessing accessibility (W3C,
2004). For users with special needs, an accessible
Web site provides perceivable content; operable
content components; understandable content and
controls; and content compatibility with other
technologies ( Caldwell, Chisholm, Slatin, &
Vanderheiden, 2005).
Publishing a Web site makes it available to the
world, but it is not possible to design for the world.
Certainly, there are generic commonalities across
Web sites. Any Web site can have hyperlinks,

controls, and information architecture. But, any

Web site will also have unique features to meet
the needs of its targeted audience. Designers need
to know the characteristics of the sites targeted
audience to promote their success and satisfaction at the site. They must also project future
user characteristics. For example, the intended
audience needs to include users with disabilities,
even if none of the current users has a disability.
Todays able user may have a disability tomorrow.
Age brings disabilities (OConnell, in press). New
employees may have disabilities.
A Web site goal is to enable its users to experience success and satisfaction by accommodating
each user attribute of the set of targeted users as
well as the unique constellation of attributes defining each individual user. However, the average
user is a myth. Designers need to understand the
range of each attribute within a group of users and
to recognize that users will fall at all points along
the range (e.g., low to high spatial ability).
To help us accommodate these ranges while
still focusing on targeted users, we group users into
classes, rather than design for them as individuals.
We define user classes by peoples participation in
a set of user attributes. For example, for a theaterticket vending site, the senior citizen user class
will include the attributes, aged 65 or over and
partial memory loss. The first attribute qualifies
these users for a discount; the second tells UEs
that the interface needs to remind users of the time
of the show. Other user classes for this site would
include adults, children, and members of theater
clubs, each class having its own attributes.

User Participation in Web Site

Design
It is essential to include users in the process of Web
site design because users will do what they will,
not necessarily what the designers or developers
want or expect them to do. User behavior can
be anticipated up to a point, but not predicted in
exact detail. Not being intimately familiar with

195

The Usability Engineering Behind User-Centered Processes for Web Site Development Lifecycles

the design of a site, users will try to do what

seems logical to them, even though their logic
may conflict with the designers or developers
logic. According to Shneiderman (1998), users
do not make errors of intention because they are
doing what seems logical to them in the context
of the task they are trying to complete. Users do
not try to get lost, but they do get lost.
User involvement in design extends our understanding of human behavior within the context
of the site and its stakeholders goals. In some
projects, users participate in focus groups and
interviews or respond to surveys. They join in
conceptualization and design discussions. Observing users interact with a Web site provides the
most valuable new insight, taking UEs to a level
of understanding not otherwise possible. These
observations are combined and analyzed to develop profiles of the range of expected interactions.
Observing users interacting with the site identifies
unanticipated needs and behaviors. Analyzing
observational data produces requirements exceeding what programming logic, technical expertise,
and the best of intentions can identify.
Some software engineering processes eliminate user participation but purport to accommodate a user orientation. These include assigning
a user surrogate and processing user information
channeled through a nonuser, such as the users
manager. In contrast, usability engineering
relies on close interaction with real users at the
strategic points where their input is most crucial.
It documents input from the users who actually
participated in lifecycle activities.
Knowledge gained from usability engineering cannot be gained by other means. This is
not to say that users drive design. Users are not
designers. It is a key role of the UE to translate
the understanding of users and their needs into
design recommendations. It is a key role of designers and developers to incorporate usability
recommendations into interactive interfaces that
promote users success and satisfaction.

196

Users Mental Models

Along with their capabilities and limitations,
users bring with them their previous experience
of computers and Web sites. We assume they
have built mental models, that is, psychological
representations of the ways in which computers
and Web sites work (Carroll, 1990; JohnsonLaird, 1983; Van Der Veer & Del Carmen Puerta
Melguizo, 2003).
Highly experienced users can have mental
models of different categories of Web sites, for
example, sites for entertainment, information
gathering, and e-commerce. According to the
users mental model, an entertainment site should
use bright colors and animation, whereas an information site should use subdued colors and minimal
animation, for example, only to demonstrate an
integral concept. Abstracting and representing
users mental models is another job for the UE.
Tools for doing this include cognitive task analysis
(e.g., Crandall, Kline, & Hoffman, 2006).
User expectations for Web sites are based on
their experience with the Internet. They have
expectations for the behavior of controls, for
example, the back and forward button, Tab key,
and Enter key. They have expectations for hyperlinks, for example, a link once activated will
change its color. Some expect to be told whether
clicking on a link will take them outside the site.
Sometimes, users worry about whether they will
be able to get back to where they are if they click
on a link because they have been unable to form
a definite mental model for link behavior. This
uncertainty can arise when they have experienced
links behaving inconsistently across sites.

User Interface
In one sense, a user interface (UI) is software
that people use to interact with technology. For
UEs it is a matter of layers. Above the underlying

The Usability Engineering Behind User-Centered Processes for Web Site Development Lifecycles

technology is the important look-and-feel layer

(Garrett, 2002). Feel refers to more than the point
and click paradigm; it also refers to what users
hear, for example, a sound associated with an
alarm. The UI includes both the users actual and
expected mode of interaction with a Web site, for
example, keyboard, mouse, or speech. Whether
the designers intend it or not, because of mental
models the UI includes implied functionality
based on similarities to familiar software. In the
broadest sense, the UI is the virtual place where
the users mental model, meets the designers
system model (Bolt, 1984). Aligning these models is a goal of usability engineering (Norman &
Draper, 1986).

Usability Engineering
Usability engineering is a set of defined, usercentered processes, grounded in research and
experience-based principles. The purpose of
usability engineering in Web development is to
raise the potential for users success and satisfaction and, thereby, to support Web site providers
goals. The UE must understand the complex set
of variables residing in any user group and apply
this understanding to promote users success and
satisfaction. This understanding is what makes
usability engineering critical to achieving Web site
usability. Because peoples styles of interacting
with technology change as technology progresses,
usability engineering is a continually evolving
field informed by applied research in human
interaction with technology.
The UE applies expertise not usually found
in other software development team members to
make an essential contribution to the quality of
a Web site. As noted by Bias and Karat, good
usability is not standard for most Web sites
(2005, p. 2). When usability engineering is not
part of Web site development, the team faces a
high risk that, at the least, the site will not promote users success and satisfaction; at worst, it

will inhibit users success and satisfaction and

thereby prevent achievement of the site providers
goals. This outcome would be a disservice to the
providers and other stakeholders.
Usability engineering is practiced by UEs who
typically have training and experience in a variety
of disciplines. In the case of Web work, relevant
experience is in fields such as psychology, HCI,
testing protocols, and design. The UEs work is
distinguished by its orientation toward the user, but
usability engineering is more than user advocacy.
For example, in addition to understanding users
mental models and expectations, the UE must
also be well versed in technology, standards, and
laws. To create feasible design approaches, the
UE must understand organizational and project
goals. To integrate usability into the lifecycle,
the UE must be able to communicate with users,
stakeholders, and other members of the Web site
development team.
A Web site has several objectives. While delivering information and functionality, it also bears
the responsibilities of putting the site providers
best foot forward and achieving their business
goals. A Web sites usability influences users
impressions of the site providers integrity and
trustworthiness. UEs are qualified to coordinate
the factors necessary to meet the needs of both users and site providers. As we show later, the means
to this end is to incorporate usability throughout
a software development lifecycle.

Usability Engineering Principles

Usability engineering is a multi-disciplinary field.
With roots in human factors, it also draws on disciplines such as software engineering, linguistics,
biology, cognitive psychology, technology, and
graphic design. These fields diverse contributions
are documented in usability principles, a set of
research and experience-based, widely accepted
guidelines for achieving usability (e.g., Koyani et
al., 2003; Mayhew, 1992; Shneiderman & Plaisant,
2004). UEs continually update usability principles

197

The Usability Engineering Behind User-Centered Processes for Web Site Development Lifecycles

in response to new research and technologies.

Usability principles empower design strategies
to meet users needs and expectations, while
avoiding unnecessary, inappropriate features that
burden schedules and budgets.
Most usability principles trace to human
capabilities and limitations. A user with color
deficiencies can detect changes in brightness or
shape but cannot distinguish between certain
colors. Therefore one usability principle tells us
never to rely on color to convey meaning. An
important usability principle requires taking steps
to assure users always feel in control. Progress
bars and hourglasses give users a sense of where
they are in a process, contributing to their sense
of control. Another principle obliges UEs to direct
users through design, for example, leading them
through a Web-based form, by following their
expectations, for example, whether the family
name should precede or follow the last name.

or expect little chance of successfully integrating usability engineering in a Web development

project (Hix & OConnell, 2005). For example,
it is not uncommon for stakeholders to mistakenly consider usability engineering an expensive
add-on that puts users needs above all others.
To counter this misconception, the UE will give
examples of usability as a cost-effective and costjustifiable, integral contributor to achieving the
site providers goals while instilling usability (e.g.,
Mayhew & Tremaine, 2005; Web Accessibility
Initiative, n.d.). Another means to overcoming
stakeholder misconceptions is to bring them in to
watch early usability evaluation sessions where
they see users encountering usability problems.
When they return for later sessions after changes
have been made to the UI, stakeholders observe
users experiencing the benefits of designing for
usability.

Stakeholders

Software Engineering

Everyone who has any interest in a Web site is a

stakeholder, whether they are company officers,
the marketing team, or the system administrators
in charge of maintaining the site. Site providers
are the stakeholders who finance the Web site and
set the business goals. They stipulate the purpose
of the site from the business point of view. They
allocate resources and set policy.
Although some consider UEs to be principally user advocates, UEs have another equally
important job. That is to address the concerns
of stakeholders during software development.
It is the UEs responsibility to make sure that
the design, while promoting users success and
satisfaction, also promotes the aims of the Web
sites stakeholders.
Sometimes stakeholders have misconceptions
about the worth of usability engineering or how it
integrates with software engineering. Recognizing and addressing this fact is an essential task
for UEs; they must remedy these misconceptions

Software engineering is the application of a

systematic, disciplined, quantifiable approach
to the development, operation, and maintenance
of software (IEEE, 1990, p. 70). A software
engineer plays two roles, computer scientist and
project manager. On a large Web project, software
engineers typically oversee both the technical and
management aspects of development. They accomplish this by following lifecycles that stipulate,
for example, a projects phases, methods, activities, inputs, outputs, milestones, documentation,
and risk mitigation strategies.
The software engineering view of usability
has not always coincided with the UEs definition
or outlook. Historically, software engineerings
attention to usability was largely confined to
summative testing activities (Hix & Hartson,
1993), validation at projects end that people can
use a Web site. Now, software engineers more
often consider usability a valuable part of Web
site development lifecycles.

198

The Usability Engineering Behind User-Centered Processes for Web Site Development Lifecycles

Integrating usability
Engineering into Software
Engineering Lifecycles
Software engineering lifecycles are hospitable to
usability engineering. Commonalities between
software engineering and usability engineering
facilitate this compatibility. The two professions
share tools such as use cases, although they sometimes employ them differently. They have the
common goal of delivering quality Web sites, on
time and on budget, to satisfied users, customers,
and other stakeholders. They share terminology,
but sometimes with different meanings or connotations. For example, in software engineering,
the word interface primarily means a connection
between two components of a software system,
whereas, to a UE, interface first and foremost
denotes the human-computer interface.
Software engineering and usability engineering processes can occur in parallel because their
activities and outputs are compatible. Sometimes
these processes are rigid, but the constraints of
developing Web sites in real time against tight
schedules and tighter budgets drive a trend toward
adaptability. This trend emphasizes the need for a
UE. In this fast-paced environment, users on the
development team can be few and their involvement infrequent. In such a case, a UE draws on
knowledge of the field, for example, usability
principles and knowledge of users, to aid in the
development of usable Web sites.

Lifecycle Activities
Usability engineering has corresponding activities
for most software engineering activities. Not all
lifecycles incorporate the same activities. Activity
sequence and frequency can vary. Some activities can be simultaneous. Each activity has goals,
inputs, processes, and products.
In Figure 1, we present a high-level view of
a user-centered software engineering lifecycle.

We use a generic lifecycle where all activities

are connected to each other, feeding output into
subsequent activities. The sequence of activities is adaptable. Within each activity, usability
engineering and software engineering processes
occur simultaneously.
Table 1 sets out some activities and processes
of software development lifecycles. In addition to
usability in general, the goal or objective of these
activities and processes is to provide accessibility for people with disabilities. This table is not
comprehensive, but it demonstrates how some
usability engineering processes can be integrated
into a software development lifecycle.
Although we maintain the importance of usability engineering throughout Web site development (e.g., Murphy, Marquis, Nichols, Kennedy,
& Mingay, 2001), our discussion zooms in on the
most critical iterative process areas. These core
activities are project planning, requirements definition, design, and evaluation/testing. Although
these are a subset of all possible software engineering activities, we designate these activities as core
because they represent the make it or break it
points in a typical software engineering lifecycle
where usability engineering must be addressed.
However, none of these activities stands alone.
So, we place them within the context of other
typical iterative lifecycle activities such as product
conceptualization and version production.

Project Planning
A Web site project starts as an idea. This can be
a new idea encapsulating the site providers vision. More often, at the end of a lifecycle, a team
returns to product conceptualization when they
evaluate an existing site and begin to plan future
versions. In either case, the team needs a blueprint
for the steps between the original concept and the
insertion of the final product into the workplace.
This blueprint is called the project plan.
Formal software engineering lifecycles start
with project planning. Successful project planning

199

The Usability Engineering Behind User-Centered Processes for Web Site Development Lifecycles

Table 1. Usability engineering activities and processes during a software development lifecycle (Partially
based on Addelston & OConnell, 2005; Mayhew, 1992)
Software & Usability
Engineering
Lifecycle Activities
Product
Conceptualization

Project Planning

Requirements
Definition

Design

Version Production
Evaluation/Testing
(Across versions)

Launch & Site

Maintenance

200

Examples of Usability Engineering Processes

Literature review, for example, to check latest research results
Expert review of competition or legacy Web site
Brainstorming
Collection and analysis of users feedback on legacy Web site
Overview planning, for example, project summary with requirements
Project organization planning, for example, identifying and assigning resources,
roles, and responsibilities
Usability risk management planning
Technical process planning, for example, accessibility processes, equipment, and
tools
Test planning
Incorporating UE responsibilities into plan
Interviews
Focus groups
User profiling
User class definition
Translating users needs into requirements
Behavioral and cognitive task analysis
Persona development
Setting usability goals for evaluation/testing
Post-release user surveys
Analyzing usage statistics
Contextual user observations
User-centered design
Interaction design (e.g., use cases)
Participatory design
Style guide writing
Story boarding
Card sorting
Instantiating design in Web-authoring tools
User interface evaluation with user participation (measurement of users success and
satisfaction)
Expert review
Automated testing with accessibility tools
No usability engineering processes

The Usability Engineering Behind User-Centered Processes for Web Site Development Lifecycles

depends on a clear understanding by all team

members of the site providers vision as well as
an appreciation of the concerns of other stakeholders. Usability engineering and user involvement
must be addressed from the start in the plans
schedule and budget. If not, they are excluded
early, with little chance of later inclusion. This is
true whether development focuses on a new site
or an existing site.
Project planning aims to ensure on-time, onbudget delivery of a Web site that meets its providers organizational goals and fulfills requirements.
Usability engineering adds the goals of users
success and satisfaction. Another primary goal
is to gain site provider and team acceptance of
usability engineering (Hix & OConnell, 2005).
Input to project planning includes output from
product conceptualization such as market analyses and everything learned while maintaining a
legacy site.
Typical processes involved in project planning
are setting schedules; identifying and allocating resources; and stipulating activities, inputs,
outputs, and milestones. Usually, planning is
the responsibility of software engineers and
managers. However, the UE contributes to planning by identifying the points where usability
engineering is necessary and identifying the
inputs and outputs needed to proceed from one
step to the next. For example, the UE will point
out the need to design and produce an accessible
version with alt tags (alternative text). An alt tag
is HTML code associating words with Web site
graphics to describe their content. Alt tags must
be present before evaluation with a screen reader,
an assistive device that reads aloud the contents
of Web pages, including alt tags. Screen reader
users include people with little or no eyesight.
Assistive hardware, such as a screen magnifier,
or software devices, such as screen readers,
help people with visual disabilities interact with
technology. The UE also alerts the team to the
need to schedule users with visual disabilities to
participate as evaluators.

The product of these processes is a project plan

that integrates usability engineering activities with
other scheduled project activities and defines all
players roles, responsibilities, milestones, inputs,
and outputs. The unique goals of usability engineering during planning are that the Web sites
design will consider the user and be hospitable to
user-centered processes, and that the project will
fund, schedule, and provide usability engineering.
If these goals are to be achieved, the plan must
require usability engineering.

Requirements Definition
From the software engineering perspective,
requirements definition is a set of processes that
identify and document a Web sites goals in terms
of how the site will fulfill its providers vision
by delivering information and/or functionality.
It focuses on user needs assessment. Usability
engineering also looks at the site from the users
perspectives as well to verify that the users needs
and expectations are being met. It addresses user
requirements.
Web sites have functional, system performance, and usability requirements. Functional
requirements define what a Web site is supposed
to do. For example, a functional requirement for
an e-commerce site that sells printers stipulates
that the site must display photos of the printers.
System performance is a measure of how well
the Web site does what it is supposed to do. In
our example, a system performance requirement stipulates that the site will deliver a specified photo over a 56K modem in less than two
seconds. Usability requirements are sometimes
called non-functional requirements. This term is
misleading, however, because it diminishes the
importance of usability.
Usability is a measure of users success and
satisfaction with their experience at the site. But
usability must also address the goals of the sites
providers. For example, the providers of an ecommerce site benefit from having customers

201

The Usability Engineering Behind User-Centered Processes for Web Site Development Lifecycles

revisit their site frequently. To achieve this goal,

usability requirements stipulate that a user be able
to purchase a product, making no more than one
error, and give a satisfaction rating of seven on a
scale where nine is the maximum positive rating.
Users success and satisfaction benefit both the
users and the sites providers.
Some interface requirements are yes/no, for
example, the Web site will provide anchor links
on every page at its top two levels. Others are
quantified, for example, users will be able to
navigate, with only one click, from a product
description page to a page where they can order
the product.
Some requirements are standards-driven.
Standards can be defined by the site provider, for
example, an in-house standard for displaying the
company logo. Standards can come from groups
such as the ISO, which, for example, gives standards for representing names of languages in its
ISO
639 (ISO, 2002).
Sometimes requirements are legally mandated.
Take the case of special needs, a term that refers
to the needs of users with disabilities. The UI of
a U.S. federal government site must comply with
Section 508 of the Rehabilitation Act of 1973 as
amended in 1998. These requirements aim to
provide users who have special needs a quality
of information accessibility comparable to that
of users without disabilities.
The primary goal of requirements definition
for the UE is setting usability requirements, the
Web sites obligations to address users needs. In
usability engineering, users needs and Web site
requirements are often the same thing.
Ideally, filling requirements will meet users
needs. However, a Web site can fulfill all functional and system performance requirements,
yet still not be usable. In the end, it is the users
experiences of a Web site that determine whether
the site has achieved the site providers vision.
So, usability engineering promotes a perspective that incorporates the product concept, but
expands it with an understanding of targeted users

202

and their needs. Usability engineering brings an

understanding of factors that may otherwise not
come into the mix, for example, users mental
models in the context of their capabilities and
limitations.
Usability engineering processes during requirements definition start by considering users
and their needs within the context of the Web sites
intended purposes. Inputs to this process are the
providers goals for the site as well as existing
information about targeted users and their needs.
As other team members set performance and
system function requirements, UEs learn whether
proposed content will meet not only providers
goals, but also users needs. If the providers goal
is to inform users about weather conditions, are
users people who need to know only tomorrows
temperature and precipitation forecast, or do they
include fishermen interested in tides as well? If
the project involves an existing site, UEs address
what worked and what did not work for its current users. This understanding empowers UEs
to leverage the fact that each users experience
of the site is a reflection of that users needs and
expectations.
Requirements definition brings the UE face-toface with targeted users during interviews, focus
groups, and observations in their workplaces, that
is, any place where people interact with the Web.
The UE develops surveys and studies existing
documentation such as usage statistics for a legacy
site. However, existing documentation often omits
the users real-world practices. Therefore, close
interaction with users is key. However, the Web
introduces the requirement of designing for a
wider range of users than a project can usually
involve. This factor dictates that the UE rely also
on usability principles and knowledge of human
capabilities and limitations within the context
of Web use. The UE consults resources, such as
Web sites on Web usage statistics, to keep abreast
of changing interaction behaviors (e.g., Internet
World Stats Market Research, 2005; Refsnes Data,
2005). Such resources inform production concep-

The Usability Engineering Behind User-Centered Processes for Web Site Development Lifecycles

tualization as well as requirements definition.

The UE aims for user-informed requirements,
but not necessarily user-driven requirements.
Although users input is integral and respected,
the UE must also draw on knowledge of usability engineering to inform recommendations for
requirements. For example, when information
must be stressed, users may have a preference
for many colors and other emphasis techniques
on the same display. The UE knows that a userdriven requirement for large blocks of text in upper case, presented in a large variety of strongly
contrasting colors will result in adverse effects
on user performance (e.g., Koyani et al., 2003).
The usability-engineered requirement will specify
what portion of the text to emphasize and how to
emphasize it in a way that promotes users success. For example, the requirement will stipulate
initial capitals and no more than three or four
compatible colors that can be distinguished by
most users with visual color deficiencies.
Once the UE has collected data on users, data
analysis occurs. The first step is to define user
profiles, descriptions of the Web sites target population. User profiles record user attributes such
as computer literacy; experience in the subject
matter and functionality of the site; physical and
cognitive capabilities and limitations; special
needs; education; mental models; interaction
styles; goals at the site; and tasks all important
factors that impact on user interaction with the
site. In another process, user class analysis (also
called user group analysis), the UE allocates user
profiles into groups according to shared attributes,
thereby defining user types, such as patients or
policy holders.
In behavioral and cognitive task analysis
(also called process analysis), the UE develops
descriptions of users goals and the tasks they
perform to accomplish those goals. These definitions are developed within the context of the site
providers intended purposes for the site. During
this key requirements definition activity, the UE
studies existing documents and observes users

to learn the steps and sequences that they take to

accomplish goals. Behavioral analysis documents
observable tasks such as receiving information
and inputting to the computer. Cognitive task
analysis documents users mental transformations and decisions. Doing mental arithmetic is
an example of a mental transformation. Do users
have to keep a running total in mind of their bill
on an e-commerce site or does the Web site do it
for them? When comparing products as the basis
for decision making, is the product information
displayed in a way that facilitates comparison, or
does the user have to navigate between pages to
find comparative data?
Task analyses include steps and work flows.
They describe the users experience from beginning to end. For example, when users must fill
out a form on a secure Web site, the UE documents the experience by identifying the initial
steps users need to go through, such as receiving an information package in the mail, locating
the URL for the site, locating the users unique
identifying number for login, all long before the
user navigates to the form (Gordon & Gill, 1997;
Hackos & Redish, 1998; Kirwan & Ainsworth,
1992; Redish & Wixon, 2003; Schraagen, Chipman, & Shalin, 2000; Vicente, 1999).
During requirements analyses, the UE can
develop personas, imaginary representatives of
highly specified classes of users (e.g., Head, 2003;
Pruitt & Grudin, 2003). Personas reflect what
has been learned about users during requirements analysis. They are detailed descriptions of
typical users, often given names and illustrated
with commercial photographs. Personas become
almost real to the team, serving as reminders of
typical users needs.
The UE tracks the findings of these processes
to the sites intended content to assure that it is
presented in a manner that empowers users to
achieve their goals at the site. For example, during user class definition, the UE specifies groups
of people with disabilities, associating needed
assistive devices with the user group. The UE

203

The Usability Engineering Behind User-Centered Processes for Web Site Development Lifecycles

introduces requirements for the site to present

content in a way that makes it accessible via
these devices.
In translating user needs into requirements, the
UE draws on a variety of sources. The sites goals
are viewed through the lens of users needs. For
example, consider a case where one of the sites
goals is to enhance the organizations image by
showing its chief technology officer talking about
a new technical approach. Considering users with
hearing impairments and users with low bandwidths that rule out multimedia, the UE introduces
two usability requirements: (1) captions and (2)
a text version of the presentation. Meeting these
requirements benefits the organization by spreading the message to some who would not otherwise
be able to receive it. Thus, the UE addresses the
needs of both the users and the organization that
provides the Web site.
Usability requirements become goals for later
usability evaluation. Setting specific, measurable
usability goals with the client provides a quantitative basis for assessing the design against users
needs (Whiteside, Bennett, & Holtzblatt, 1990).
Although requirements definition produces
helpful artifacts such as user profiles, user classes,
user task descriptions, and personas, the most
important products of requirements definition are
the usability requirements that specify the sites
look and feel. Usability requirements become a
checklist for everything that must be accomplished
to promote successful and satisfactory users
experiences at the site.
A project benefits from understanding users,
the motives behind their actions, and the rationales behind their opinions about the site. When
no requirements reflect users and their needs, a
project is at high risk of developing an inappropriate site a site that does not meet its intended
users needs and expectations. It risks wasting
time and resources meeting inappropriate goals.
It risks negative impacts on schedule and budget
because of the need to retrofit, that is, redesign
and recode the UI to correct usability problems.

204

Although requirements are established at

the beginning of a project, they are iteratively
reviewed and updated as more is learned about
users and the ways that the Web sites look, feel,
functioning, and performance impact users' success and satisfaction.

Design
Incorporating users input from requirements
definition, the UE participates in developing the
sites information architecture. Information architecture is like a road map; it sets out the paths
that users follow to their destinations on a Web
site. It is at the heart of design. Impacting more
than site navigation, the information architecture
impacts a pages content and layout. The UEs
role is to assure that the information architecture
facilitates navigation and makes finding information natural for users.
Important UCD processes, collectively called
interaction design, consider the ways that real
users attempt to accomplish goals at a Web site.
UEs base interaction design on all that they have
learned about the users, for example, their agebased capabilities, mental models, and expectations within the context of the goals of the sites
providers. Usability principles provide UEs with
rules of thumb that inform UCD decisions. Consider a site intended for senior citizens who expect
a prominent link to articles about leisure activities
for seniors. The UE considers usability principles
on legibility for older users with decreased visual
acuity. These principles recommend a large font
and a strong contrast between the font and background colors (e.g., Czaja & Lee, 2003).

User-Centered Design
Best practices in usability engineering include
UCD, a set of usability engineering processes that
focus on understanding users, their goals, their
strengths and limitations, their work processes
all user attributes that impact how users will

The Usability Engineering Behind User-Centered Processes for Web Site Development Lifecycles

interact with a Web site. The goal of UCD is to

achieve users success and satisfaction by incorporating the users perspective into design.
The UEs multi-disciplinary background adds
value to interface design. For example, understanding technology is a prerequisite for designing
an accessible Web site. Publishing a Web site is
simple nowadays. Adding unneeded features is
tempting, just because it is so easy to do. The UE
knows how to manage the impact of features,
such as animations, on users with disabilities.
It is a simple matter to give an information site
a bright red background with flashing blue bold
titles. The UE understands the biological impacts
of such an approach, the potential for the eye to
become fatigued because it is unable to focus
(Travis, 1991). The UE also knows that animations increase download time and, that therefore,
on an informational site, animations can reduce
users satisfaction.
The UE brings to UCD an understanding of
disciplines such as psychology and semiotics, the
science of signs and symbols. When incorporating
icons into a UI design, for example, it is important
to use standard icons to mean what they usually
mean and to test any new designs for user comprehension. If a standard icon is used to mean
something different from what the users expect,
it is likely to cause confusion. Many users will
have no idea how to interpret an ambiguous icon
in the context of their tasks. With effort, users can
learn arbitrary meanings for icons, but they easily
forget arbitrary meanings. Icons need text labels
to clearly indicate the actions that will occur when
they are activated (Horton, 1994).
The UE applies usability principles to participatory design, a UCD process in which users
comment on design concepts and perhaps generate
their own sketches. Users offer opinions on mockups or prototypes. A prototype Web site is like a
preview of coming attractions at the cinema. It
includes a sampling, but not all of the features and
functions of the planned site. Sometimes it is an
experiment to investigate UI concepts.

A typical participatory design process is card

sorting, where users sort terms that are going to be
used in the Web site into groups which they name.
The UE combines results from all participants
through a statistical technique. Applying usability principles, the UE then derives a meaningful
organization of topics for the Web site to inform
the information architecture.
We distinguish between user-centered design
and inappropriate user-driven design where users
input translates directly into design directives.
Although user-driven design has the admirable
quality of being user-focused, it excludes the input
of a UE. In turning users requests into design decisions without looking at them in light of usability
principles, practitioners of user-driven design run
a high risk of producing Web sites that, in the end,
do not meet users needs. Another pitfall of userdriven design is requirements creep that extends
schedules and strains budgets as users add inappropriate features and functions that, at worse,
will have a negative impact on their experience
at the site (Andre & Wickens, 1995).
Ideally, a team of designers, developers, and
UEs document UCD decisions in a style guide to
promote consistent design. For example, a style
guide specifies conventions for screen layout;
size, spacing, and location rules for screen elements; and fonts, icons, and color palettes. The
style guide evolves, continually updated to record
new design decisions.
Design is rarely a one-time effort. Design
versions are iteratively evaluated and revised
throughout the lifecycle. In many lifecycles,
design and version production are interspersed
or even simultaneous. In these situations, the
UE performs design consultations and evaluates
iterative products, informing design decisions
with knowledge of users and usability principles.
UCD results in a plan for the ways that users will
interact with the Web site. The principal product
of design is not necessarily a ready-to-publish
Web site. It can be a prototype.

205

The Usability Engineering Behind User-Centered Processes for Web Site Development Lifecycles

Traditionally, during design, the UE has been

a user advocate who consults on decisions involving the UI. In this fast-paced age, UEs are more
and more involved in design creation and version
production, using Web site creation tools.
Although design iterations occur throughout
the lifecycle, once part of the site is implemented,
programmers are naturally reluctant to make
changes. Engineering usability up front reduces
the need to request changes after programming.
This is why UEs and programmers need to work
closely together during UC design.
The principal benefit of having a UE in the
loop is that design is more likely to speak to users needs. Results include increased productivity,
shorter learning times, longer and repeated visits,
increased profits, and decreased costs (e.g., Bias
& Mayhew, 2005; Kalin, 1999; Mayhew 1999).

Use Cases
A use case is a formal description of ways a product
can be used. It consists of a statement of goals
with a description of the users and the processes
the designers expect them to perform to achieve
those goals. Sometimes, a use case is expressed
in a sketch. Use cases first come into play in task
analysis activities during requirements definition.
They are referred to during design.
Use cases provide an example of how a UE
can prevent a well-intentioned practice from misrepresenting users. Use cases are the product of
a process analysis technique to develop a simple,
high-level statement of users goals and processes.
Use cases are common to the tool kits of both
software engineers and usability engineers.
Basing Web design on use cases has strengths
and weaknesses. For each module of a system,
common processes are written up with the prerequisites for each process, the steps to take for
the users and the system, and the changes that
will be true after the process is completed. Use
cases help to ensure that frequent processes are
supported by the system, that they are relatively

206

straightforward, and that the system architecture

reflects the process structure.
Use cases, however, do not account for all possible user interactions at the Web site. Use cases
tend to stress behavioral tasks, but do not capture
cognitive tasks. Use cases do not leave room for
users unexpected actions at the Web site. Users
will naturally do what seems apparent to them,
based on the cues given by the UI. A use case
does not necessarily represent a natural action
for users in the context of the moment. Use cases
can put users in the position of having to provide
unexpected input to the computer input that the
computer needs but that users do not necessarily
know they are supposed to provide.
Use cases make assumptions about the users,
for example, that they understand the internal logic
of the system the way developers do. Consider
novice users who try to use the back button only
to discover it does not work because a second
browser instance has launched unannounced
and without taking focus. The use case depicts
users navigating between browser instances, but
does not accommodate their expectation to use
the back button or the fact that they do not know
about the second browser.
The limitations of use cases demonstrate the
need for usability engineering. If design relies
on use cases but omits a use case for a certain
goal/process set, the site will lack important functionality or information. Conversely, use cases
not derived from understanding users can result
in unnecessary features or information. The UE
adds value to use cases by making user-centered
recommendations that would not be in the picture
otherwise. The UE adds the human dimension
to an otherwise limited view of the user as the
provider of input to the computer and the recipient
of output from the computer. The UE knows that
factors such as the user's thoguht processes and
physical abilities are key to system success.
As with any other tool, use cases must evolve
as a project progresses. During updates, the UE
introduces the user perspective, incorporating

The Usability Engineering Behind User-Centered Processes for Web Site Development Lifecycles

what has been learned since the last iteration about

how users will interact with the site. Without a
UE, the project risks misapplying use cases. For
example, avoiding a one-to-one relationship between use case and screen, the UE assures that
screens accommodate users decision-making
strategies and work flows, not someone elses
model of discrete interactions.

Evaluation/Testing
Verification and validation (V&V) are software
engineering terms for testing. Verification is iterative testing against requirements. Validation is
the final testing against requirements at the end of
the lifecycle. Usability evaluation is a set of V&V
processes that occurs in conjunction with other
V&V activities and is an integral component of an
overall V&V approach. In addition to checking for
conformance to usability requirements, usability
evaluation has the added goal of assessing a wide
range of users experiences at the site. The UE
keeps the door open for new requirements based
on the way real users interact with the site. New
requirements become input to the next cycle. At
projects end, they become input for new product
conceptualization.
Key user-centered, usability evaluation processes entail observing users interacting with a
Web site. Activities for formal user observation
include writing a test plan; identifying participant
users; working with site providers to set usability goals for each user group and task; defining
tasks; writing statements of goals that never tell
users how to achieve those goals; preparing a
user satisfaction survey; preparing ancillary
materials such as consent forms; carrying out
the observations; analyzing data, and writing a
report (Lazar, Murphy, & OConnell, 2004). These
formal processes entail structuring evaluation
activities to reflect the tasks identified during
requirements definition.
Evaluation draws on the products of all earlier
activities. For example, usability goals are based

on input from task analysis (Whiteside et al.,

1990). Designers also identify features or functions about which they have usability concerns.
The UE makes sure that these concerns are addressed in the scenarios to generate information
needed to inform UCD decisions.
During a usability evaluation session, users
work with scenarios, sometimes behind one-way
glass. On the other side of the glass, UEs often
employ click-capture software; record numbers
and types of user errors; document latency periods when users pause for a significant period
of time trying to figure out what to do next; and
record critical incidents where users must stop
work because of difficulties with the Web site.
Ideally, the providers, developers, and stakeholders observe users as they try to accomplish tasks
at the site. Observing users interacting with the
site can show them the need for changes. Nothing
speaks louder about the quality of a Web site than
the experiences of its users.
Using a method called think aloud, the UE encourages users to talk about their expectations and
reactions while they work with the Web site. The
output is metric data on users success accompanied by anecdotal data, the users own comments
on what they were doing and why and what they
think of the site. UEs are specially trained to put
users at ease during observations and to facilitate
the users evaluation experience, without telling
users how to accomplish their tasks.
After usability observations, users often
complete satisfaction surveys on their experience
with the Web site. In an effort to foster reliability,
the UE only collects and processes feedback on
those elements of the site that the user has experienced. These surveys typically collect ratings
on a numerical scale to produce metric data. They
also offer opportunities for users to elaborate on
their experiences.
The UE never relies solely on satisfaction data,
but uses it to inform analysis of performance
data collected during user interactions with the
Web site. Users often report higher levels of

207

The Usability Engineering Behind User-Centered Processes for Web Site Development Lifecycles

satisfaction than would be expected from their

observed performance. This is one reason why
UEs interview users about their experience at
the end of the session. Another reason is to give
users opportunities to bring up points that no one
else has anticipated. It is common for developers
and other stakeholders to talk with users at this
point, too.
Formal usability observations can take place
wherever users interact with the Web site. With
increasing personal computer sales and the proliferation of mobile devices, people are viewing the
same sites from different locations: a conventional
office, a crowded cyber caf, or a subway train.
Each location impacts the user experience differently. A conventional office can have lights that
glare on the screen, ringing phones, and frequent
interruptions from colleagues. The cyber caf can
have background noise and poor lighting. The
subway can cause breaks in connectivity and the
need to view only small chunks of information
on a small mobile phone screen. Because of this,
when possible, UEs try to hold observations in
environments where the site will be used. Otherwise, when possible, they simulate the work
environment within the evaluation environment
to produce more valid findings.
It is unusual to hold formal usability observations at every development iteration. Indeed,
some projects find even one or two rounds of user
observations to be cost prohibitive. However, other
simple, less expensive processes incorporate users perspectives. In an expert review, one or more
UEs independently assess an interface against
their understanding of users; usability principles;
and applicable laws and standards. If more than
one UE has performed the expert review, the UEs
then meet to discuss and prioritize their findings
before discussing them with the rest of the team.
Another kind of expert review employs automated
accessibility tools, for example, InSight/InFocus
(SSB Technologies, 2004). Such tools inspect the
Web site code for the UI for conformance with

208

accessibility regulations. They identify violations

and recommend remedies.
A lifecycle typically includes several V&V
iterations. The product of usability evaluation is
a set of recommendations to improve the potential
for users success and satisfaction with the UI. The
principal benefit is an understanding of how users
interact with the site (Dumas & Redish, 1999).
A unique benefit of usability engineering is the
coordination of these recommendations with other
stakeholder, organizational, and project goals.
Without a UE, the team risks relying on processes that appear to be usability evaluation, but
actually fall short of delivering user-centered
products. Examples of pseudo-usability engineering include having stakeholders other than real
users provide feedback on interacting with the
Web site. Another example is simply asking users
to say what they like and dislike. The Web is rife
with misinformation about usability engineering.
Someone other than a UE trying to engineer usability based on such misinformation can arrive
at invalid recommendations that fail to improve
the user experience and, at worst, can degrade it.
Nothing takes the place of actual success and satisfaction data collected from representative users
and interpreted by trained usability engineers.

Summary
We maintain that usability engineering is rigorous, process-based, and addresses needs of stakeholders, such as site providers, as well as users.
We have set out a typical software engineering
process and discussed key usability engineering
contributions. We have demonstrated simultaneous, complementary activities whose products
benefit later activities without adversely affecting schedules. We have shown what would be
lacking without usability engineering and how
the potential of users success and satisfaction
increases with usability engineering. We stress

The Usability Engineering Behind User-Centered Processes for Web Site Development Lifecycles

that usability engineering is the means to providing successful and satisfactory experiences for
Web site users while fulfilling the goals of the
sites providers. The UEs contribution is integral
to Web site development.

draft. Retrieved July 13, 2005, from http://www.

w3.org/TR/WCAG20/#robust

References

Crandall, B., Kline, G., & Hoffman, R. R. (2006).

Working minds: A practitioners guide to cognitive
task analysis. Cambridge, MA: MIT Press.

Addelston, J. D., & OConnell, T. A. (2004). Usability and the agile project management process
framework. Cutter Consortium Agile Project
Management Executive Report, 5(10).
Addelston, J. D., & OConnell, T. A. (2005). Integrating accessibility into the spiral model of the
software development lifecycle. In Proceedings
of the 11th International Conference on HumanComputer Interaction (Volume 8 Universal
Access in HCI: Exploring New Dimensions of
Diversity). Las Vegas, NV: Mira Digital Publishers. CD-ROM.
Andre, A. D., & Wickens, C. D. (1995). When users want whats NOT best for them. Ergonomics
in Design, 4, 10-14.
Baddeley, A. (1990). Human memory. Boston:
Allyn & Bacon.
Bias, R. G., & Karat, C. M. (2005).

Justifying
cost-justifying usability. In R. G. Bias & D. J.
Mayhew (Eds.), Cost justifying usability: An update for the Internet age (2nd ed., pp. 2-16). San
Francisco: Elsevier.
Bolt, R. A. (1984). The human interface: Where
people and computers meet. Toronto, Canada:
Wadsworth.
Brown, E. L., & Deffenbacher, K. (1979). Perception and the senses. New York: Oxford University
Press.
Caldwell, B., Chisholm, W., Slatin, J., & Vanderheiden, G. (Eds.). (2005, June 30). Web content
accessibility guidelines 2.0 (WCAG 2.0). Working

Carroll, J. M. (1990). Mental models in humancomputer interaction. In M. Helander (Ed.),

Handbook of human-computer interaction (pp.
45-65). New York: Elsevier B.V.

Czaja, S. J., & Lee, C. C. (2003).

Designing computer systems for older adults. In J. A. Jacko &

A. Sears (Eds.), The human-computer interaction
handbook: Fundamentals, evolving technologies,
and emerging applications (pp. 413-427). Mahwah, NJ: Erlbaum.
Dumas, J. S., & Redish, J. C. (1999). A practical
guide to usability testing (Rev. ed.). Portland,
OR: intellect.
Garrett, J. J. (2002). The elements of user experience: User-centered design for the Web. Thousand
Oaks, CA: New Riders.
Gordon, S. E., & Gill, R. T. (1997). Cognitive
task analysis. In C. Zsambok & G. Klein (Eds.),
Naturalistic decision making (pp. 131-140). Hillsdale, NJ: Erlbaum.
Hackos, J., & Redish, J. (1998). User and task
analysis for interface design. New York: Wiley.
Head, A. J. (2003). Personas: Setting the stage
for building usable information sites. Retrieved
July 16, 2005, from http://www.infotoday.com/
online/jul03/head.shtml
Hix, D., & Hartson, H. R. (1993). Developing user
interfaces: Ensuring usability through product &
process. New York: Wiley.
Hix, D., & OConnell, T. A. (2005). Usability engineering as a critical process in designing for accessibility. In Proceedings of the11th International
Conference on Human-Computer Interaction
(Volume 8 Universal Access in HCI: Exploring

209

The Usability Engineering Behind User-Centered Processes for Web Site Development Lifecycles

New Dimensions of Diversity). Las Vegas, NV:

Mira Digital Publishers. CD-ROM.
Horton, W. (1994). The icon book. New York:
Wiley.

model for students to experience usability issues

in the federal workplace. Journal of Informatics
Education Research, 6(3), 57-78.
Mayer, R. E. (1992). Thinking, problem solving,
cognition (2nd ed.). New York: W. H. Freeman.

Institute of Electrical and Electronics Engineers

(IEEE). (1990). IEEE standard glossary of software engineering terminology (IEEE Standard
610.12-1990). New York: IEEE.

Mayhew, D. J. (1992). Principles and guidelines

in software user interface design. Englewood
Cliffs, NJ: Prentice Hall.

International Organization for Standardization

(ISO). (1998). Guidance on usability. (ISO 924111). Geneva, Switzerland: ISO.

Mayhew, D. J. (1999). The usability engineering

lifecycle: A practitioners handbook for user interface design. San Francisco: Morgan Kaufmann.

International Organization for Standardization

(ISO). (2002). Codes for the representation of
names of languages. (ISO 639). Retrieved July
20, 2005, from http://www.ics.uci.edu/pub/ief/
http/related/iso639.txt

Mayhew, D. J., & Tremaine, M. M. (2005). A basic

framework. In R. G. Bias, & D. J. Mayhew (Eds.),
Cost-justifying usability: An update for the internet age (pp. 41-101). San Francisco: Elsevier.

Internet World Stats Market Research. (2005).

Internet User Statistics and Population Stats for
51 Countries and Regions North America,
Central America, South America and the Caribbean. Retrieved July 11, 2005, from http://www.
internetworldstats.com /stats2.htm#north
Johnson-Laird, P. N. (1983). Mental models. Cambridge, UK: Cambridge University Press.
Kalin, S. (1999, April 1). Usability mazed and
confused. CIO Web Business Magazine. Retrieved
July 16, 2005, from http://www.cio.com/archive/
webbusiness/040199_use.html#price
Kirwan, B., & Ainsworth, L. K. (Eds.). (1992).
A guide to task analysis. Philadelphia: Taylor &
Francis.
Koyani, S., Bailey, R. W., Nall, J. R., Allison, S.,
Mulligan, C., Bailey, K., et al. (2003). Researchbased Web design & usability guidelines (NIH
Publication No. 03-5424). Washington, DC: U.S.
Department of Health and Human Services
Lazar, J., Murphy, E. D., & OConnell, T. A. (2004).
Building university-government collaborations: A

210

Miller, G. A. (1956). The magical number seven

plus or minus two: Some limits on our capacity
for processing information. Psychological Review,
63, 81-97.
Murphy, E., Marquis, K., Nichols, E., Kennedy,
K., & Mingay, D. (2001). Refining electronic datacollection instruments and data-collection tools
through usability testing. Research in Official
Statistics, 4(2), 23-33.
Norman, D. A., & Draper, S. W. (Eds.). (1986).
User centered system design. Hillsdale, NJ:
Erlbaum.
OConnell, T. A. (in press). The why and how of
senior-focused design: 21st century user interface
design strategies to empower users born in the
20th century. In J. Lazar (Ed.), Universal usability.
West Sussex, UK: Wiley.
Pruitt, J., & Grudin, J. (2003). Personas: Practice
and theory. ACM. Retrieved July 20, 2005, from
http://www.aiga.org/resources/content/9/7/8/
documents/pruitt.pdf
Redish, J., & Wixon, D. (2003). Task analysis. In J.
A. Jacko & A. Sears (Eds.), The human-computer

The Usability Engineering Behind User-Centered Processes for Web Site Development Lifecycles

interaction handbook: Fundamentals, evolving

technologies, and emerging applications (pp.
922-940). Mahwah, NJ: Erlbaum.
Refsnes Data. (2005). Browser statistics. Retrieved
July 11, 2005, from http://www.w3schools.com/
browsers_stats.asp
Rehabilitation Act, 29 U.S.C. 794d (1973). As
amended by the Workforce Investment Act, PL
105-220 (1998).
Schraagen, J. M., Chipman, S. F., & Shalin, V.
L. (2000). Cognitive task analysis. Mahwah, NJ:
Erlbaum.
Shneiderman, B. (1998). Designing the user
interface: Strategies for effective human-computer interaction (3rd ed.). Reading, MA: Addison
Wesley Longman.
Shneiderman, B., & Plaisant, C. (2004). Designing the user interface: Strategies for effective
human-computer interaction (4th ed.). Reading,
MA: Pearson Addison Wesley.
SSB Technologies. (2004). InSight (Version 4.0)/
InFocus (Version 4.0). [Computer software]. San
Francisco: SSB Technologies.
Travis, D. (1991). Effective color displays: Theory
and practice. New York: Academic Press.

Van Der Veer, G. C., & Del Carmen Puerta Melguizo, M. (2003). Mental models. In J. A. Jacko &
A. Sears (Eds.), The human-computer interaction
handbook: Fundamentals, evolving technologies,
and emerging applications (pp. 52-80). Mahwah,
NJ: Erlbaum.
Vicente, K. J. (1999). Cognitive work analysis:
Toward safe, productive, and healthy computerbased work. Mahwah, NJ: Erlbaum.
Web accessibility initiative (WAI). Retrieved July
16, 2005, from http://www.w3.org/WAI/
Whiteside, J., Bennett, J., & Holtzblatt, K. (1990).
Usability engineering: Our experience and evolution. In M. Helander (Ed.), Handbook of humancomputer interaction (pp. 791-817). New York:
Elsevier B.V.
World Wide Web Consortium (W3C). (2004).

Endnote
1

This material is released to inform interested

parties of ongoing research and to encourage
discussion of work in progress. The views
expressed are those of the authors and not
necessarily those of the U.S. Census Bureau.

This work was previously published in Human Computer Interaction Research in Web Design and Evaluation, edited by P.
Zaphiris, pp. 1-21, copyright 2007 by Information Science Publishing (an imprint of IGI Global).

211

212

Chapter 1.16

Personalization Systems and

Their Deployment as Web Site
Interface Design Decisions
Nanda Kumar
Baruch College, The City University of New York, USA

Abstract

Introduction

This chapter reviews the different types of personalization systems commonly employed by Web
sites and argues that their deployment as Web
site interface design decisions may have as big
an impact as the personalization systems themselves. To accomplish this, this chapter makes a
case for treating Human-Computer Interaction
(HCI) issues seriously. It also argues that Web
site interface design decisions made by organizations, such as the type and level of personalization
employed by a Web site, have a direct impact on
the communication capability of that Web site.
This chapter also explores the impact of the
deployment of personalization systems on users
loyalty towards the Web site, thus underscoring the
practical relevance of these design decisions.

Organizations, in anticipation of the emergence

of Web interface as a major point of contact between companies and customers, are beginning
to employ a wide variety of technologies to build
meaningful relationships with their customers.
While Web interface may not be the only point
of contact that customers use, organizations are
aware of the advantages of using the Web to cater
to the needs of the customers. This concept of
self-service not only reduces costs for the company in the long run, but also increases customer
satisfaction by addressing the transactional and the
relational needs of the customer. In this chapter,
I review the different types of personalization
systems commonly employed by Web sites and
argue that their deployment as Web site interface
design decisions may have as big an impact as the
personalization systems themselves. To accom-

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Personalization Systems and Their Deployment as Web Site Interface Design Decisions

plish this, I make a case for treating HCI issues

seriously and argue that Web site interface design
decisions made by organizations, such as the type
and level of personalization employed by a Web
site, has a direct impact on the communication
capability of that Web site. I also focus on the
practical relevance of these Web site design decisions by examining their effect on users loyalty
towards the Web site.

its customers to the point where any and all of

these dialogues are seamlessly integrated with the
databases historical and transactional information. Based on the data stored in these databases
and recent history (the pages customers viewed in
the last session), Web sites automatically attempt
to improve their organization and presentation of
content. These Web sites, armed with a host of
appropriate tools including intelligent agents,
recommendation engines and the like attempt
to anticipate the context of the interaction with
their customers and personalize each customers
shopping experience (Andre & Rist, 2002; Billsus,
Brunk, Evans, Gladish, & Pazzani, 2002).
Personalization is a process of providing
special treatment to a repeat visitor to a Web site
by providing relevant information and services
based on the visitors interests and the context
of the interaction (Chiu, 2000; Cingil, Dogac, &
Azgin, 2000). Personalization is needed to successfully manage customer relationships, promote
the right product the customer is interested in, and

Background: Personalization
Systems
Most of the technologies and tools that companies
use to manage their relationship with their customers usually fall under the banner of Customer
Relationship Management (CRM) System. Even
though personalization is just one piece of the
CRM pie, it is a very crucial piece as effective
personalization significantly enhances the ability
of the organization to initiate a discourse with

Figure 1. Overview of personalization techniques

Collection of
Visitor Information

Personalization Processes

Implicit
Profiling

Three Filtering Techniques:

Simple, Content Based and
Collaborative Filtering

Explicit
Profiling
Legacy
Data

Business Rules

Dynamic Assembly and Display

of Content as Web Page

213

Personalization Systems and Their Deployment as Web Site Interface Design Decisions

manage content. Most of the advanced personalization might require sophisticated data mining
techniques and the ability to display dynamic
content without seriously compromising system
resources (dynamic display of content will usually
mean increased download time).
There are a few well-known techniques for
personalization. Rules-based personalization
modifies the content of a page based on specific
set of business rules. Cross-selling is a classic
example of this type of personalization. The key
limitation of this technique is that these rules must
be specified in advance. Personalization that uses
simple filtering techniques determines the content that would be displayed based on predefined
groups or classes of visitors and is very similar to
personalization based on rules-based techniques.
Personalization based on content-based filtering
analyzes the contents of the objects to form a
representation of the visitors interest (Chiu,
2000). This would work well for products with
a set of key attributes. For example, a Web site
can identify the key attributes of movies (VHS,
DVD) such as drama, humor, violence, etc., and
can recommend movies to its visitors based on
similar content. Personalization based on collaborative filtering offers recommendations to
a user based on the preferences of like-minded
peers. To determine the set of users who have
similar tastes, this method collects users opinion on a set of products using either explicit or
implicit ratings (Chiu, 2000). Please see Figure
1 for an illustration of how a Web site could use
all three personalization methods to best serve
the customer.

Deployment of
Personalization Systems
Profiling
An intelligent way to make the Web site adaptive is to use not only the information provided

214

by the user (such as rating the music and log-in

information), but also information that could
be collected based on the click-stream trail left
behind by the user. These two different sources
of collecting information about the consumer are
known as explicit and implicit profiling. As the
name implies, explicit profiling collects information about a user by directly asking him or her
information about himself or herself and product
likes and dislikes. This information is collected
over a period of time and is stored in the customer
database as a profile. Typically, the user would
need to log-in in order for the Web site to access
the profile and provide personalized content. Even
though cookies can be used to store this information on a users hard disk, companies prefer to use
the log-in approach as this allows the Web site to
identify the unique visitor (cookies wont help if
the computer is shared within a family or if the
customer accesses the Web site from a different
location say from the office).
Implicit profiling typically tracks the actual
behavior of the customer while browsing the Web
site. This method of collecting information is
transparent to the user. While less intrusive, this
method of collecting information has implications
for the users privacy. Typically, information is
collected about the pages the consumer visited, the
products he or she looked at and the time that the
user spent on these pages. If a (brick and mortar)
company has good information systems, the data
from explicit and implicit profiling can be merged
with the off-line customer information (see legacy
user data in Figure 1) to effectively present a
seamless Web interface to the customer.
Ideally, a company should use all sources of
information it has about the customer. However,
when a user visits a shopping Web site (even
a repeat user), it would be unsound business
practice to expect the user to log-in every time
to access personalized content. Hence, a good
Web site would use implicit profiling and make
a few assumptions about the likes and dislikes of
the customer to provide adaptive content to the

Personalization Systems and Their Deployment as Web Site Interface Design Decisions

customer. For example, if a customer visits a

specific product page, it is a good idea to assume
that the customer is interested in that particular
product and provide content personalized to that
users need. Of course, in most cases, even if
the user logs in, the Web site may have little else
other than previous purchase history if the user
has not provided any specific information on the
products he or she likes.

Personalization as an Interactive
Dialogue Between a Web Site and Its
Users
The level and extent of personalization offered
by the Web site will have an effect on the communication characteristics of the media. This
research argues that different levels of support
provided for personalization will specifically impact on the adaptiveness [similar to contingency
used by (Burgoon et al., 2000)] of the Web site.
This is best illustrated by discussing a real life
example using Amazon.com. Appendices 1 to 3
include three screen shots that show the different ways Amazon.com attempts to personalize
the experience of the customer. When the user
enters the Web site, he or she is invited to log
in if desired. Once the user logs in, Appendix 1
shows the Web page that is dynamically created
by Amazon.com. This page recommends products to the user based on past purchase history
and on the explicit ratings provided by the user
to a set of select items. Appendix 2 shows the
product page for a book the user is interested in.
The column on the left hand side of this page
shows the associated related content about the
product that is displayed on this page. Appendix
3 shows the page tailor-made for the user based
on his recent browsing history and past purchase
history. Of course, the scenario described above
assumes that the user logged into the Web site at
the outset. An intelligent Web site can still adapt
its content in its product page by assuming that

the user is interested in the product he or she is

browsing. Accordingly, the product page shown
in screen shot 2 can be personalized even without
an explicit log-in by the user.
If the same user were to shop for the book that
he is interested in a physical store, he might have
approached the sales clerk (or even a friend he had
taken along for the shopping trip) for help locating
the product. Now, when he mentions to his friend
that he is interested in this specific book, music or
movie, then it is possible to imagine a conversation happening along the lines discussed above.
Of course, the above discourse with the Web site
is limited by the need for a shared context. The
conversation will not be totally indeterminable in
terms of context and content and may not move
along in any arbitrary direction as is possible in
a conversation with a friend. But, this research
argues that there are enough cues in the discourse
initiated by the personalization system of Amazon.
com that is enough to give the user the impression that the conversation is contingent within
that shared context.

Virtual Communities as Personalized

Communities
To enhance the relationship with the customers,
companies can also provide support for virtual
communities, as this will facilitate access to
free-flowing and unstructured information beyond what is provided by the computer agents
(Jones, 1997; Preece, 2001, 2002). For example,
companies can aggregate the opinions of consumers on a particular product and present them to
a new user who is browsing that product page.
Depending on the level of support provided by the
Web site, the new user can also get in touch with
another consumer he or she might identify with,
as is the case with Amazon.com. A recent study
(Brown, Tilton, & Woodside, 2002) shows that
community features create value for a shopping
Web site. Their study showed that community

215

Personalization Systems and Their Deployment as Web Site Interface Design Decisions

users accounted for about one-third of the visitors

to the e-tailing sites surveyed and that they also
generated two-thirds of the sales (2000 transactions worth one million dollars). Practioners have
long argued that having a vibrant community in the
form consumer reviews is crucial for the success
of e-commerce Web sites such as Amazon.com
and Ebay.com (Brown et al., 2002; Kirkpatrick,
2002). Hence providing support for consumer
reviews facilitates formation of one type of virtual
community and integrating high level of support
(user rating and information about the user) for
consumer reviews on the product page increases
personalization afforded by Web sites as these
are relevant comments and opinions by different
users presented on the product page.

Deployment of Personalization as
Web Site Design Decisions
Reeves, Nass and their colleagues at the Center
for the Study of Language and Information at
Stanford have shown that even experienced users
tend to respond to computers as social entities
(Nass, Lombard, Henriksen, & Steur, 1995; Nass,
Moon, Fogg, Reeves, & Dryer, 1995; Nass & Steur,
1994). These studies indicate that computer users
follow social rules concerning gender stereotypes
and politeness, and that these social responses
are to the computer as a social entity and not to
the programmer. When explicitly asked by the
researchers, most users consistently said that
social responses to computers were illogical and
inappropriate. Yet, under appropriate manipulation, they responded to the computer as though it
were a social entity. This, in fact, is the essence
of the Theory of Social Response (Moon, 2000;
Reeves et al., 1997). Thus I argue that there is
value in conceptualizing the Web site as a social
actor and that the Web site can be equated to the
agents mentioned above in terms of source
orientation. There are several points-of-contact
between a Web site and its users that will result

216

in responses by the users not unlike the way they

would respond to a social interaction.
In the light of the above discussions, Web sites
should also view deployment of personalization
systems as important Web site design decisions
that will facilitate or hinder this interactive dialogue between a Web site and its users. In a recent
study conducted by this author, the personalization
systems deployed by four major Web sites (Amazon.com, BarnesandNoble.com, CDNow.com and
Chapters.ca) were compared along with the Web
sites support for virtual communities. The results
of the study showed strong support showing that
level of support for personalization systems had
an impact on customer loyalty. The results also
suggested that Web sites should pay close attention to the way they deploy these personalization
systems. Specifically, the results showed that by
implementing consumer review support on the
product page (as done by Amazon.com), we could
simulate support for personalization systems in
the absence of personalization systems.

Conclusion
In practice, it is advantageous for the Web sites to
offer some form of support for personalization or
virtual community as this makes the Web site to
be perceived as more adaptive. This will facilitate better communication between the Web site
and the shoppers, thus leading to higher levels
of customer loyalty. Companies do understand
that in practical terms it takes a lot more money
and effort to acquire a new customer than to keep
an existing customer and arguments presented in
this chapter throws new light on the role of support for personalization and consumer reviews
in increasing customer loyalty.
Good personalization systems can be very
expensive to set up. Enabling an e-commerce
Web site with the necessary tools to build a vibrant
community costs little (especially when compared

Personalization Systems and Their Deployment as Web Site Interface Design Decisions

to personalization systems), as the community

members provide the content. The results of the
study offer evidence that Web sites by providing
support for consumer reviews could reap the benefits in terms of increased adaptiveness despite
offering very little personalization. However, the
ways and means of implementing support for
personalization and virtual communities deserve
further research. Future research should examine
the implementation of these features in finer detail. This will help the organizations understand
more in depth the trade-offs involved in providing
different types and levels of personalization and
virtual communities.

References
Andre, E., & Rist, T. (2002). From adaptive hypertext to personalized Web companions. Communications of the ACM, 45(5), 43-46.
Billsus, D., Brunk, C. A., Evans, C., Gladish, B.,
& Pazzani, M. (2002). Adaptive interfaces for
ubiquitous Web access. Communications of the
ACM, 45(5), 34-38.
Brown, S., Tilton, A., & Woodside, D. (2002).
Online communities pay. McKinsey Quarterly,
19(1), 17.
Burgoon, J. K., Bonito, J. A., Bengtsson, B., Cederberg, C., Lundeberg, M., & Allspach, L. (2000).
Testing the interactivity model: Communication
processes, partner assessments, and the quality
of collaborative work. Journal of Management
Information Systems, 16(3), 33-56.
Chiu, W. (2000). Web site personalization. IBM
High-Volume Web Site Team, WebSphere Software Platform.

Cingil, I., Dogac, A., & Azgin, A. (2000). A

broader approach to personalization. Communications of the ACM, 43(8), 136-141.
Jones, Q. (1997). Virtual-communities, virtual
settlements & cyber-archaeology: A theoretical
outline. Journal of Computer Mediated Communication, 3(3).
Kirkpatrick, D. (2002). Esther Dyson: Living the
networked life. Fortune, 145(11), 168-172.
Moon, Y. (2000). Intimate exchanges: Using computers to elicit self-disclosure from consumers.
Journal of Consumer Research, 26, 323-339.
Nass, C., Lombard, M., Henriksen, L., & Steur,
J. (1995). Anthropocentrism and computers.
Behavior and Information Technology, 14(4),
229-238.
Nass, C., Moon, Y., Fogg, B. J., Reeves, B.,
& Dryer, C. (1995). Can computers be human
personalities? International Journal of HumanComputer Studies, 43(2), 223-239.
Nass, C., & Steur, J. (1994). Voices, boxes and
sources of messages: Computers and social actors. Human Communication Research, 19(4),
504-527.
Preece, J. (2001). Sociability and usability: Twenty
years of chatting online. Behavior and Information Technology, 20(5), 347-356.
Preece, J. (2002). Supporting community and
building social capital. Communications of the
ACM, 45(4), 37-39.
Reeves, B., & et al. (1997). The media equation.
Stanford, CA: CSLI Publications.

217

Personalization Systems and Their Deployment as Web Site Interface Design Decisions

Appendix 1

Appendix 1: Screen Shot 1

Appendix 2

218

Personalization Systems and Their Deployment as Web Site Interface Design Decisions

Appendix 3

This work was previously published in Web Systems Design and Online Consumer Behavior, edited by Y. Gao, pp. 147-155,
copyright 2005 by IGI Publishing, formerly known as Idea Group Publishing (an imprint of IGI Global).

219

220

Chapter 1.17

Communication +
Dynamic Interface =
Better User Experience
Simon Polovina
Sheffield Hallam University, UK
Will Pearson
Sheffield Hallam University, UK

Introduction
Traditionally, programming code that is used
to construct software user interfaces has been
intertwined with the code used to construct the
logic of that applications processing operations
(e.g., the business logic involved in transferring
funds in a banking application). This tight coupling
of user-interface code with processing code has
meant that there is a static link between the result
of logic operations (e.g., a number produced as the
result of an addition operation) and the physical
form chosen to present the result of the operation to the user (e.g., how the resulting number
is displayed on the screen). This static linkage
is, however, not found in instances of natural
human-to-human communication.
Humans naturally separate the content and
meaning that is to be communicated from how

it is to be physically expressed. This creates the

ability to choose dynamically the most appropriate
encoding system for expressing the content and
meaning in the form most suitable for a given
situation. This concept of interchangeable physical output can be recreated in software through
the use of contemporary design techniques and
implementation styles, resulting in interfaces that
improve accessibility and usability for the user.

Background
This section accordingly reviews certain theories
of communication from different disciplines and
how they relate to separating the meaning being
communicated from the physical form used to
convey the meaning.

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Communication + Dynamic Interface = Better User Experience

Claude Shannon (1948), a prominent researcher

in the field of communication theory during the
20th century, put forward the idea that meaning is
not transmitted in its raw form, but encoded prior
to transmission. Although Shannon was primarily
working in the field of communication systems
and networks such as those used in telephony,
his theory has been adopted by those working
in the field of human communications. Shannon
proposed a five-stage model describing a communication system. Beginning with the first stage
of this model, the sender of the communication
creates some content and its intended meaning.
In the second stage, this content is then encoded
into a physical form by the sender and, in the third
stage, transmitted to the receiver. Once the communication has been received by the receiver from
the sender, it is then at its fourth stage, whereby it
is decoded by the receiver. At the fifth and final
stage, the content and meaning communicated by
the sender become available to the receiver.
An example of how Shannons (1948) model
can be applied to human communication is speechbased communication between two parties. First,
the sender of the communication develops some
thoughts he or she wishes to transmit to the
intended receiver of the communication. Following on from the thought-generation process,
the thoughts are then encoded into sound by the
vocal cords, and further encoded into a particular
language and ontology (i.e., a set of mappings
between words and meaning) according to the
senders background. This sound is subsequently
transmitted through the air, reaching the receivers
ears where it is decoded by the receivers auditory
system and brain, resulting in the thoughts of the
sender finally being available to the receiver.
This split between meaning, its encoding,
and the physical transmission of the meaning is
recognised in psychology. Psychology considers
that there are three stages to receiving data: (a)
the receiving of sensory stimuli by a person, (b)
the perception of these stimuli into groups and
patterns, and (c) the cognitive processing of the

groups and patterns to associate cognitively the

meaning with the data (Bruno, 2002). Thus, for
example, a receiver may see a shape with four
sides (the data) and associate the name square
(the meaning) with it. There is accordingly a
split between the input a person receives and the
meaning he or she cognitively associates with
that input.
Consider, for example, the words on this
page as an example of the psychological process
through which meaning is transmitted. The first
stage of the process is where the reader receives
sensory stimuli in the form of black and white
dots transmitted to the eyes using light waves
of varying wavelength. Upon the stimuli reaching the reader, the brain will perceptually group
the different dots contained within the received
stimuli into shapes and, ultimately, the reader will
cognitively associate the names of letters with
these shapes and extract the meaning conveyed
by the words.
Semiotics, which is the study of signs and their
meanings (French, Polovina, Vile, & Park, 2003;
Liu, Clarke, Anderson, Stamper, & Abou-Zeid,
2002), also indicates a split between meaning
and its physical presentation. Within semiotics,
the way something is presented, known as a sign,
is considered to be separate from the meaning it
conveys. Accordingly, in semiotics there are three
main categories of signs: icons, indexes, and symbols. This delineation is, however, not mutually
exclusive as a particular sign may contain elements
of all categories. Vile and Polovina (2000) define
an icon as representative of the physical object it
is meant to represent; a symbol as being a set of
stimuli, that by agreed convention, have a specific
meaning; and indexes as having a direct link to a
cause, for example, the change of a mouse pointer
from an arrow shape to an hourglass to reflect the
busy state of a system.
This classification of the physical representation according to its relationship with the content
and meaning it conveys provides further opportunities to distinguish content and meaning

221

Communication + Dynamic Interface = Better User Experience

from its physical presentation, and to classify the

different elements of presentation. For example, a
shop selling shoes may have a sign outside with
a picture of a shoe on it. The image of the shoe is
the sign, or the physical presence of the meaning,
which in this case is an icon, while the fact that it
is a shoe shop is the intended meaning. Equally,
this could be represented using the words shoe
shop as the physical sign, in this case a symbol of
the English language, while the meaning is again
that of a shoe shop.
This split of content and meaning from its
physical presentation, which occurs naturally
in human communication, allows for the same
content and meaning to be encoded in a variety
of different forms and encoding methods. For
example, the meaning of no dogs allowed can
be encoded in a variety of visual images. For
instance, there might be (a) an image of a dog
with a cross through it, (b) the words no dogs
allowed, (c) an auditory sequence of sounds
forming the words no dogs allowed, or (d) the
use of tactile alphabets such as Braille, which is
used to encode printed writing into a form for
the blind. However the content and meaning is
conveyed, it remains the same regardless of how
it is physically presented.

Software Architectures for

Content Separation
For the true separation of presentation from content to occur therefore in software, the content
(namely the data or information itself as well as the
applications operations, i.e., its business logic as
indicated earlier) is stored in a neutral format. This
neutrality is achieved when the content is untainted
by presentation considerations. This allows any
given content to be translated and displayed in
any desired presentation format (e.g., through an
HTML [hypertext markup language] Web browser
such as Microsofts Internet Explorer, as an Adobe
Acrobat PDF [Portable Document Format], as an

222

e-book, on a mobile phone, on a personal digital

assistant [PDA], or indeed on any other device not
mentioned or yet to be invented). The theories of
detaching content and meaning from its physical
presentation thus give a framework to separate
content from presentation. Once that conceptual
separation can be made, or at least continually
realisable ways toward it are achieved, then this
approach can actually be deployed in the design
and implementation of computer systems.
There are a number of methods offered by contemporary software languages and architectures
to achieve this detachment between the content
and meaning, and how the content can thus be
displayed. In the sphere of Web development, the
extensible markup language (XML) is one such
example. XML provides a useful vehicle for separating presentation from content (Quin, 2004a).
Essentially, unlike HTML in which the tags are
hard coded (e.g., Head, Body, H1, P, and so forth),
XML allows designers or developers to define their
own tags particular to their domain (e.g., Name,
Address, Account-number, Transactions, Debits,
Credits, and so forth in, say, a banking scenario).
How this content is presented has, of course, to
be defined by the designer or developer; he or she
can no longer rely on the browser to format it by
simply recognising the hard-coded HTML tags.
The extensible stylesheet language (XSL) is the
vehicle to achieve this (Quin, 2004b). Equally, the
scaleable vector graphics (SVG) format, based
on XML, is another World Wide Web format
capable of separating content and meaning from
presentation. SVG specifies drawing objects,
their dimensions, colour, and so forth, but leaves
the determination of presentation modality to
the client viewer application (Ferraiolo, Jun, &
Jackson, 2003).
Within enterprise systems, this separation can
be achieved through the use of object-orientated
and n-tier design methodologies. Object orientation works through its embodiment of the four
goals of software engineering (Booch, 1990;
Meyer, 1988; Polovina & Strang, 2004). These

Communication + Dynamic Interface = Better User Experience

four goals of software engineering, namely (a)

abstraction, (b) cohesion, (c) loose coupling, and
(d) modularity, determine the principled design of
each object that makes up the system. They seek
to ensure that the object only performs the functions specific to its role, for example, to display
a piece of information or to perform a calculation. Accordingly, these goals seek to ensure that
presentation objects only present the information,
while logic objects only perform calculations and
other business-logic operations. These content
objects thus do not concern themselves with
how the information is presented to the user;
instead these content objects communicate their
information via presentation objects to perform
this function.
In addition to embodying the four goals of
software engineering, object orientation builds
on these by providing three further principles:
(a) encapsulation, (b) inheritance, and (c) polymorphism (Booch, 1990; Meyer, 1988; Polovina
& Strang, 2004). Inheritance allows an object
to inherit the characteristics and behaviours of
another object. Utilising this feature, it is possible
to extend the functionality of an object to include
new functionality, which may be new buttons or
other interface elements within a user interface.
Polymorphism is used to select an object based
on its ability to meet a given set of criteria when
multiple objects perform similar functions. For
example, there may be two objects responsible
for displaying the same interface element; both
display the same content and meaning, but using
different languages. In this scenario, the concept
of polymorphism can be used to select the one appropriate for the language native to the user. Thus,
object-orientated design can be used to naturally
compliment the process of separating content and
meaning from its method of presentation.
A common practice within the field of software engineering is to base software designs on
common, predefined architectures, referred to as
patterns. One pattern, which lends itself well to
the separation of content and meaning from its

method of presentation, is the n-tier architecture.

The n-tier architecture separates the objects used
to create the design for a piece of software into layers (Fowler, 2003). The objects contained within
each layer perform a specific group of functions,
such as data storage. In the three-tier architecture,
for example, one layer is responsible for handling
the softwares input and output with the user,
another handles its business-logic processes,
and the final layer handles the persistent storage
of information between sessions of the software
being executed. Through the use of an n-tier
architecture and the separation of the different
areas of an applications design that it creates, it
is possible to separate the content from its mode
of presentation within software design.
Software engineerings ability to separate
content and meaning from its physical presentation can be aided by some contemporary implementation methods. These methods are based on
component architectures that aim to create reusable segments of code that can be executed. This
enhances object orientation, which seeks to create
reusable segments of software at the source-code
level. While there is not much difference in the
design, having reusable segments of executable
code translates to faster time to change segments,
further enhancing the plug-and-play nature of
software. Microsofts Component Object Model
(COM) is a client-side Windows-based component
architecture (Microsoft Corporation, 1998). This
architecture enables programs to be built as individual components that are linked together using
a client application to form a complete software
program. This approach to software implementation provides the ability to construct similar
pieces of software using the same components,
where the functionality is common between the
pieces of software. For example, if the storage
and logic elements of a piece of software were
to remain the same but the user interface were
to be changed due to the differing needs of user
groups, the same components forming the storage
and logic sections could be used for all versions of

223

Communication + Dynamic Interface = Better User Experience

the software. Furthermore, this could occur while

different components were created to provide the
different user interfaces required. This method
would reduce the time taken to build and deploy
the software amongst a group of diverse users.
Another implementation technique, built
around distributed components located on different physical machines, are Web services (MacDonald, 2004). Instead of the components used
to build the software being located on the same
machine, different components can be placed
on different machines. This results in users being able to share and access the same physical
instance of objects. This enhances COM, which
although it gives access to the same components,
forces each user to use different instances of them.
One advantage of Web services is that they allow
the existence of different user interfaces while
letting users access the same physical objects
used for the logic and storage processes. This
type of deployment will ensure that all users are
accessing the same data through the same logic
processes, but allows the flexibility for each user
or user group to use an interface that is the most
optimal for their needs, be they task- or devicedependant needs.

The Human-computer
interaction Benefits
The human race rarely uses fixed associations
between content or meaning and its physical
representation. Instead, people encode the meaning into a form appropriate for the situation and
purpose of the communication. Communication
can be encoded using different ontologies such
as different languages and terminology. Communication is thus able to take different physical
channels (e.g., sound through the air, or writing
on paper), all of which attempt to ensure that the
content or meaning is communicated between the
parties in the most accurate and efficient manner available for the specific characteristics of

224

the situation. Currently, this is not the case with

computer interfaces; contemporary interfaces
instead tend to adopt a one size fits all approach
for the majority of the interface.
In taking this one-size-fits-all approach, content and meaning may not be transmitted to the
user in the most accurate form, if it is communicated at all. The characteristics of the situation
and participants are not taken into account. This
makes the interface harder to use than might be,
if it can be used at all. Some users, such as those
with a sensory disability or those with a different native language, may not be able to access
the information as it has been encoded using an
inaccessible physical form (e.g., visual stimuli are
inaccessible for the blind). Or it has been encoded
using a foreign language, which the user does not
understand. This immediately prevents the user
from accessing the content and meaning conveyed
by that form of presentation.
Equally, terminology can be prohibitive to the
ease of use of a user interface. The set of terms
that we know the meaning for (i.e., ontology) is
based on factors such as the cultural, educational,
and social background of the user as well as the
geographic area the user inhabits. This leads to
different groups of people being familiar with
different terminology from those in other groups,
although there is some degree of overlap in the
ontologies used by the different groups. The user
is forced to learn the terminology built into the
interface before they can extract the meaning that
it conveys. This imposes a learning curve on the
user, unless they are already familiar with the
particular set of terms used. Hence, by using a
one-size-fits-all user interface, some users will
find it difficult or impossible to use.
By utilising the facilities offered by contemporary software-engineering practices, it is
possible to avoid this one-size-fits-all approach
and its inherent disadvantages in terms of humancomputer interaction. By allowing the encoding
scheme used to present software interfaces to
change with different users, interfaces will begin

Communication + Dynamic Interface = Better User Experience

to mimic the processes used to encode content

and meaning that are found in natural human-tohuman communication. This change will result in
interfaces that are accessible by those who could
not previously access them, and will also result in
greater ease of use for those who previously had
to learn the terminology used within the interface,
hence improving interface usability.

future trends
One emerging trend is the use of explicit user
modeling to modify the behaviour and presentation of systems based on a users historic use of
that system (Fischer, 2001). Explicit user modeling
involves tracking the preferences and activities of
a user over time, and building a model representing that behaviour and associated preferences.
This, coupled with the concept of presenting the
content and meaning in the form most suitable for
the user, holds the ability to tailor the content to a
specific individuals needs. By monitoring how a
user receives different types of information over
time, a historic pattern can be developed that can
subsequently be used to present the content and
meaning based on an individuals actual requirements, not on a generalized set of requirements
from a specific group of users.

Conclusion
Currently, by entwining the association between
content and meaning and the physical form used to
represent it, software user interfaces do not mimic
natural human-to-human communication. Within
natural communication, the content and meaning
that is to be conveyed is detached from its physical form, and it is only encoded into a physical
form at the time of transmission. This timing of
the point at which the content and meaning are
encoded is important. It gives the flexibility to
encode the content and meaning in a form that

is suitable for the characteristics of the situation

(e.g., the channels available, the languages used by
the parties, and the terminology that they know).
This ensures that humans communicate with each
other in what they consider to be the most appropriate and accurate manner, leading to encoding
schemes from which the parties can access the
content and meaning in an easy method.
This is not currently the case for software
user interfaces, which use a too tightly coupled
association between the content and meaning and
the physical form used to encode it. By utilising
contemporary Web-based or object-orientated
component architectures, this problem of fixed
encoding schemes can be overcome. Therefore,
software user interfaces can more closely mimic
natural language encoding and gain all the benefits
that it brings.

References
Booch, G. (1990). Object oriented design with
applications. Redwood City, CA: Benjamin-Cummings Publishing Co., Inc.
Bruno, F. (2002). Psychology: A self-teaching
guide. Hoboken, NJ: John Wiley & Sons.
Ferraiolo, J., Jun, F., & Jackson, D. (Eds.). (2003).
SVG 1.1 recommendation. The World-Wide Web
Consortium (W3C). Retrieved October 4, 2005
from http://www.w3.org/TR/SVG
Fischer, G. (2001). User modeling in humancomputer interaction. User Modeling and UserAdapted Interaction, 11(1-2), 65-86.
Fowler, M. (2003). Patterns of enterprise application architecture. Boston: Addison Wesley
Press.
French, T., Polovina, S., Vile, A., & Park, J.
(2003). Shared meanings requirements elicitation
(SMRE): Towards intelligent, semiotic, evolving architectures for stakeholder e-mediated

225

Communication + Dynamic Interface = Better User Experience

communication. Using Conceptual Structures:

Contributions to ICCS 2003, 57-68.
Liu, K., Clarke, R., Anderson, P., Stamper, R., &
Abou-Zeid, E. (2002). Organizational semiotics:
Evolving a science of information systems. Proceedings of IFIP WG8.1 Working Conference.
MacDonald, M. (2004). Microsoft .NET distributed applications: Integrating XML Web services
and .NET remoting. Redmond, WA: Microsoft
Press.
Meyer, B. (1988). Object-oriented software
construction. Upper Saddle River, NJ: PrenticeHall.
Microsoft Corporation. (1998). Microsoft component services: A technical overview. Retrieved
August 29, 2004, from http://www.microsoft.
com/com/wpaper/compsvcs.asp
Polovina, S., & Strang, D. (2004). Facilitating
useful object-orientation for visual business organizations through semiotics. In K. Liu (Ed.),
Virtual, distributed and flexible organisations:
Studies in organisational semiotics (pp. 183-184).
Dordrecht, The Netherlands: Kluwer Academic
Publishers.
Quin, L. (2004a). Extensible markup language
(XML). Retrieved March 26, 2004, from http://
www.w3.org/XML/
Quin, L. (2004b). The extensible stylesheet language family (XSL). Retrieved April 29, 2004,
from http://www.w3.org/Style/XSL/
Shannon, C. (1948). A mathematical theory of
communication. The Bell System Technical Journal, 27, 379-423, 623-656.
Vile, A., & Polovina, S. (2000). Making accessible
Websites with semiotics. Retrieved August 15,
2004, from http://www.polovina.me.uk/publications/Semiotics4AWDshortpaper.PDF

226

Key Terms
Accessibility: The measure of whether a
person can perform an interaction, access information, or do anything else. It does not measure
how well he or she can do it, though.
Content: The information, such as thoughts,
ideas, and so forth, that someone wishes to communicate. Examples of content could be the ideas
and concepts conveyed through this article, the
fact that you must stop when a traffic light is
red, and so on. Importantly, content is what is
to be communicated but not how it is to be communicated.
Encoding: Encoding is the process by which
the content and meaning that is to be communicated is transformed into a physical form suitable
for communication. It involves transforming
thoughts and ideas into words, images, actions,
and so forth, and then further transforming the
words or images into their physical form.
Object Orientation: A view of the world
based on the notion that it is made up of objects
classified by a hierarchical superclass-subclass
structure under the most generic superclass (or
root) known as an object. For example, a car is
a (subclass of) vehicle, a vehicle is a moving
object, and a moving object is an object. Hence,
a car is an object as the relationship is transitive
and, accordingly, a subclass must at least have the
attributes and functionality of its superclass(es).
Thus, if we provide a generic user-presentation
object with a standard interface, then any of its
subclasses will conform to that standard interface.
This enables the plug and play of any desired
subclass according to the users encoding and
decoding needs.
Physical Form: The actual physical means
by which thoughts, meaning, concepts, and so
forth are conveyed. This, therefore, can take the

Communication + Dynamic Interface = Better User Experience

form of any physical format, such as the writing

or displaying of words, the drawing or displaying
of images, spoken utterances or other forms of
sounds, the carrying out of actions (e.g., bodily
gestures), and so forth.

a standard interface between the tiers. This enables the easy swapping in and out of presentation
components, thus enabling information to be
encoded into the most appropriate physical form
for a given user at any given time.

Software Architecture: Rather like the

architecture of a building, software architecture
describes the principled, structural design of computer software. Contemporary software architectures are multitier (or n-tier) in nature. Essentially,
these stem from a two-tier architecture in which
user-presentation components are separated from
the information-content components, hence the
two overall tiers. Communication occurs through

Usability: A measure of how well someone

can use something. Usability, in comparison to
accessibility, looks at factors such as ease of
use, efficiency, effectiveness, and accuracy. It
concentrates on factors of an interaction other
than whether someone can perform something,
access information, and so forth, which are all
handled by accessibility.

This work was previously published in the Encyclopedia of Human Computer Interaction, edited by C. Ghaoui, pp. 85-91,
copyright 2006 by Information Science Reference, formerly known as Idea Group Reference (an imprint of IGI Global).

227

228

Chapter 1.18

Pervasive Computing:
What Is It Anyway?

Emerson Loureiro
Federal University of Campina Grande, Brazil
Glauber Ferreira
Federal University of Campina Grande, Brazil
Hyggo Almeida
Federal University of Campina Grande, Brazil
Angelo Perkusich
Federal University of Campina Grande, Brazil

Abstract

Inside Chapter

In this chapter, we introduce the key ideas related to the paradigm of pervasive computing.
We discuss its concepts, challenges, and current
solutions by dividing it into four research areas.
Such division is how we were able to understand
what really is involved in pervasive computing
at different levels. Our intent is to provide readers with introductory theoretical support in the
selected research areas to aid them in their studies
of pervasive computing. Within this context, we
hope the chapter can be helpful for researchers of
pervasive computing, mainly for the beginners,
and for students and professors in their academic
activities.

The recent advances in hardware and wireless

technologies have leveraged the creation of the
first experimental pervasive computing scenarios.
Due to the belief that these scenarios will be an
integral part of future living, research in this field
is increasing at a fast pace. Therefore, theoretical
and mainly practical studies are of great use as a
way of supporting this belief.
Performing such studies, however, implies
identifying the intricacies behind pervasive
computing. Although its concept is quite simple,
understanding these intricacies is a task which
scatters across different research fields. Computer

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Pervasive Computing

networks, distributed and cognitive systems,

software engineering, and user interface design
are some of these fields.
Therefore, in this chapter our main objective
is to identify and discuss, at an introductory level,
some of these intricacies. More specifically, we
define four major research areas in pervasive
computing, namely pervasive networking, context
awareness, pervasive systems development, and
pervasive computing middleware. Based on this
view, we then take the reader on a journey through
the universe of pervasive computing, discussing
concepts, challenges, and current solutions.

Introduction
Today, computing is facing a significant revolution. There is a clear migration from the traditional desktop-based computing to the ubiquitous

era, where computing will be spread all around

us and seamlessly integrated into our lives. It
is this new stage of computing that researchers
have named pervasive computing. We can say
that it is the accomplishment of the so-called
concept of calm technology (Weiser & Brown,
1995), or as Weiser (1993) has said, it envisions
computation primarily in the background where
it may not even be noticed (p. 1). Not surprisingly, these ideas require us to view computers
in a totally different way, not only as something
we log onto, work on, and log out of when we are
finished (Saha & Mukherjee, 2003). Instead, we
should see a computer as a portal to a repository
of computational resources, making use of them
to work on the background and fulfill tasks according to our needs and preferences.
Pervasive computing, also known as ubiquitous
computing (Weiser, 1991), has been recognized
as the third wave in computer science, following

Figure 1. Sales of mainframes, personal computers, and ubiquitous computing devices

229

Pervasive Computing

the mainframe and the personal computer ages.

Therefore, even if not fully conceived, pervasive
computing will be the prevailing paradigm of the
21st century. Observing the graph shown in Figure
11, one can see the sales associated with ubiquitous computing devices follow a fast exponential
growth. As more and more facilities, or services,
will be available for users of such devices, this
growth, even that in a lower rate, will be expected
to continue. After all, it is not for nothing that
academy and mainly industry are so confident
on the pervasive computing paradigm.
Getting a ride on this new trend, the purpose
of this chapter is to conduct the reader behind the
scenes of pervasive computing, introducing the
main concepts and challenges involved in it. The
structure of the chapter is illustrated in Figure 2.
We start with a first glance at pervasive computing by describing a sample scenario in order to
provide the reader with the general concepts. It

Figure 2. Overview of the chapter

230

is presented as an overview of the technological

advances that have leveraged the development
of pervasive systems, as well as the challenges
imposed by pervasive computing scenarios. In the
Pervasive Networking section, we present two key
concepts for pervasive environments, mobility and
host discovery. The notion of context and its importance to pervasive computing will be outlined in
the Context in Pervasive Computing section. Next,
we present some methods that have been used for
developing pervasive systems. More specifically,
some techniques that application developers need
in order to deal with the inherent characteristics of
software for pervasive computing are discussed.
Based on this discussion, we then outline in the
Middleware for Pervasive Computing section
the main features that should be presented by a
pervasive computing middleware and how they
can aid the development of pervasive applications. Additionally, some pervasive computing

Pervasive Computing

middleware solutions are presented. We conclude

the chapter by summarizing the actual state of
pervasive computing research, and also discuss
possible future directions.

A First Glance at Pervasive

Computing
Imagine yourself porting a mobile device, like
a handheld, while walking through a shopping
mall. Now imagine you are very interested in
having a cappuccino. You think it is a pity there
is no place in this mall offering cappuccinos.
Fortunately, your handheld knows that you
like cappuccino, and it becomes aware that the
shopping mall has just opened a coffee shop. And
guess what? Cappuccinos are sold there. Based
on this information, your mobile device notifies
you about this news, and now you can have your
desired cappuccino. While you are savoring it,
you are reminded of that book you are interested
in. Without hesitation, you take your device out
and check to see if any store in the mall has such
a book to offer. When the search is finished, you
find out that two bookstores are selling the book
you want. The search returns all the information
you need, such as the price of the book in both
stores, discounts, and payment options. With such
information at hand, you select the offer best suited
for you and request the book. From your mobile
device, you provide all the information to complete
the purchase. Now, all you need to do is go to the
bookstore and get your brand new book.
Wonderful, is it not? Just by porting a mobile
device you were able to savor a cappuccino and
buy the book you wanted. And both tasks were
completed in a very natural way, as if computing
had been fully woven into our lives. This is a typical example of a pervasive computing scenario.
First introduced by Mark Weiser in his seminal
paper (Weiser, 1991), pervasive computing is part
of an evolution chain consisting of distributed
systems and mobile computing (Satyanarayanan,

2001). It envisions a world where computing and

applications are embedded in everyday objects.
Clothes, televisions, air conditionings, and cars
are examples of such objects. They will be capable of seamlessly interacting with each other
in order to perform tasks on behalf of the users
by taking intelligent actions or making available
relevant information at the right place and at the
right time.
Weiser affirmed that pervasive computing can
be achieved through three major technologies:
cheap and low-power devices, a network infrastructure for communicating these devices, and
pervasive applications. At the time this was said,
hardware technology was not fully available to
support pervasive computing. Wireless networking, as we have today, was neither available nor
deployed in mobile devices. Consequently, pervasive applications could not be developed.
This started to change with the introduction of
more powerful mobile devices, such as the current
smart cellular phones and handhelds, that allowed
for the development of more complex applications
for such devices. Also, the embedding of wireless
networking technologies, like Bluetooth (Bray &
Sturman, 2000) and Wi-Fi (Reid & Seide, 2002),
on mobile devices has promoted the availability
of mobile applications. These technologies have
permitted us to give the first steps toward the vision
of pervasive computing. This has caused a rush
for the first solutions in the field, and many works
with this purpose have been developed. Oxygen
(http://www.oxygen.lcs.mit.edu), Aura (Garlan,
Siewiorek, Smailagic, & Steenkiste, 2002), Smart
Space (Stanford, Garofolo, Galibert, Michel, &
Laprun, 2003), Portolano (Esler, Hightower, Anderson, & Borriello, 1999), and Wings (Loureiro,
Bublitz, Barbosa, Perkusich, Almeida, & Ferreira,
2006) are some examples of works related to the
branch of pervasive computing.
However, a long road is still ahead. Despite the
hardware advances in the last years, there are still
a new set of problems associated with software
systems for pervasive environments. For pervasive

231

Pervasive Computing

computing to become a true reality, applications

need to have full access to the information about
the users and the environments in which they are
situated. This is in a broad sense what has been
named context, although many variations for the
concept exist nowadays. The current lighting
condition, temperature level, and the number of
users around a mobile device are some examples
of the information associated with the word context. The great challenge that remains within this
scope is how to model context information, and
mainly how to effectively exploit it. The effective
use of context information is one of the key issues
to achieve Weisers vision of invisible computing
(Satyanarayanan, 2001). Still, acting on behalf of
users requires pervasive systems to be ready for
changes in their interests. Changes in the local
and remote resources available should also be
considered, as they are important for achieving
such pervasiveness.

Going down to the networking level, we find

that mobility and host discovery are two important
features for pervasive environments. Whereas the
former allows embedded applications to perform
their tasks uninterruptedly; that is, even when the
user is moving through different networks, host
discovery permits a device to discover network
hosts, and also to be discovered by them. Due
to the discovery of such hosts, a device is then
able to query for the information and resources
they share, informing the user about the most
relevant ones.
It is clear the preceding challenges need to be
first well understood and solved. Only then can
dependable pervasive systems emerge. Therefore,
from this point on, we start delineating such
challenges, as well as mentioning some of the
current solutions for them. However, if any of the
introductory ideas are not clear enough, the reader
can refer to the literature presented in Table 1,

Table 1. The introductory literature that has been used and their main contributions

(Weiser, 1991)

Reference

Main Contribution

(Weiser, 1993)

Overview of pervasive computing

(Saha & Mukherjee, 2003)

(Satyanarayanan, 2001)
(Weiser & Brown, 1995)
(Garlan, Siewiorek, Smailagic, & Steenkiste, 2002)

Challenges brought on by pervasive computing

The concept of calm technology

(Stanford, Garofolo, Galibert, Michel, & Laprun, 2003)

(Esler, Hightower, Anderson, & Borriello, 1999)
(Loureiro, Bublitz, Barbosa, Perkusich, Almeida, &
Ferreira, 2006)

232

Some current solutions for pervasive computing

Pervasive Computing

where we indicate the contribution of references

used throughout the first two sections.

Pervasive Networking
Pervasive networking is about the plumbing involved in the communication of devices in pervasive computing environments. Therefore, studies
within this area range from the design and energy
consumption techniques of wireless interfaces to
the development of high level protocols, such as
routing and transport ones. At this high level, mobility and host discovery play fundamental roles
as enablers of pervasive environments. Research
in these areas has considerably advanced, and as a
result, some practical solutions are already available today. Therefore, in this section we present a
review of the concepts associated with mobility
and host discovery.

Mobility is certainly a fundamental element

for pervasive environments. The possibility for
providing users with on the move networking
enables applications to work in the background
by invisibly searching for some relevant content.
However, the use of mobility in computing systems
inherently leads them to face a set of new and
challenging problems, which can be grouped in
the following way (Satyanarayanan, 1996):

Mobility
You probably receive your mail at your residence,
right? Now, consider that you are moving to a
new house. Among other concerns, you would
probably want to change the mailing address associated with correspondences like your credit
card bill. In this case, you must notify your credit
card company that you have just moved, and that
consequently your mailing address has changed.
Either you do this or your credit card bill will
be delivered to the old address, which is not a
desirable situation.
A scenario similar to the above one is basically what happens in computing environments
enhanced with mobility. In other words, mobility
must allow a device to change its physical location
and still be capable of receiving network packages from the other hosts. Note that, by physical
location, we are referring to the network a device
is connected to. Therefore, moving through different networks is what requires a node to have
its address changed.

Resource poverty of mobile computing

devices: It is a fact that mobile devices are
resource-poor when compared to personal
computers. Processor speed and memory/
disk capacities are considerably higher
in static computers than in mobile ones.
Therefore, software for mobile computing
need to be well designed in order to save
processor usage and storage space.
Energy restrictions: Static computers are
plugged to some energy network, which is
theoretically an unlimited source of energy.
Mobile devices, on the other hand, depend
on limited capacity batteries. Therefore,
techniques for saving energy should be applied in mobile applications.
Variability of wireless links: Wireless
connectivity is still highly variable in terms
of performance and reliability. Whereas
some buildings provide high-bandwidth
and reliable wireless connections, others
may provide considerably less bandwidth
and reliability. This can be even worse in an
open environment, where connection may be
shared by lots of users. Undoubtedly, these
changes in wireless connectivity need to be
addressed in pervasive computing systems
by, for example, implementing some network
congestion control algorithm.
Security of wireless connections: Due to
the broadcast nature of wireless links, they
are easier to eavesdrop with than wired ones.
Therefore, if security is already an important

233

Pervasive Computing

feature of fixed networks, for wireless ones

it is an even more important feature.

The Degrees of Mobility

The different degrees of mobility have a direct
impact over the topology of a network. The
more mobile are the network nodes, the more
flexible the network needs to be. In the case of
Ethernet networks, for example, nodes are too
static. Therefore, only in sporadic situations is
it necessary to change the network address of a
node. Consequently, the network topology does
not necessarily need to be flexible. In this case,
protocols like DHCP (dynamic host configuration protocol) seamlessly solve the problem of
delivering new network addresses to nodes. At
the other extreme, a network may be populated
by highly mobile nodes. Such a level of mobility allows users to move around areas that, for
various reasons, have no fixed network coverage.
In these situations infrastructureless networks
are more appropriate. That is, nodes should be
capable of establishing connections with each
other whenever needed. In this case, the network
would be formed opportunistically as more and
more mobile devices get together.
Within this context, as Sun and Savoula (2002)
have already pointed out, three modes of communication can be distinguished when it comes
to the degree of mobility: nomadic, cellular, and
pervasive communication. In the first case, no
connection is necessary when the device is migrating from one network to another. A typical
example of the nomadic communication is a user
who uses a notebook for connecting to a network
both at work and at home. Note that there is no
need to keep network connections while the users are moving from work to their house. Only
when getting home should the notebook receive
a new address for accessing the network. In the
cellular communication mode, the network is
organized in cells, where each cell is located

234

adjacent to a set of others. All cells have a central element, which provides connectivity for the
nodes within them. Therefore, a mobile device
can move through different cells and maintain a
connection with their central element, becoming
thus accessible even when moving. Current mobile
telephony networks are an example of this kind
of communication, where the base stations act
as the central elements. Finally, pervasive communication can be mainly characterized by the
lack of a fixed network infrastructure different
from the two previous ones. Therefore, nodes
should establish connections directly with each
other whenever they come close enough. These
features are what characterize the so-called ad
hoc networks (Chlamtac, Conti, & Liu, 2003), and
will be of great importance in the deployment of
pervasive computing environments.
Among the current solutions for mobility, we
could cite Mobile IP (Perkins, 1997), GPRS (General Packet Radio System), and Bluetooth. Basically, Mobile IP and GPRS are mobility solutions
respectively for IP and mobile telephony networks.
Bluetooth, on the other hand, is a standard for
short-range and low-cost wireless communication
in an ad hoc way. Further description concerning
these technologies can be found on the Web sites
listed in the Useful URLs section.

Host Discovery
Putting it simply, host discovery is about finding other hosts in the network, and also being
found by them. This apparently simple concept
is of great importance for pervasive computing
environments, and can be found in technologies
such as Bluetooth and UPnP (Universal Plug and
Play). As an example of its usage, consider the
acquisition of context information in decentralized
environments, such as the available services. By
using host discovery, a device can, for example,
find the available hosts in the environment and
query them for the services they provide. This

Pervasive Computing

Figure 3. Host discovery approaches

Figure 3(a). Notification-based host discovery.

is the service discovery mechanism used by the

Bluetooth technology.
Host discovery can be performed either by
using a notification-based approach or a querybased one. In the former, a host is discovered
when it notifies itself to the others. This requires
the host to send its advertisement through the
network, so that the others can be aware of it.
Such advertisements contain information such as
the address, name, and description of the hosts.
The advertising task can be executed a singe time
(e.g., when the host joins the network) or periodically (e.g., each second). The notification-based
host discovery is illustrated in Figure 3(a). On the
other hand, the query-based approach, illustrated
in Figure 3(b), is based on sending discovery
messages and waiting for their responses, which
contain information about the discovered hosts.
Therefore, by retrieving the information contained
in these responses, a host is able to contact the
discovered hosts.

Figure 3(b). Query-based host discovery.

Context in Pervasive
Computing
A fundamental functionality of pervasive computing applications is to present users with relevant
information or services at the right place and in
the right time, in a seamless way. Such information can be, for instance, a landmark for tourists
to visit based on their preferences. In this process, two key inputs are involved: the needs and
interests of the user and the information available
both in the environment and in their devices. The
former allows the applications to define what sort
of information would be relevant to the user. The
latter is the source from where such information
will be retrieved. Let us get back to our first example, the one presented in the First Glance on
Pervasive Computing section. In that case, your
desire for a cappuccino and the book you wanted
to buy were your needs and interests. Whereas the
former could be acquired, for instance, by keep-

235

Pervasive Computing

ing historic information of your preferences, the

information about the book has been explicitly
provided by you. Based on both information, the
application running in the handheld was able to
determine what would be considered relevant
information for you. By gathering information
from the environment, the application decided
that the opening of a new coffee shop and the
bookstores in the surrounding area were relevant
enough information to present you.

A Definition of Context
The preceding discussion should provide at least
a first impression of what context really means. In
pervasive computing literature, context has been
defined in a number of ways. Some researchers
have defined context by categorizing the different
information associated with it. Gwizdka (2000),
for example, identifies two types of context:
internal and external. Internal context provides
information about the state of the users, such as
their current emotional state. External context,
on the other hand, describes the environment on
which a user is immersed, for example, informing about the current noise or temperature level.
In the work of Petrelli, Not, Strapparava, Stock,
and Zancanaro (2000), two types of context are
identified: material and social. Material context
is associated with location (e.g., at home), devices
(e.g., a handheld, a cellular phone) or the available infrastructure (e.g., available networks).
Social context, on the other hand, encapsulates
the information about the current social state of
the user, for example, in a meeting or a movie
theater. Another work, by Schilit and Theimer
(1994), defines three categories for grouping
context information: computing context, user
context, and physical context. A refinement of
these categories is presented by Chen and Kotz
(2000), through the addition of a fourth category,
time context. The information associated with
each category is presented as follows.

236

Computing context: Network bandwidth,

the cost involved in communication, and
available resources, such as printers, displays, and workstations.
User context: People in the vicinity and the
location, profile, and current social situation
of the user.
Physical context: Lighting and noise levels,
current temperature, and traffic conditions.
Time context: Time of the day, day of the
week, month and season of the year.

Note that the above ideas do not really define

what is context, but instead try to give it a meaning by enumerating the sort of information that
could be related. The problem in defining context in this way is that it may be hard to affirm
whether some information can be considered
context information or not. Additionally, a more
general definition of context would certainly enable a better understanding of its role in pervasive
computing. Therefore, for the purposes of this
chapter, we consider a context as defined by Dey
(2001, p. 45).
Context is any information that can be used to
characterize the situation of an entity. An entity
is a person, place, or object that is considered
relevant to the interaction between a user and an
application, including the user and applications
themselves.

Context Aware Applications

Considering the current context for determining
the actions that can be taken is very natural for
us. We commonly use information such as the
place we are at and the people around us to guide
our actions. When we are in a movie theater, for
example, we know how bothersome it is to speak
loudly, and so most of us generally do not do so.
In this scope, applications that make use of this
kind of information are called context aware ap-

Pervasive Computing

plications. Such applications, however, require

functionalities for acquiring and interpreting the
context information, and to choose and execute an
action based on it. More precisely, three elements
are involved throughout this process: context
acquisition, context representation, and context
reasoning.
Context acquisition concerns the way context
information is obtained, namely sensed, derived,
or explicitly provided (Mostfaoui, Rocha, &
Brzillon, 2004). Sensed context information is
gathered from physical sensors, such as lighting
or temperature ones. Derived context information
is computed on demand, like the time of the day
and the number of people around. Finally, when
context information is explicitly provided, the
user is responsible for providing it. An example
of this acquisition can be viewed on applications
which provide a form for users to fill in with their
preferences (i.e., favorite kinds of books, movies,
food, and entertainment).
Once acquired, context information needs to
be made available to the interested applications.
This implies that it must be represented in an
agreed format, so that the interested applications
can understand the information they received
from the providers. As already pointed out by Held,
Buchholz, and Schill (2002), the representation of
the context should be structured, interchangeable,
composable/decomposable, uniform, extensible,
and standardized. Structuring is important for
enabling applications to filter pieces of information from the context. Interchangeability is related
to the possibility of applications to exchange
context information with each other. Therefore,
in order to provide this characteristic, a context
representation must be serializable. Composition and decomposition enables to compose and
decompose context information from different
sources. This allows transferring only sub-parts
of the information, for example, when it has been
updated, to avoid sending the whole context representation. Uniformity claims that different kinds
of context information (e.g., users profile, device

profiles, resource profiles) should be represented

in a similar manner in order to ease interpretation by the applications which use them. As the
number of terms and variables of a context is difficult to predict, even in quite restricted domains,
extensibility is also a fundamental characteristic
for context information representation. Finally, as
devices and applications can come from different vendors, context information must be based
on standards. This would certainly improve the
exchanging of context information among pervasive computing applications. Current solutions for
pervasive computing represent context in different
ways; for example, using key-value pairs, XML
documents (Boyera & Lewis, 2005; Ryan, 1999),
object-oriented models (Henricksen, Indulska,
& Rakotonirainy, 2002), and ontology-based
models (Chen, Finin, & Joshi, 2003; Henricksen,
Livingstone, & Indulska, 2004; Masuoka, Labrou,
Parsie, & Sirin, 2003).
Considering that the context information is
represented in a way that applications understand,
it is possible to make use of this information and
perform context reasoning. Basically, context
reasoning is the use of contextual information
for guiding the actions an application will take.
As Satyanarayanan (2001) has already pointed
out, the effective use of context information is
a fundamental element in pervasive computing,
as a means for achieving the invisibility feature
envisioned by Weiser. The context reasoning
mechanism of a pervasive computing system
can be as simple as if-then-else statements, or
as complex as rule-based (Nishigaki, Yasumoto,
Shibata, Ito, & Higashino, 2005) and case-based
methods (Ma, Kim, Ma, Tang, & Zhou, 2005).
An important characteristic of context reasoning systems is the ability to deal with uncertain
context information. As the information acquired
by sensors (i.e., sensed context information) is
prone to errors, applications should consider the
quality of the acquired context information when
performing their reasoning tasks. To this end,
different approaches have been proposed using,

237

Pervasive Computing

for example, Bayesian networks (Gu, Pung, &

Zhang, 2004a) and Fuzzy logic (Ranganathan,
Muhtadi, & Campbell, 2004).

Development of Pervasive
Computing Systems
Based on our discussion until now, it is easy to
realize that the intrinsic features of pervasive
computing have an impact on the way software
is designed and developed. For example, adaptability, customization, and context sensitivity are
some of the characteristics that are constantly
associated with pervasive computing systems
(Raatikainen, Christensen, & Nakajima, 2002).
Different software engineering techniques have
been used when dealing with them. In this way, we
will now review some of these techniques, as well
as how they can be applied in pervasive computing
systems. More precisely, we will discuss how the
component and plugin-based approaches can be
used to provide such systems with adaptability
and customization. In addition, we show how they
can be aware of changes in the context, through
the generation and notification of events.

Figure 4. Component-based architecture

238

The Component-Based Approach

Component-based software engineering addresses the development of systems as an assembly of
components. More precisely, its focus is on the
development of components as reusable entities,
as well as on the maintenance and upgrade of systems through the customization and replacement
of such components. The main advantages of this
reuse and assembly-based paradigm is a more effective management of complexity, reduced time
to market, increased productivity, and improved
quality (Crnkovic, 2001).
In a general way, a component is a software
implementation which can be executed in a logic
or physical device and can be reused in several
applications of the same domain (Bachman, Bass,
Buhman, Dorda, Long, Robert, Seacord, &
Wallnau, 2000). The well-defined interface of
the component describes the services or events
that implement its functionalities. Such interface enables encapsulation of the components
functionalities, reducing the coupling among
them, and also improving the flexibility of the
software design.

Pervasive Computing

Because the components must be connected to

assemble the application, it is necessary software
to ensure the interaction among components,
managing their service and event dependencies.
Generally, such entity is implemented as a software framework. To guarantee the components
will behave as expected by the framework, some
interfaces, called contracts, are defined (see Figure
4). These contracts, which components are forced
to implement, guarantee that the development
of independent components satisfies certain
standards, allowing the framework to use such
components without being aware of their internal
implementation details (Bachman et al., 2000).

The Component-Based Approach in

Pervasive Computing
Within the scope of pervasive computing, the
application of the component-based approach is
straightforward in the development and maintenance of software. Due to the dynamics and
heterogeneity they present (e.g., different services
available, several kinds of protocols, devices with
different processing power, and storage capabilities), the reuse and flexibility characteristics of
components are mandatory in pervasive computing software design. The combination of these
features provides an efficient way for enabling
an application to seamlessly adapt. This can be
performed either by dynamically changing a
component by an equivalent, or by assembling a
new functionality to the application.
In this context, reuse is important, for example,
due to the increasing number of technologies related to pervasive networking, such as Bluetooth,
UPnP, Zeroconf (Guttman, 2001), and JXTA
(Gong, 2001). Since such technologies are based
on standard protocols, they can be implemented
as software components in order to be reused in
several applications. Reuse is not only interesting
for communication technologies. Components can
also be implemented for many other purposes,
such as audio and video streaming and context

information retrieval, and yet provide the same

reuse feature.
Examples of pervasive computing systems
which make use of the component approach include Aura (Garlan, et al., 2002), Runes (Costa,
Coulson, Mascolo, Picco, & Zachariadis, 2005),
and PCom (Becker, Handte, Schiele, & Rothermel, 2004).

Plug-in-Based Architectures
Applications based on the plug-in approach are
characterized by having a functional core with
well-defined hooks where extensions (i.e., plugins) can be dynamically plugged (see Figure
5) (Mayer, Melzer, & Schweiggert, 2002). The
functional core contains only the minimum set of
functionalities the application needs to run. Plugins, on the other hand, are intended to enhance
the application by adding features to it. Therefore,
plug-in-based applications can be executed even
when no extensions have been installed. Besides,
features that are not in use can be safely removed,
by plugging out the associated plug-in.
A more revolutionary view of plug-in-based
architectures is to consider everything as a plugin. In this new form of plug-in architectures, the
application becomes a runtime engine for managing each plug-ins life cycle. As a consequence,
end user functionalities are entirely provided by
means of plug-ins. For such kinds of application,
the extension of plug-ins through other plug-ins is
thus a fundamental feature (Birsan, 2005).

Plug-in-Based Architectures in
Pervasive Computing
The application of the plug-in approach in pervasive computing systems provides them with
the needed characteristic of customization. From
minimum, but functional software, users can
gradually download specific plug-ins to their
daily activities, choosing the ones which best
supply their needs. Take as an example an envi-

239

Pervasive Computing

Figure 5. General view of a plug-in-based architecture

ronment filled with services of video streaming,

delivered in avi format. Consider now a mobile
device, with a video player installed, capable of
receiving video streaming through the network
and displaying it to the user. Such a player,
however, can only decode mpeg formats, and
the environment delivers video in avi. Once the
user wants to play some movie available in the
environment, the player has to transparently find
out a way of playing avi video formats. Therefore,
considering that Internet access is available in the
environment, the video player can download an
avi plug-in from a repository, install it, and play
the required video in a transparent way. Note that,
although the player did not know the user would
ever need to receive a video stream in avi format,
it was prepared for such situation, permitting it
to fulfill the process smoothly.
It is interesting to note how the pure plug-in
approach fits well when applied in pervasive computing. The plug-in runtime environment, obviously equipped with other features, like context
sensitivity, can be mapped to a pervasive computing infrastructure. Applications would then be
viewed as plug-ins, which could be extended by

240

other plug-ins, and so on. Therefore, plug-ins, in

this case, applications, could be installed in the
users device on demand, and be removed from
it when no longer needed.
Within this scope, the fact that plug-ins can be
removed without affecting its host application is
also important for pervasive computing. Mobile
devices are restricted concerning disk and memory
capacity, and thus it would be helpful to remove
non-used plug-ins in order to save some space.
Practical examples concerning the usage of
the plug-in concepts in pervasive computing can
be found in middlewares like Wings (Loureiro,
et al., 2006), BASE (Becker, Schiele, Gubbels,
& Rothermel, 2003), ReMMoC (Coulson, Grace,
Blair, Duce, Cooper, & Sagar, 2005), and PluginORB (dAcierno, Pietro, Coronato, & Gugliara,
2005).

Event-Based Systems
An event-based system is the one in which the
communication among some of its components is
performed by generating and receiving events. In
this process, initially a component fires an event,

Pervasive Computing

and after that, such an event will be delivered to

all the components interested in it. In an eventbased system, a component can assume the role
of producer, consumer, or both. The producer is
responsible for generating and firing events. The
consumer, on the other hand, is a component which
registers itself for the occurrence of a particular
event, and is notified when such event occurs.
The process of event notification can be performed in two ways: through the event-based or
the callback cooperation models (Fiege, Mhl, &
Grtner, 2002). In the former, a key element is the
event service. Such an element is responsible for
receiving an event from a producer and forwarding it to consumers. To exemplify this process, we
have presented an example, which is illustrated in
Figure 6. In such a figure, we have six components
(A, B, C, D, E, and F), the event service, and two
kinds of events (X and Z). Some components act
only as producers (components A and C) whereas

others only as consumers (components B, D, and

F). Finally, component E acts as both a producer
and a consumer. The arrows in the figure indicate
the flow of event announcements and notifications
within the system. In the callback model, on the
other hand, the consumer subscribes directly to
the provider of the event. Therefore, the provider
must keep track of the consumers for each event
it provides. When some component is interested
in various events, it must thus subscribe to each
component providing the event. The so-called
Observer design pattern (Gamma, Helm, Johnson, & Vlissides, 1995) is an abstraction of the
callback event cooperation model.

The Event-Based Approach in

Pervasive Computing
In the context of pervasive computing, the eventoriented approach can be very helpful in the no-

Figure 6. Event delivery through the event service

241

Pervasive Computing

tification of changes in the context. An example

would be a user, located in a shopping mall, who
is interested in book discounts. Applications
dispersed in the environment could frequently
deliver events associated with the discounts
available in the mall. Through a handheld, the
users could then register their interests on events
associated, for example, with books discounts. As
soon as one of such events is generated by some
application, a centralized event service would be
able to deliver the event to the users handheld,
which in turn would notify the user. The event
could then provide the user with information
such as the store offering the discounts and the
associated books.
Another use of events in pervasive computing
is for notifying applications about the battery level
of a device. In this scenario, through the callback
cooperation model, an application could specify a
value associated with the remaining battery (e.g.,
5%) and register itself to an event which is fired
every time the battery level reaches such a value.
In this way, an application could be notified when
the device is about to be run out of energy, in order
to save some critical data before the battery be
completely consumed.
The event approach has been successfully used
in the development of pervasive software systems.
Maybe the best example we can cite is the UPnP
specification. Its event mechanism, which uses the
General Event Notification Architecture (GENA)
(Cohen, Aggarwal, & Goland, 2000), permits the
notification of changes in a device, such as the
services that have been inserted or removed from
it. Other examples include the Scooby (Robinson
& Wakeman, 2003) and Solar (Chen & Kotz, 2002)
pervasive computing middlewares.

Middleware for Pervasive

Computing
The task of building pervasive computing applications can be too tedious if performed from

242

scratch. In other words, the developer will need to

deal with low level networking protocols, the way
the context is acquired and monitored, notification of changes in the context, and methods for
enabling flexibility in their applications. This, of
course, deviates the attention of the developers to
tasks that are not the purpose of the application.
Instead, they should only concentrate on the application logic, that is, the tasks the application
must perform. This is where middleware for
pervasive computing comes in.
By providing a high level abstraction for application developers, middlewares can considerably
speed up development time and also decrease
the number of errors in software implementation
(Mascolo, Capra, & Emmerich, 2002). Besides
abstracting low level details, an ideal middleware
should also provide the developer with robustness, efficiency, and security (Aldestein, Gupta,
Richard, & Schwiebert, 2005).
These are, however, characteristics that should
be presented by any kind of middleware, targeted
at pervasive computing or not. From the discussion we have presented until now, it is clear that
pervasive computing middleware should not stop
at this point. Adaptability, for example, is part of
any application in pervasive environments, and
consequently, pervasive computing middlewares
should present it. Take as an example the RPC
(remote procedure call) protocols they use for
invoking remote services available in the environment. Considering the diversity of such protocols
(e.g., XML-RPC, RMI, SOAP), the middleware
could be faced with situations in which it does
not implement the RPC protocol used by a certain
service. Middlewares for pervasive computing
should be capable of transparently overcoming
this problem, by downloading the specific RPC
protocol implementation, installing it, and finally
invoking the service. Moreover, middlewares for
pervasive computing should naturally support the
development of flexible applications on top of it. In
other words, they should provide all the necessary
tools for developers to build inherently adaptable

Pervasive Computing

applications. Based on our previous discussions,

the component and plug-in-based approaches can
be very effective, both for the middleware and the
applications running over it.
Another characteristic is that pervasive computing middlewares need to provide a way for applications to retrieve information from the context,
and also to be notified about changes related to it.
In middlewares for traditional distributed systems
this was not required, as the context on which
they were executed was too static. Therefore, the
sporadic changes in the context could be easily
managed by the middleware, with no need to be
exposed to the applications. As a result of the
dynamics involved in pervasive environments,
middlewares can not efficiently take decisions
on behalf of applications (Mascolo et al., 2002).
It is more reasonable, thus, to expose the context
to applications and let them take their own actions. At this point, context representation and
event notification are very useful for pervasive
computing middlewares.

Pervasive Computing Middleware

Solutions
Current solutions for pervasive computing middlewares have been focused on different aspects, such
as service discovery, service composition, context
sensitivity, and networking heterogeneity. These
solutions range from general purposes middlewares, like Jini (Waldo, 1999) and Aura (Garlan,
et al., 2002), to application domain specific ones,
such as healthcare (Bardram & Christensen, 2001)
and e-learning (Apostolopoulos & Kefala, 2003).
Due to their importance in the development of
pervasive computing applications, in this section
we outline some of these solutions.

Jini
Jini is a middleware focused on service discovery
and advertisement. Jini services are advertised in

service catalogs through a Java interface. When the

service catalog address is known a priori, Jini clients make requests directly to the service catalog
in order to discover services. The service catalog
returns a proxy for each discovered service, which
is used by clients for remotely invoking it. If the
service catalog address is not known, requests
are performed in order to find it. As the service
providerand consequently its servicesmay
not be available in the network, Jini implements
a lease mechanism to help clients avoid finding
unavailable services. Therefore, when providing
a service, its provider receives a lease, which
must be renewed at a specified time interval. If
the lease is not renewed, the service is removed
from the service catalog, and thus can no longer
be found by Jini clients.

OSGi
The open services gateway interface (OSGi) is a
specification supporting the development of Java
service-based applications through the deployment of components known as bundles (Lee,
Nordstedt, & Helal, 2003). A major advantage of
the OSGi specification is that such bundles can
be installed, uninstalled, and updated without the
need to stop and restart the Java applications. In
the scope of pervasive computing, this is a fundamental feature, as it enables pervasive computing systems to adapt themselves in a completely
transparent way to their users.
The main idea behind OSGi is the sharing and
discovery of services. In other words, bundles are
able to advertise and discover services. When
advertising a service, a bundle can define a set
of key-value pairs, representing the services
properties. Such properties can thus be useful for
other bundles, in order to discover the services
they need. These advertisement and discovery
processes are both performed through a registry,
managed by the OSGi implementation. In this
way, it is able to keep track of all services cur-

243

Pervasive Computing

rently advertised and being used, thus enabling

a bundle to be updated even when other bundles
are executing and using its services.

RCSM
Reconfigurable context-sensitive middleware
(RCSM) (Yau, Karim, Yu, Bin, & Gupta, 2002)
addresses context-awareness issues through an
object-based framework. The context-independent
information of an application is implemented in
programming languages such as C++ and Java.
The context-sensitive information is implemented
as an interface, using the context-aware interface
definition language (CA-IDL). This interface has
a mapping of what actions should be executed
according to each activated context. In this way,
the application logic is isolated from the context
specification.

SOCAM
Service-oriented context-aware middleware
(SOCAM) (Gu, Pung, & Zhang, 2004b) supports
the development of context-aware services where
ontologies are used for representing context
information. There are two types of ontologies:
high-level ontologies describe generic concepts
which are domain-independent, such as person,
activity, location, and device, and domain-specific
ontologies define concepts which concern specific
domains, such as vehicle and home domains.

Aura
The main focus of Aura is to minimize the distraction of users by providing an environment in
which adaptation is guided by the users context
and needs. This project, developed at the Carnegie Mellon University, has been applied in the
implementation of various applications, such as
a wireless bandwidth advisor, a WaveLan-based
people locator, an application which captures the

244

users intent in order to provide task mobility

among different environments. All these applications use components provided specifically by the
Aura project, as well as other components, such
as the service registry functionalities provided by
Jini and Coda (Satyanarayanan, 2002).

Current and Future Trends

in Pervasive Computing
Research
Undoubtedly, research in the pervasive computing
field has considerably advanced. At the networking level, for example, wireless communication
is already possible through technologies like
Bluetooth, Wi-Fi, and Zigbee. Wi-Fi, although
providing a data rate of up to 54 Mbps, has high
energy consumption. Bluetooth and Zigbee consume considerably less energy than Wi-Fi, but
provide less transmission rates, respectively of
1 Mbps and 250 Kbps at most. The tradeoff of
these characteristics must then be analyzed when
deploying a pervasive computing environment.
Despite these advances, enhancements must still
be achieved in this area. Among them, we could
point out the power consumption of wireless
interfaces. Are current solutions well suited for
todays appliances? Can power consumption be
improved? Or better, can the relation between
power consumption and transfer rate of a wireless
interface be improved? Considering this variability of power consumption and transfer rate in
wireless interfaces, what would be an ideal configuration for the network and the appliances (i.e.,
the set of wireless interfaces in the environment
and the devices) to minimize power consumption
and maximize the transfer of data?
Mobility and ad hoc networking protocols
have also been developed, like Mobile IP, GPRS,
UPnP, and Zeroconf. The mix of wireless interfaces and these protocols have already leveraged
the deployment of some experimental pervasive

Pervasive Computing

environments. One point to be considered in

this scope is the degree of pervasiveness of such
solutions. In other words, how seamless are they
concerning configuration, initialization, and finalization? Do we need to start and stop them all the
time? One could still think about the performance
of these solutions in current mobile devices. Do
they consume too much energy? If yes, how can
they be improved?
When considering the context-awareness, undoubtedly some efforts have been made in specific
application domains. As examples, we could cite
tourist information, healthcare, sports, learning,
multimedia, and intelligent houses. However,
context is far from being used as it is intended.
For example, some of the current solutions focus
on discovering needed services, but do not well
address the assembly of new functionalities based
on the requirements of the users and the resources
available. In this case, what sort of methods should
be applied in order to improve this assembly? Is the
context information represented in a reasonable
way, so that it simplifies such assembly? Work in
this area should involve more precise representation of the context, including the resources and the
interests of the user, as well as more intelligent
methods for better determining what is relevant
to the users, and what is not. Determining what is
relevant to the user also requires the application
to be capable of capturing the users preferences,
needs, and different states (i.e., social, emotional,
physical). User preferences are mostly acquired
by providing forms which users fill in. However,
how boring can it be for a user to update such
forms? Considering that this is a boring task, are
they willing to pay this price for some degree of
pervasiveness? Concerning the needs of the user,
they are commonly obtained explicitly; that is, the
users provide to the system what functionality they
are interested in. Can this acquisition be enhanced?
Is it possible to draw a profile of the user from its
reoccurring needs in order to enable pro-activity?
What about the information concerning the states

of the user? Is it reasonable to require the users

to explicitly set them in their personal devices?
Can this information be acquired in a different
way, through sensors placed in the users body
for example?
Another point to be analyzed concerns the social
impact pervasive computing may cause. Consider,
for example, the way people will react to sensors
that detect their presence and suggest actions for
them. Will they be pleased for these suggestions or
angry because there is always some boring gadget
telling them what to do? One should consider also
whether people will accept having sensors in their
bodies for obtaining information like their health
and emotional state. Will you? What about privacy
and security? Many people are still mistrustful of
checking their banking accounts through the Internet, what would they say about sharing personal
information in an open environment inhabited by all
kinds of people? Certainly, all these social factors
are as important as technological ones, and thus,
must be deeply investigated.

Pervasive Computing for

Knowledge and Learning
Management
As we have seen throughout this chapter, the
evolution that has taken place in hardware and
networking technologies, along with the dissemination of mobile devices like cellular phones, make
pervasive computing the prevailing paradigm for
the next computing systems. Within the scope
of this trend, one of the aspects that pervasive
computing can greatly contribute is knowledge
management, mainly when we consider the increased demand for learning independently of time
and place (Lytras, Pouloudi, & Poulymenakou,
2002). The use of information technology (IT) in
the context of knowledge management is already
commonplace. As already stated by Marwick
(2001), many different IT solutions can be used

245

Pervasive Computing

for this purpose, with the goal of enabling knowledge to be acquired, created, and disseminated
by transforming it from tacit to explicit forms
and vice-versa. One of the great advantages of
IT, in this case, is the possibility to transpose
the barriers of time and space. Based on this, it
is not hard to realize how pervasive computing
can aid in further transposing such barriers. With
pervasive computing technology, members of
an organization are able to establish meetings,
anytime and anywhere, with the goal of sharing their knowledge. In this scenario, video and
audio over the Internet could be used to give the
impression that the members are in a real meeting.
Therefore, people from organizations geographically distributed would not need to travel to meet
each other. Another use of pervasive computing
technology is for the generation of reports with
the intent of disseminating someones knowledge.
With the possibility of having Internet connection
whenever needed, members of an organization
could prepare reports about a specific topic and
then distribute them to the others, no matter where
they are. Within this process, any needed document, either internal or external to the organization, could also be accessed, enabling knowledge
to be acquired, created, and disseminated in a
ubiquitous way.
Due to this possibility of extending the limits
of knowledge acquisition, creation, and dissemination, it is not surprising to see the first solutions
trying to combine pervasive computing and
knowledge management. An example is the work
of Will, Lech, and Klein (2004), which proposes
a tamagotchi-based solution for supporting mobile workers in finding relevant information for
the work they perform. Basically, this solution
works by enabling a mobile device to interact
with information suppliers in a seamless way,
through a continuous and proactive matching
between the information they provide and the one
needed by the mobile workers. Another work in
this category is the p-learning Grid, also known

246

as the pervasive learning Grid (Liao, Yang, &

Hsu, 2005). As its name indicates, the p-Learning Grid is targeted to support mobile learners
in pervasive environments. The overall idea of
this work is to represent learning objects (LOs)
using Grid services (Foster, Kesselman, Nick, &
Tuecke, 2002), distributing them among several
connected computers (i.e., Grid infrastructure).
In this way, through such an infrastructure, LOs
can be dynamically discovered by client devices,
using, for example, the current learning needs of
users. Other solutions in this field include the work
of Amman, Bright, Quirchmayr, and Thomas
(2003) and the GetSmart system (Marshall, Zhang,
Chen, Lally, Shen, Fox, & Cassel, 2003).

Conclusion
In this chapter, we have discussed some concepts
surrounding pervasive computing. We have provided the reader with a high level understanding
of the pervasive computing paradigm at different
levels, and thus we have ranged from networking to software engineering issues. By using this
approach, we presented a broad vision related
to pervasive computing to aid researchers, students, and professors in research and teaching
activities.
Based on the concepts we have discussed,
it is possible to conclude that the application of
pervasive computing in the real world is still in its
beginning. Many efforts have still to be made in
order to bring the primary vision of pervasiveness
to real life. One could ask whether such vision
will really be conceived. This is, for now, a question which is still unanswered. Answering it will
require deep theoretical and, mainly, practical
studies. This is what researchers should focus
on, and thus, this work has been an introductory
theoretical contribution to this end. We believe
the concepts presented here will be summed up in
other works, and also be useful when developing

Pervasive Computing

real world applications, with the intent of reaching

a reasonable vision of pervasive computing.

References
Aldestein, F., Gupta, S.K.S., Richard, G.G., &
Schwiebert, L. (2005). Fundamentals of mobile
and pervasive computing. McGraw-Hill.
Amann, P., Bright, D., Quirchmayr, G., & Thomas,
B. (2003). Supporting knowledge management
in context aware and pervasive environments
using event-based co-ordination. In Proceedings
of the 14th International Workshop on Database
and Expert Systems Applications, Prague, Czech
Republic.
Apostolopoulos, T.K., & Kefala, A. (2003). A configurable middleware architecture for deploying
e-learning services over diverse communication
networks. In V. Uskov (Ed.), Proceedings of the
2003 IASTED International Conference on Computers and Advanced Technology in Education
(pp. 235-240). Rhodes, Greece. Acta Press.
Bachman, F., Bass, L., Buhman, C., Dorda, S.C.,
Long, F., Robert, J., Seacord, R., & Wallnau, K.
(2000). Technical concepts of component-based
software engineering (vol. 2). Pittsburgh, PA.
EUA: Carnegie Mellon Software Engineering
Institute. Retrieved October 10, 2006, from
http://www.sei.cmu.edu/pub/documents/00.reports/pdf/00tr008.pdf
Bardram, J.E., & Christensen, H.B. (2001). Middleware for pervasive healthcare: A white paper.
In G. Banavar (Ed.), Advanced topic workshop:
Middleware for mobile computing, Heidelberg,
Germany.
Becker, C., Handte, M., Schiele, G., & Rothermel, K. (2004). PCOM: A component system for
pervasive computing. In Proceedings of the 2nd
IEEE International Conference on Pervasive

Computing and Communications (pp. 67-76).

Orlando, FL: IEEE Computer Society.
Becker, C., Schiele, G., Gubbels, H., & Rothermel,
K. (2003). BASE-A micro-broker-based middleware for pervasive computing. In Proceedings
of the First IEEE International Conference on
Pervasive Computing and Communications (pp.
443-451). Fort Worth, USA: IEEE Computer
Society.
Birsan, D. (2005). On plug-ins and extensible architectures. ACM Queue, 3(2), 40-46, 197-207.
Boyera, S., & Lewis, R. (2005, May 2). Device
independence activity. Retrieved October 10,
2006, from http://www.w3.org/2001/di
Bray, J., & Sturman, C.F. (2000). Bluetooth: Connect without cables. Prentice Hall.
Chen, H., Finin, T., & Joshi, A. (2003). An ontology for context-aware pervasive computing
environments. Knowledge Engineering Review,
18(3), 197-207.
Chen, G., & Kotz, D. (2000). A survey of contextaware mobile computing research (Tech. Rep. No.
TR2000-381). Hanover: Dartmouth College.
Chen, G., & Kotz, D. (2002). Solar: A pervasive
computing infrastructure for context-aware mobile applications (Tech. Rep. No. TR2002-421).
Hanover: Dartmouth College.
Chlamtac, I., Conti, M., & Liu, J.J.N. (2003).
Mobile ad hoc networking: Imperatives and challenges. Ad hoc Networks, 1(1), 13-64.
Cohen, J., Aggarwal, S., & Goland, Y.Y. (2000,
September 6). General event notification architecture base: Client to arbiter. Retrieved October 10,
2006, from http://www.upnp.org/download/draftcohen-gena-client-01.txt
Costa, P., Coulson, G., Mascolo, C., Picco, G. P., &
Zachariadis, S. (2005). The RUNES middleware:
A reconfigurable component-based approach to

247

Pervasive Computing

network embedded systems. In Proceedings of the

16th Annual IEEE International Symposium on
Personal Indoor and Mobile Radio Communications (pp. 11-14). Berlin, Germany.

Garlan, D., Siewiorek, D., Smailagic, A., & Steenkiste, P. (2002). Project Aura: Toward distraction-free pervasive computing. IEEE Pervasive
Computing, 1(2), 22-31.

Coulson, G., Grace, P., Blair, G., Duce, D., Cooper,

C., & Sagar, M. (2005). A middleware approach
for pervasive grid environments. Paper presented
at the Workshop on Ubiquitous Computing and
E-research, Edinburgh, Scotland.

Gong, L. (2001). JXTA: A networking programming environment. IEEE Internet Computing,

5(3), 88-95.

Crnkovic, I. (2001). Component-based software

engineering: New challenges in software development software. Software Focus, 4, 127-133.
dAcierno, A., Pietro, G.D., Coronato, A., & Gugliara, G. (2005).
Plugin-orb for applications in
a pervasive computing environment. In Proceedings of the 2005 International Conference on
Pervasive Systems and Computing. Las Vegas,
Nevada: CSREA Press.
Dey, A.K. (2001). Understanding and using
context. Personal and Ubiquitous Computing,
5(1), 4-7.
Esler, M., Hightower, J., Anderson, T., & Borriello,
G. (1999). Next century challenges: Data-centric
networking for invisible computing (The Portolano Project at the University of Washington). In
Proceedings of the 5th International Conference
on Mobile Computing and Networking (pp. 2435). Seattle, Washington: ACM Press.
Fiege, L., Mhl, G., & Grtner, F.C. (2002).
Modular event-based systems. The Knowledge
Engineering Review, 17(4), 359-388.
Foster, I., Kesselman, C., Nick, J., & Tuecke,
S. (2002). Grid services for distributed system
integration. Computer, 35(6), 37-46.
Gamma, E., Helm, R., Johnson, R., & Vlissides,
J. (1995). Design patterns: Elements of reusable object-oriented software. Addison-Wesley
Professional.

248

Gu, T., Pung, H.K., & Zhang, D.Q. (2004a).

A
Bayesian approach for dealing with uncertain
contexts. In G. Kotsis (Ed.), Proceedings of
the 2nd International Conference on Pervasive
Computing, Vienna, Austria. Austrian Computer
Society.
Gu, T., Pung, H.K., & Zhang, D.Q. (2004b).
Toward an OSGi-based infrastructure for contextaware applications. IEEE Pervasive Computing,
3(4), 66-74.
Guttman, E. (2001). Autoconfiguration for IP
networking: Enabling local communication. IEEE
Internet Computing, 5(3), 81-86.
Gwizdka. J. (2000). Whats in the context? (Position Paper for Workshop on the What, Who,
Where, When, Why, and How of Context-Awareness). The Hague: The Netherlands.
Held, A., Buchholz, S., & Schill, A. (2002).
Modeling of context information for pervasive
computing applications. In Proceedings of the 6th
World Multiconference on Systemics Cybernetics
and Informatics, Orlando, Florida.
Henricksen, K., Indulska, J., & Rakotonirainy,
A. (2002).
Modeling context information in
pervasive computing systems. In F. Mattern &
M. Naghshineh (Eds.), In Proceedings of the
First International Conference on Pervasive
Zurich, Switzerland:
Computing (pp. 167-180).
Springer-Verlag.
Henricksen, K., Livingstone, S., & Indulska, J.
(2004).
Towards a hybrid approach to context

Pervasive Computing

modeling, reasoning, and interoperation. In

Proceedings of the First International Workshop
on Advanced Context Modeling, Reasoning,
and Management, Nottingham, UK (pp. 54-61).
ACM Press.
Lee, C., Nordstedt, D., & Helal, S. (2003). Enabling smart spaces with OSGi. IEEE Pervasive
Computing, 2(3), 89-94.
Liao, C. J., Yang, F. C., Hsu, K. (2005). A service-oriented approach for the pervasive learning
grid. Journal of Info. Science and Engineering,
21(5), 959-971.
Loureiro, E., Bublitz, F., Barbosa, N., Perkusich,
A., Almeida, H., & Ferreira, G. (2006). A flexible middleware for service provision over heterogeneous networks. In Proceedings of the 4th
International Workshop on Mobile Distributed
Computing, Niagara Falls, New York. IEEE Computer Society. (Accepted for publication)
Lytras, M., Pouloudi, A., & Poulymenakou, A.
(2002). Knowledge management convergence:
Expanding learning frontiers. Journal of Knowledge Management, 6(1), 40-51.
Ma, T., Kim, Y.D., Ma, Q., Tang, M., & Zhou,
W. (2005). Context-aware implementation
based
on CBR for smart home
. In Proceedings of IEEE
International Conference on Wireless and Mobile
Computing, Networking and Communications,
Montreal, Canada (pp. 112-115). IEEE Computer
Society.
Marshall, B., Zhang, Y., Chen, H., Lally, A., Shen,
R., Fox, E., & Cassel, L. (2003). Convergence
of knowledge management and e-learning: The
GetSmart experience
. In Proceedings of the Joint
Conference on Digital Libraries, Houston, Texas
(pp. 135-146). IEEE Computer Society.
Marwick, A.D. (2001). Knowledge management
technology. IBM Systems Journal, 40(4), 814831.

Mascolo, C., Capra, L., & Emmerich, W. (2002).

Mobile computing middleware. Advanced
Lectures on Networking (pp. 20-58). SpringerVerlag.
Masuoka, R., Labrou, Y., Parsia, B., Sirin, E.
(2003). Ontology-enabled pervasive computing
applications. IEEE Intelligent Systems, 18(5),
68-72.
Mayer, J., Melzer, I., & Schweiggert, F. (2002).
Lightweight plug-in-based application development. In M. Aksit, M. Mezini, & R. Unland (Eds.),
Revised Papers from the International Conference
on Objects, Components, Architectures, Services,
and Applications for a Networked World, Erfurt,
Germany (pp. 87-102). Springer-Verlag.
Mostfaoui, G.K., Rocha, J.P., & Brzillon, P.
(2004).
Context-aware computing: A guide for
the pervasive computing community. In F. Mattern & M. Naghshineh (Eds.), Proceedings of
the 2004 IEEE/ACS International Conference on
Pervasive Services, Beirut, Lebanon (pp. 39-48).
IEEE Computer Society.
Nishigaki, K., Yasumoto, K., Shibata, N., Ito, M.,
& Higashino, T. (2005). Framework and rule-based
language for facilitating context-aware computing
using information appliances. In Proceedings of
the First International Workshop on Services
and Infrastructure for the Ubiquitous and Mobile
Internet (pp. 345-351). Columbus, Ohio: IEEE
Computer Society.
Perkins, C.E. (1997). Mobile IP. IEEE Communications, 35(5), 84-99.
Petrelli, D., Not, E., Strapparava, C., Stock, O.,
& Zancaro, M. (2000).
Modeling context is like
taking pictures. In Proceedings of the Workshop
on Context Awareness. The Hague: The Netherlands
Raatikainen, A.K., Christensen, H.B., & Nakajima, T. (2002). Application requirements for

249

Pervasive Computing

middleware for mobile and pervasive systems.

Mobile Computing Communication Review,
6(4), 16-24.
Ranganathan, A., Muhtadi, J.A., & Campbell,
R.H. (2004). Reasoning about uncertain contexts
in pervasive computing environments. IEEE
Pervasive Computing, 3(2), 62-70.
Reid, N.P., & Seide, R. (2002). Wi-Fi (802.11)
network handbook. McGraw-Hill.
Robinson, J., & Wakeman, I. (2003). The scooby
event based pervasive computinginfrastructure.
In Proceedings of the Postgraduate Networking
Conference, Liverpool, UK.
Ryan, N. (1999, August 6). ConteXtML: Exchanging contextual information between a mobile client
and the FieldNote server. Retrieved October 10,
2006, from http://www.cs.kent.ac.uk/projects/mobicomp/fnc/ConteXtML.html

and meeting room projects: Signals, acquisition,

annotation and metrics. In Proceedings of the
2003 IEEE Conference on Acoustics, Speech,
and Signal Processing, Hong Kong, China. IEEE
Computer Society.
Sun, J.Z., & Savoula, J. (2002). Mobility and
mobility management: A conceptual framework.
In Proceedings of the 10th IEEE International
Conference on Networks (pp. 205-210). Singapore:
IEEE Computer Society.
Waldo, J. (1999). The Jini architecture for network-centric computing. Communications of the
ACM, 42(7), 76-82.
Weiser, M. (1991). The computer for the 21st century. Scientific American, 265(3), 94-104.
Weiser, M. (1993). Ubiquitous computing. Computer, 26(10), 71-72.

Saha, D., & Mukherjee, A. (2003). Pervasive

computing: A paradigm for the 21st century.
Computer, 36(3), 25-31.

Weiser, M., & Brown, J.S. (1995). Designing calm

technology. PowerGrid Journal, 1(1). Retrieved
October 10, 2006, from http://www.ubiq.com/
weiser/calmtech/calmtech.htm

Satyanarayanan, M. (1996). Fundamental challenges in mobile computing. In Proceedings of

the 15th Annual ACM Symposium on Principles
of Distributed Computing (pp. 1-7). Philadelphia:
ACM Press.

Will, O.M., Lech, C.T., & Klein, B. (2004). Pervasive knowledge discovery: Continuous lifelong
learning by matching needs, requirements, and
resources. In Proc. of the 4th International Conf.
on Knowledge Management, Graz, Austria.

Satyanarayanan, M. (2001). Pervasive computing:

Vision and challenges. IEEE Personal Communication, 8(4), 10-17.

Yau, S.S., Karim, F., Yu, W., Bin, W., & Gupta,
S.K.S. (2002). Reconfigurable context-sensitive
middleware for pervasive computing. IEEE Pervasive Computing, 1(3), 33-40.

Satyanarayanan, M. (2002). The evolution of

coda. ACM Transactions on Computer Systems,
20(2), 2-25.
Schilit, B., & Theimer, M. (1994). Disseminating
active map information to mobile hosts. IEEE
Network, 8(5), 22-32.
Stanford, V.
, Garofolo, J., Galibert, O., Michel,
M., & Laprun, C. (2003). The

NIST smart space

250

Endnote
1

Courtesy of the Xerox Palo Alto Research

Center - http://www.ubiq.com/hypertext/
weiser/UbiHome.html

Pervasive Computing

Appendix I: Internet Section:

Ubiquitous Computing Grand
Challenge
The Ubiquitous Computing Grand Challenge is a
community of researchers from different parts of
the world targeting ubiquitous computing (http://
www-dse.doc.ic.ac.uk/Projects/UbiNet/GC).
Their research is focused at different levels of
ubiquitous computing, from social to technological. With this purpose, this community has been
proposing a set of, as they call it, foothill projects,
within the scope of ubiquitous computing. This
community still provides a mailing list used by
their members to discuss the projects and directions of the community, among other adjacent
topics. The registration to the list is open.

Interaction
Select one of the foothill projects presented on the
Ubiquitous Computing Grand Challenge Web site
and prepare one of the following items:
1.
2.

A research paper with at least 3000 words,

without counting images, or
A presentation at least 40 minutes long.

Both the paper and the presentation should

present in what way the selected project contributes to the ubiquitous computing research, the state
of the art, and practical applications. Remember
that it could be helpful to access the mailing list,
for example, to generate some interesting discussion concerning the selected foothill project.

Appendix II: Case Study

A Usual Day with Pervasive
Computing
At half past six in the morning, Jessicas alarm
clock wakes her up, as programmed. As soon as
she gets off the bed, the curtains of the bedroom
are automatically opened and the alarm clock stops
yelling. Within one hour, she dresses for work,
takes her breakfast, gets her stuff, and leaves.
When arriving at work, a camera positioned at
the entrance of the parking lot recognizes Jessica
and her car, and thus, the gates are automatically
opened. She greets the gateman and enters. At
this point, her car automatically selects the best
parking options. She chooses one of them, and
leaves the parking lot.
Jessica is about to enter the company when
a sensor in the entry of the building detects her
presence, and by knowing the elevator she usually
takes to go up, warns her that it is on maintenance.
Based on the number of people waiting at each
elevator, it suggests the best option for Jessica.
Finally, Jessica gets to her office, and at this
moment environmental conditions, like lighting
or air conditioning levels or curtains opening,
are automatically adjusted according to her
preferences. Furthermore, a coffee machine,
which knows that Jessica usually drinks a cup
of coffee in the morning, greets her and asks if
she would like some. She has a meeting within a
few minutes, but she thinks she has enough time
to taste her daily coffee. Then, through a voice
command, Jessica orders the machine to prepare
it. After finishing the coffee, she then leaves the
room for another day of work.
Late in the night, Jessica gets ready to return
home. When leaving the parking lot, her car
informs her that the usual way she takes home is
too congested and it automatically provides an
alternative route.

251

Pervasive Computing

On her way home, she receives a phone call

on her cellular phone. As she is currently driving, the phone redirects the callers talk to an
audio output located in the car. Jessica can use
an available microphone to talk to the caller by
forwarding the audio streaming to the cellular
phone. Therefore, the focus on the driving is not
compromised.
When she gets home, the mail box identifies
her and notifies her about the three letters left in
it. She takes the correspondences and enters her
home, when again the environment is adjusted
according to her preferences. Furthermore, the
answering machine detects her presence and automatically informs her of missed calls. Through
a voice command, Jessica starts listening to each
message left. She then finds a message by her
mother, asking her to return the call as soon the
she gets home. Jessica stops listening to the missed
calls and orders the answering machine to dial
her mothers number through her cellular phone.
After talking to her, Jessica has dinner and finally
gets ready to sleep. She lies down on her bed, and
automatically the air conditioning is turned on,
the curtains are closed, and the alarm clock is set
to wake her up in the morning to have one more
usual day with pervasive computing.

Questions
1.

2.

3.

252

What other pervasive computing features

would be helpful for Jessicas daily activities?
For each feature provide a detailed description of how it would be fit into the above
scenario.
Do you think the current society is ready
for this kind of technology? Explain your
answer.
Considering the hardware and software technologies we have today, identify/describe
issues of creating the pervasive computing
scenario described.

Appendix III: Useful URLs

UPnP Forum Web site
http://www.upnp.org
Wi-Fi Alliance
http://www.wi-fi.org
Bluetooth Special Interest Group
https://www.bluetooth.org
What is General Packet Radio Service?
http://www.gsmworld.com/technology/gprs/intro.
shtml
Zigbee Alliance
http://www.zigbee.org
Home Networking with Zigbee
ht t p://w w w.embedded.com /showA r t icle.
jhtml?articleID=18902431
Mobile IP for IPv4
http://www.ietf.org/html.charters/mip4-charter.
html
Zero Configuration Networking (Zeroconf)
http://www.zeroconf.org
OSGi Alliance
http://www.osgi.org

Appendix IV: Further Reading

Banavar, G., & Bernstein, A. (2002). Software
infrastructure and design challenges for ubiquitous computing applications. Communications of
the ACM, 45(12), 92-96.
Gupta, R., Talwa, S., & Agrawal, D.P. (2002).
Jini home networking: A step toward pervasive
computing. Computer, 35(8), 34-40.

Pervasive Computing

Kallio, P., Niemel, E., Latvakoski, J. (2004).

October
UbiSoft pervasive software. Retrieved
10, 2006, from http://www.it.lut.fi/kurssit/0405/010651000/Luennot/T2238.pdf
Kindberg, T., & Fox, A. (2002). System software
for ubiquitous computing. IEEE Pervasive Computing, 1(1), 70-81.
Landay, J.A., & Borriello, G. (2003). Design

patterns for ubiquitous computing. Computer,

36(8) 93-95.
Milanovic, N., & Milutinovic, V. (2003). Ad hoc
networks and the wireless Internet. In V. Milutinovic & F. Patricelli (Eds.), Mastering e-business
infrastructure (pp. 255-335). Kluwer Academic
Press. Retrieved October 10, 2006, from http://informatik.hu-berlin.de/~milanovi/chapter6.zip
Plymale, W.O. (2005). Pervasive computing goes
to school. EDUCAUSE Review, 40(1), 60-61.
Tarkoma, S., Balu, R., Kangasharju, J., Komu, M.,
Kousa, M., Lindholm, T., Mkel, M., Saaresto,
M., Slavov, K., & Raatikainen, K. (2004). State
of the art in enablers of future mobile wireless
Internet. HIIT Publications. Retrieved October

10, 2006, from http://www.hiit.fi/publications/

pub_files/fc-state-of-the-art-2004.pdf
Weiser, M. (1994). The world is not a desktop.
Interactions, 1(1), 7-8.
Weiser, M., Gold, R., & Brown, J.S (1999). The
origins of ubiquitous computing research at
PARC in the late 1980s. IBM Systems Journal,
38(4), 693-696.

Appendix V: Possible Paper

Titles/Essays

Pervasive computing: a utopia of human

mind or a promising paradigm for everyday
life?
Social impacts of pervasive computing
Bringing pervasiveness to the real world:
practical applications of pervasive computing
Dealing with privacy of context information
in pervasive computing systems
The impact of pervasive computing on
software development

This work was previously published in Ubiquitous and Pervasive Knowledge and Learning Management: Semantics, Social
Networking and New Media to Their Full Potential, edited by M. Lytras and A. Naeve, pp. 1-34, copyright 2007 by IGI Publishing, formerly known as Idea Group Publishing (an imprint of IGI Global).

253

254

Chapter 1.19

Pervasive and Grid Computing

Merging
Loreno Oliveira
Federal University of Campina Grande, Brazil
Emerson Loureiro
Federal University of Campina Grande, Brazil
Hyggo Almeida
Federal University of Campina Grande, Brazil
Angelo Perkusich
Federal University of Campina Grande, Brazil

INTRODUCTION
Nowadays, we are experiencing an increasing use
of mobile and embedded devices. These devices,
aided by the emergence of new wireless technologies and software paradigms, among other
technological conquests, are providing means to
accomplish the vision of a new era in computer
science. In this vision, the way we create and use
computational systems changes drastically for a
model where computers loose their computer appearance. Their sizes were reduced, cables were
substituted by wireless connections, and they are
becoming part of everyday objects, such as clothes,
automobiles, and domestic equipments.

Initially called ubiquitous computing, this paradigm of computation is also known as pervasive
computing (Weiser, 1991). It is mainly characterized by the use of portable devices that interact
with other portable devices and resources from
wired networks to offer personalized services to
the users. While leveraging pervasive computing,
these portable devices also bring new challenges
to the research in this area. The major problems
arise from the limitations of the devices.
At the same time that pervasive computing was
attaining space within the research community,
the field of grid computing (Foster, Kesselman,
& Tuecke, 2001) was also gaining visibility and
growing in maturity and importance. More than

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Pervasive and Grid Computing Merging

just a low cost platform for high performance

computing, grid computing emerges as a solution
for virtualization and sharing of computational
resources.
In the context of virtual organizations, both
grid and pervasive computing assemble a number
of features that are quite desirable for several
scenarios within this field, notably the exchanging of information and computational resources
among environments and organizations. The
features of these technologies are enabling system
designers to provide newer and enhanced kinds of
services within different contexts, such as industry, marketing, commerce, education, businesses,
and convenience. Furthermore, as time goes on,
researchers have made attempts of extracting and
incorporating the better of the two technologies,
thus fostering the evolution of existing solutions
and the development of new applications. On the
one hand, pervasive computing researchers are
essentially interested in using wired grids to hide
the limitations of mobile devices. On the other
hand, grid computing researchers are broadening
the diversity of resources adhered to the grid by
incorporating mobile devices.
This chapter presents part of our experiences
in the research of both pervasive and grid computing. We start with an overview about grid
and pervasive technologies. Following, there are
described and discussed approaches for combining
pervasive and grid computing. These approaches
are presented from both perspectives of grid and
pervasive computing research. Finally, in the last
section, there are presented our criticisms about
the approaches discussed and our hopes about the
future steps for this blend of technologies.

GRID AND PERVASIVE COMPUTING

Grid computing (in short grids) was born as a low
cost alternative for high performance computing,
rapidly evolving to a solution for virtualization

and sharing of resources. By spreading workload

across a large number of computers, users take
advantage of enormous computational resources
that would otherwise be prohibitively expensive
to attain with supercomputers. To day, grids try
to leverage a world where the access to any computational resources across sites is standardized,
ubiquitous and reliable.
Grids are characterized by the use of heterogeneous and non-dedicated resources, scattered
under different administrative domains and linked
through low-bandwidth channels. There are no
guarantees neither whether a certain resource will
be available at a certain moment nor, once available, it will keep available during any interval of
time. Due to these features, grids are best suited
for applications that can adapt themselves to deal
with intermittent resources.
More recently, grids are going towards defining
a service-oriented architecture, OGSA1, where
resources are advertised and used as services.
OGSA extends traditional service-oriented architectures by the standardization of a number of
helper services, concerned with the maintenance
of the whole infrastructure. Such services include
load balancing, QoS2 assurance, redundancy of
services, and so forth.
Pervasive computing is also a kind of distributed computing. The general vision is that
computation is embedded in objects of everyday
life, such as clothes, rooms, automobiles and
domestic devices. These objects and environments collaborate on behalf of people, balancing
proactivity, context information and pre-known
information about users profiles to provide personalized services to users, on the right place and
on the right moment.
Pervasive computing materializes in the form
of smart spaces. These environments are characterized by collaboration between embedded
devices, wired infrastructure and users. Normally,
users interact with smart spaces through portable
devices. These devices carry information about

255

Pervasive and Grid Computing Merging

the users profiles and may tell the environment

how the user wishes to interact with it. Alternatively, the environment itself may know about
users profiles, and query them by sensing users
through sensors. Regardless of how the environment gets aware of users profiles and how it
senses the presence of users, the basic principle of
pervasive computing is to disappear from human
perception. In other words, users should benefit
from decisions made by environments without
being prompted every time about what they wish
the environment does for them.

THE PERVASIVE PERSPECTIVE

Pervasive computing is mainly characterized by
the use of resource constrained mobile devices.
Thus, from the perspective of pervasive computing, the main research focus is on using resources
of grids for supplying lacks of mobile devices.
Current approaches for bringing together grids
and pervasive computing will be discussed in the
following.

Using Grids for Hiding Limitations of

Mobile Clients
In the perspective of pervasive computing, using
grids for supplying the lack of resources of mobile
devices seems to be the major contribution that
grids may bring to the pervasive computing research. Grids have plenty supply of resources that
are needed by mobile clients. Nevertheless, current attempts to virtualize the capacity of mobile
clients (Clarke & Humphrey, 2002; Hingne, Joshi,
Finin et al., 2003; Kumar & Song, 2005;) still falls
into strong assumptions about the communication between mobile client and grid. Particularly,
successful applications of this approach are yet
restricted for those cases where the interaction
between mobile client and grid requires low data
exchange or where some alternative is found for
supplying the system with needed data.

256

For instance, Kumar and Song (2005) proposed

a grid-based infrastructure where mobile clients
run only helper applications linked to the grid.
As proof of concept, a virtual file system was
developed linking some folders in the PDAs file
system to a bigger hard disk in the grid. Thus,
files and full programs may be moved to the grid,
staying in the PDA only those files and programs
more frequently used. All content moved to the
grid is consumed on demand through streams.
The approach of Kumar and Song (2005)
reveals the main weaknesses of integrating pervasive and grid computing. Due to limitations of
battery autonomy, portable devices should avoid
using intensively their network interfaces. Therefore, a successful integration between grid and
pervasive computing should consider applications
where there are not large amounts of data being
exchanged between portable devices and grid.
The issues of restricted data exchange between
portable devices and grid are better targeted by
Hingne et al. (2003). In this approach, the authors
consider a niche of applications where data needed
by portable clients require intensive processing
over large amounts of data. However, the authors
assume these data can be obtained from different
fonts than users devices, such as sensor networks
and databases, both directly linked to the grid.
In this scenario, users use portable devices
to perform queries for data that need to be transferred, and eventually created, on-demand. Data
can be directly gathered from data sources, as a
databank or some sensor, or may need to be created from the processing of some other data. In
the last case, a grid is used to process data and
thus reduce the overall client waiting time.
Besides, on the one hand, the approach of
Hingne et al. (2003) presents a solution for the
issues discussed before, on the other hand it may
be posing an economical challenge. Deploying
an entire infrastructure, for example a sensor
network, to substitute the mobile device role of
acting as the systems entrance of data may be
prohibitively expensive. Also, this solution may

Pervasive and Grid Computing Merging

restrict user mobility in the sense that users become restricted to interact with the system only
where the infrastructure was deployed.

Using Grids as Test Bed of

Pervasive Applications
Adequate test beds are used for revealing problems
very difficult to identify in traditional development
environments. The development of any kind of
application achieves better results when prototypes are tested and developed in situations and
scenarios close to the real ones. There are many
examples of well succeeded test beds for classes
of applications, for example PlanetLab, which is
an overlay network over the Internet intended to
be a test bed of distributed applications, specially,
the peer-to-peer ones.
Keeping this in mind, Davies, Friday, and
Storz (2004) set up their own approach for getting
benefit from grids. The authors underline common
characteristics of grids and pervasive computing
and use this as a starting point for defining a gridbased test bed of pervasive applications.
The ability of testing and developing pervasive
applications in an environment that simulates real
environments is certainly a great contribution.
Problems that are difficultely detected inside
traditional development environments are easier
identified and fixed when applications are tested
in appropriate environments.
Nevertheless, some questions remain open.
For instance, grids are traditionally built over
wired networks and smart spaces are composed
of wireless devices. In order to try to be closer
to real smart spaces, such a grid-based test bed
must simulate the network interfaces of mobile
clients, including bandwidth, power consumption
and coverage area. Another point of investigation
is the gap of performance between small portable
devices and computers of wired grids. Tests
performed in hosts rather powerful than the real
devices where the applications should run may

hide problems that should be identified in this

phase of tests. After all, it is still needed to know
whether the tested application is the same one to
be deployed in mobile devices. This becomes an
issue due to differences in architectures, as well
as hardware and software-specific optimizations
that may contrast the results given in the test and
deployment phases.

Using Grids to Collect and Process

Data from Sensor Networks
Embedding computation in everyday objects
means, most of time, deploying sensors and
actuators across environments. Sensors are responsible for bringing to the computational world
information about the physical world. Conversely,
actuators obtain information from the computational world and are responsible for transforming
information into actions in the physical world.
Sensors may produce large amounts of data that
must be processed and interpreted before an action is taken. Processing large amounts of data
generally implies the use of large amounts of
computational power. Therefore, grids may play
an important role in gathering and processing
information from sensor networks (Barratt et al.,
2003; Gaynor, Moulton, Welsh et al., 2004).
In this context, Gaynor et al. (2004) discuss
challenges and research directions for integrating
sensor networks to a standard wired grid. Despite
all challenges, this approach seems to balance
advantages from both technologies. In fact, this
alternative requires not much effort to be applied,
since it is a direct application of grids. As posed
by Gaynor et al. (2004), data standardization is
possibly a challenging task in providing such
architecture. Once collected from sensors, it is
important to share data in a uniform manner. Thus,
different applications and devices may consume
data without being overburden with logics for
transforming data to an understandable format.

257

Pervasive and Grid Computing Merging

THE GRID PERSPECTIVE

Traditionally, grid computing research has been
focusing on high performance. Even with the aim
change for sharing any kind of computational
resources instead of only CPU cycles and storage space, the grid community is still hungry for
more computational power. In this context, the
main focus of the grid community is to use mobile
devices in the grid for solving computational tasks
and storing files. In what follows, approaches for
getting benefit of pervasive computing inside the
grid research are summarized.

Augmenting Grid Resources through

Mobile Devices
The general idea is to use small mobile devices
as nodes of high performance grids (Litke, Skoutas, & Varvarigou, 2004; McKnight, Howison,
& Bradner, 2004; Phan, Hung, & Dulan, 2002).
Nodes connect to the grid as soon as they approximate to some access point and are disconnected
from it when they leave the access point coverage
area. While connected to the grid, mobile devices
can receive and submit tasks.
This may sound a little atypical considering
all restrictions of small mobile devices. However,
research groups around the world bet in the
whole number of small devices working together
instead of the computational power of one lone
device. Phan et al. (2002) define an architecture
as including mobile devices into a wired grid.
Mobile devices enter the grid through proxies. This
proxy runs an appropriate grid middleware, such
as Globus or Condor, to publish itself as a node.
Mobile devices are not visible to the rest of the
grid. They only contribute to enlarge the computational power of the proxy, which is responsible
for decomposing and spreading received work
across the mobile devices linked to it.
Besides all drawbacks discussed before, this
approach has yet the serious problem of how a
given task can be broken into smaller tasks for

258

sharing with mobile devices. Definitely, this is

a non-trivial problem and may overburden grid
users, who will need to provide the means for a
proxy node to split a task into small ones.
In the grid perspective, augmenting the number of devices supported by the infrastructure is
always a good deal. Both mobile devices and grid
have an opportunity to obtain benefits from this
merging. Moreover, constant advances in mobile
technology trend to reduce the drawbacks of using
mobile devices for high processing. Nonetheless,
the actual technology still poses several challenges
for this usage of mobile devices. In addition to
issues already discussed, the mobile nature itself
of these small devices may cause loss of work. For
instance, when a portable device receives some
work to perform and, just before concluding it,
the device leaves the access point coverage area.
There is also an economic question to be solved.
Small mobile devices have clearly great potential
advantages of being part of a grid in the sense that
a grid may supply it with all the resources such
devices may need. However, while part of a grid
economy, mobile devices outwears themselves
rather more than conventional grid resources.
Therefore, the grid economic model itself should
adapt to address this new class of devices, making this integration profitable for both mobile
and grid users.

Mobile/Wireless Grids
Mobile grids have as main argument carrying
grids where cables do not reach (Chu & Humphrey, 2004; Kurkovsky & Bhagyavati, 2003;
Park, Ko & Kim, 2003). The basic idea is to use
high performance where the wired infrastructure
is not available and clustering mobile devices is
the last choice. In this context, software programs
are ready to use resources of wired grids as well
as switch to an unplugged mode, where mobile
devices collaborate to achieve better throughput
for distributed processing.

Pervasive and Grid Computing Merging

As an example of this approach, Kurkovsky

and Bhagyavati (2003) combine mobile devices
to enable higher processing capacities for small
devices. To achieve this, the defined architecture
uses resources from an access point, for maintaining information about available devices in range
and distribution of tasks among nodes. Any mobile device may submit tasks to the mobile grid.
For this, the device must ask the access point to
coordinate the distribution of tasks, re-submissions and collecting of results.
The idea of using on-demand wireless high
processing sounds good, but certainly using a set
of mobile devices in place of a wired grid does not
solve the problem of high processing. However,
this may be the only alternative on critical situations (Akogrimo, 2006).
Again, overloading these resource constrained
devices trend to overdrive them, reducing thus
their autonomy and without gaining anything
with this. If we consider situations where portable
devices accept tasks whose processing time spends
more time then its autonomy, all effort for creating
a grid in an ad-hoc fashion will be lost.

Monitoring Grids and Controlling

Applications through Portable
Devices
This is the more simplistic use of mobile devices
in grids. The approach seems to have no great
challenges to be transposed. On the other hand, the
benefits obtained are rather limited. The general
idea is quite simple. In a wired grid, it is enough
to provide means for mobile clients to monitor
relevant pieces of the grid, such as the grid healthy
and application progress (Srinivasan, 2005; Taylor, Shields, Wang & Philp, 2003). Alternatively,
users can somewhat control the grid operation,
changing the way the work is doing.
The clear advantage of this approach is the
low burden over portable devices. This approach
requires low interaction and low data exchange
between mobile clients and grid. There are also

restrictions about the kind of data to be monitored. If large amounts of data are needed to be
transferred to mobile devices then this approach
may be inadequate owning to reasons already
discussed in this chapter. This approach may be
unsuitable in those scenarios where users want
real-time access to monitoring data. Depending
on the size of data, displaying a continuous stream
of data may overdrive mobile devices. (Taylor et
al., 2003) also relates to this approach. In this
work Taylor et al. (2003) use a grid to simulate
the formation of galaxies. Mobile devices may
be used to monitor the workflow of grid services
involved in the process.

CONCLUSION
In this chapter we discussed various approaches
for combining grid and pervasive computing. The
approaches were grouped into categories that were
exemplified and criticized. The list is not exhaustive, but we believe it is sufficiently representative
of how research groups are attempting to blend
the technologies.
Resource constraints of mobile devices still
are the bottleneck for this integration. In the perspective of pervasive computing, massive search
applications (e.g., cryptographic key breakers)
are the best example of class of application with
all features for a successful integration between
grids and pervasive computingsmall pieces of
data requiring intensive processing that produce
more small pieces of data.
All efforts presented here are valid and, in
well defined scenarios, they show the technologies
can be successfully integrated. The approaches
discussed in this chaper have their own application
niche and, hence, a space where they can evolve.
Nevertheless, our hopes are that while small
mobile devices were so resource constrained,
difficultely researches will have both freedom
and tools enough for developing their ideas.
On the other hand, there are several problems

259

Pervasive and Grid Computing Merging

related to both grid and pervasive computing

research: large-scale deployment, distributed resource management, economic models, interface
standardization, heterogeneity, interoperability,
scalability, runtime adaptability, fault tolerance,
service composition, service discovery, security,
resources semantic, and so on. Hence, while all the
issues pointed out in this chapter were yet challenging tasks, both grids and Pervasive computing may contribute to each other by exchanging
experiences.

REFERENCES
Akogrimo. (2006). AkogrimoAccess to knowledge through the grid in a mobile world. Retrieved
April 8, 2006 from www.mobilegrids.org
Barratt, C., Brogni, A., Chalmers, M., Cobern,
W.R., Crowe, J., Cruickshank et al. (2003). Extending the grid to support remote medical monitoring, Proceedings of the 2nd UK e-Science All
Hands Meeting (AHM03), Nottingham, United
Kingdom.
Chu, D.C., &Humphrey, M. (2004). Mobile
OGSI.NET: Grid computing on mobile devices.
Proceedings of the 5th IEEE/ACM International
Workshop on Grid Computing (GRID04), (pp.
182-191). Pittsburgh, PA, USA.
Clarke, B., & Humphrey, M. (2002). Beyond the
device as portal: Meeting the requirements of
wireless and mobile devices in the legion grid
computing system. In Proceedings of the 2nd
International Workshop on Parallel and Distributed Computing Issues in Wireless Networks and
Mobile Computing (WPIM02), (pp. 192-199). Ft.
Lauderdale, FL, USA.
Davies, N., Friday, A., & Storz, O. (2004). Exploring the grids potential for ubiquitous computing.
IEEE Pervasive Computing, 3(2), 74-75.

260

Foster, I., Kesselman, C., & Tuecke, S. (2001).

The anatomy of the grid: Enabling scalable
virtual organizations. International Journal of
High Performance Computing Applications,
15(3), 200-222.
Gaynor, M., Moulton, S.L., Welsh, M., LaCombe,
E., Rowan, A., & Wynne, J. (2004). Integrating
wireless sensor networks with the grid. IEEE
Internet Computing, 8(4),32-39.
Hingne, V., Joshi, A., Finin, T., Kargupta, H., &
Houstis, E. (2003). Towards a pervasive grid. In
Proceedings of the 17th International Parallel and
Distributed Processing Symposium (IPDPS03),
Nice, France
Kumar, R., & Song, X. (2005). GridLite: A framework for managing and provisioning services on
grid-enabled resource limited devices. Technical
Report, Mobile and Media Systems Laboratory,
HP Laboratories Palo Alto.
Kurkovsky, S., & Bhagyavati, B. (2003). Wireless
grid enables ubiquitous computing. In Proceedings of The 16th International Conference on
Parallel and Distributed Computing Systems
(PDCS 2003), Reno, NV, USA.
Litke, A., Skoutas, D., & Varvarigou, T. (2004).
Mobile grid computing: Changes and challenges
of resource management in a m

obile grid environment. In Proceedings of the 5th International

Conference on Practical Aspects of Knowledge
Management (PAKM04), Vienna.
McKnight, L. W., Howison, J. & Bradner, S.
(2004), Wireless grids: Distributed resource sharing by mobile, nomadic and fixed devices. IEEE
Internet Computing, 8(4), 2431.
Park, S.M., Ko, Y.B., & Kim, J.H. (2003). Disconnected operation service in mobile grid computing. Lecture Notes in Computer Science, 2910,
499-513.

Pervasive and Grid Computing Merging

Phan, T., Hung, L., & Dulan, C. (2002). Challenge: Integrating mobile wireless devices into
the computational grid. In Proceedings of the
8th Annual International Conference on Mobile
Computing and Networking (MobiCom02), (pp.
271278). Atlanta, GA, USA.
Srinivasan, S.H. (2005). Pervasive wireless grid
architecture. In Proceedings of the 2nd Annual
Conference on Wireless On-demand Network
Systems and Services (WONS05), (pp. 83-88).
St. Moritz, Switzerland
Taylor, I., Shields, M., Wang, I., & Philp, R. (2003).
Distributed P2P Computing within Triana: A galaxy visualization test case. In Proceedings of the
17th International Parallel and Distributed Processing Symposium (IPDPS03), Nice, France
Weiser, M. (1991). The computer for the 21st century. Scientific American, 265(3), 66-75.

Key Terms
Context Information: Any relevant information regarding the environment and its users.
Smart spaces can use context information to
deliver personalized services to users.
Embedded Devices: An embedded device
is a special-purpose computer system, which is
completely encapsulated by the device it controls.
Examples of embedded devices include home
automation products, like thermostats, sprinklers
and security monitoring systems.
Grid Services: A kind of web service. Grid
services extend the notion of web services through

the adding of concepts regarding the WSRF

specification, such as statefull services.
Intermittent Resources: Both in a grid and
smart space, intermittent resources are any kind
of resource, as computers or services, which may
appear or disappear at any time.
Mobile Devices: Any low-sized portable device used to interact with other mobile devices
and resources from smart spaces. Examples of
mobile devices are cellular phones, smart phones,
PDAs, notebooks and tablet PCs.
Peer-to-Peer: Architecture for distributed
applications without central control where each
participant acts both as client and server.
Sensor Networks: Network of many, spatially
distributed devices using sensors to monitor
conditions at different locations, such as temperature, sound, vibration, pressure, motion or
pollutants.
Service-Oriented Architecture: Collection
of services communicating with each other. Services are self-contained and do not depend on the
context or state of other services.
Test Bed: In the context of software engineering, an execution environment configured for
testing software.

Endnotes
1
2

Open grid service architecture

Quality of service

This work was previously published in the Encyclopedia of Networked and Virtual Organizations, edited by G.D. Putnik and
M.M. Cunha, pp. 1223-1229, copyright 2008 by Information Science Reference, formerly known as Idea Group Reference (an
imprint of IGI Global).

261

262

Chapter 1.20

Leveraging Pervasive and

Ubiquitous Service Computing
Zhijun Zhang
University of Phoenix, USA

Abstract
The advancement of technologies to connect
people and objects anywhere has provided many
opportunities for enterprises. This chapter will
review the different wireless networking technologies and mobile devices that have been developed,
and discuss how they can help organizations
better bridge the gap between their employees
or customers and the information they need. The
chapter will also discuss the promising application areas and human-computer interaction modes
in the pervasive computing world, and propose
a service-oriented architecture to better support
such applications and interactions.

Introduction
With the advancement of computing and communications technologies, people do not have to
sit in front of Internet-ready computers to enjoy
the benefit of information access and processing.
Pervasive computing, or ubiquitous computing,

refers to the use of wireless and/or mobile devices

to provide users access to information or applications while the users are on the go. These mobile
devices can be carried by the users, or embedded
in the environment. In either case, these devices
are connected, most likely through a wireless
network, to the Internet or a local area network
(LAN).
Mobile technologies come in a large variety and
are ever changing. In order to gain the business
value of pervasive computing, and at the same
time keep the supporting cost under control, it is
important to develop an architecture solution. A
service-oriented architecture (SOA) would allow
an enterprise to easily provision functions to be
accessible by certain types of pervasive channels.
A service-oriented architecture would also make
it possible to quickly integrate data generated by
pervasive devices and make them available in the
form of an information service.
In this chapter, we will first look at the communication networks and mobile devices that
create the various information-access and information-generation touch points in a pervasive

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Leveraging Pervasive and Ubiquitous Service Computing

computing environment. Then we will discuss

the applications and interaction models for pervasive computing. Finally, we will describe a
service-oriented architecture that an enterprise
can adopt in order to effectively and efficiently
support pervasive computing.

Mobile Communication
Networks
Mobile communication technologies range from
personal area networks (PANs; a range of about 10
meters) and local area networks (a range of about
100 meters) to wide area networks (WANs; a few
kilometers). From a network-topology perspective,
most networks are based on a client-server model.
A few are based on the peer-to-peer model.

Wireless PANs
A wireless personal area network allows the
different devices that a person uses around a
cubicle, room, or house to be connected wirelessly. Such devices may include the computer,
personal digital assistants (PDAs), cell phone,
printer, and so forth.
Bluetooth is a global de facto standard for
wireless connectivity (Bluetooth SIG, 2005). The
technology is named after the 10th-century Danish
King Harald, who united Denmark and Norway
and traveled extensively.

HomeRF is an early technology for wireless

home networking, first marketed in 2000.
The Institute of Electrical Engineers (IEEE)
802.15 wireless-PAN effort (IEEE, 2005a) focuses
on the development of common standards for
personal area networks or short-distance wireless networks.One technology out of this effort
is ZigBee, which is based on the IEEE 802.15.4
standard.
ZigBee is a low-cost, low-power-consumption,
wireless communication-standard proposal (ZigBee Alliance, 2005). Formerly known as FireFly,
ZigBee is being developed as the streamlined
version of HomeRF. A streamlined version would
allow most of the functionality with less integration and compatibility issues.
ZigBees topology allows as many as 250
nodes per network, making the standard ideal for
industrial applications. Radio-frequency-based
ZigBee is positioned to eventually replace infrared links. To achieve low power consumption,
ZigBee designates one of its devices to take on
the coordinator role. The coordinator is charged
with waking up other devices on the network that
are in a sleep mode, moments before packets are
sent to them. ZigBee also allows coordinators to
talk to one another wirelessly. This will allow
for opportunities for wireless sensors to continuously communicate with other sensors and to a
centralized system.
For enterprise computing, the wireless PANs
are within the corporate firewall. They do not

Table 1. Summary of the wireless PANs

Bluetooth

2.4 GHz

Maximum
Distance
10 meters

HomeRF

2.4 GHz

50 meters

ZigBee

2.4 GHz

75 meters

Technology

Radio Frequency

Data Capacity
721 Kbps
0.4-10 Mbps, depending on
distance
220 Kbps

263

Leveraging Pervasive and Ubiquitous Service Computing

create new requirements for the enterprise architecture to extend access to applications. However,
they do require security measures to make sure
the device that is receiving information is a recognized device. It also creates an opportunity for
the computing infrastructure to potentially know
where a particular device, and most likely the associated user, is located. How these are handled
will be discussed later in the description of the
proposed service-oriented architecture.

Wireless LANs
The set of technical specifications for wireless
local area networks (WLANs), labeled 802.11
by IEEE, has led to systems that have exploded
in popularity, usability, and affordability. Now
wireless LAN can be found in many organizations
and public places.
With a wireless LAN, a users device is connected to the network through wireless access
points (APs). APs are inexpensivemany are
available for less than $100and will usually work
perfectly with little or no manual configuration.
Wireless LANs use a standard, called IEEE
802.11, that provides a framework for manufactures to develop new wireless devices. The first
two standards released for wireless LANs were
802.11b and 802.11a. The 802.11b standard was
used in most wireless devices in the early adoption
of wireless LAN. A new standard, called 802.11g,
combines data-transfer rates equal to 802.11a with
the range of an 802.11b network (Geier, 2002). It
uses access points that are backward compatible
with 802.11b devices.
Wireless technology has become so popular
that many new devices, especially laptop computers, have built-in wireless LAN capabilities.
Windows XP, Mac OS, and Linux operating
systems automatically configure wireless settings,
and software such as NetStumbler and Boingo
provides automatic connections to whatever
WLANs they encounter. What is more, community-based groups have furthered neighborhood

264

area networks (NANs) to share wireless Internet

access from one building to the next.
Besides 802.11a/b/g technologies that have
shipped products, new technologies are emerging, including 802.11h, 802.11i, and 802.1x. The
most important developments for wireless security will be contained in the 802.11i and 802.1x
specifications. The 802.11i specification addresses
encryption (securing the communication channel), whereas 802.1x will address authentication
(verifying individual users, devices, and their
access levels).
IEEE 802.1x is another authentication protocol,
not an encryption protocol. 802.1x by itself does
not fix the existing problems with WLAN security
that relate to encryption. Therefore, attackers can
still easily read network traffic on 802.1x networks.
The 802.11i standard will address communication-channel encryption.
In order to increase the throughput of wireless
LANs, a technology called Mimo (multiple inputmultiple output) has been developed. Mimo allows
for transmission rates of more than 100 Mbps,
which is much greater than existing wireless
LANs. Presently, wireless LANs use a single antenna operating at only one of a limited number of
frequencies (channel) that are shared by all users.
Mimo technology allows the use of two or more
antennas operating on that channel. Normally,
this would cause interference degradation of the
signal because the radio waves would take different pathscalled multipath distortion. However,
Mimo uses each of these different paths to convey
more information. The Mimo technology corrects
for the multipath effects. IEEE is standardizing
the technology as IEEE 802.11n.
For an enterprise, wireless LAN technologies
allow pervasive information access throughout
the campus. Employees with authorized mobile
devices such as wireless laptops and PDAs will
be able to get online wherever they are on the
campus.
Table 2 summarizes the wireless LAN technologies.

Leveraging Pervasive and Ubiquitous Service Computing

Table 2. Summary of wireless LAN technologies

Technology
802.11a
802.11b
802.11g
802.11i
802.11n
802.1x

Maximum
Data Capacity
Distance
5 GHz
20 meters
54 Mbps
2.4 GHz
100 meters
11 Mbps
2.4 GHz
100 meters
54 Mbps
A security standard for encryption on wireless LANs
Varies
Varies
> 100 Mbps
A standard security protocol for user authentication on
wireless LANs

Radio Frequency

Wireless MANs
A wireless metropolitan area network (MAN;
also referred to as broadband wireless access,
or WiMAX) can wirelessly connect business to
business within the boundary of a city. It is becoming a cost-effective way to meet escalating
business demands for rapid Internet connection
and integrated data, voice, and video services.
Wireless MANs can extend existing fixed
networks and provide more capacity than cable
networks or digital subscriber lines (DSLs). One
of the most compelling aspects of the wireless
MAN technology is that networks can be created
quickly by deploying a small number of fixed-base
stations on buildings or poles to create high-capacity wireless access systems.
In the wireless MAN area, IEEE has developed
the 802.16 standard (IEEE, 2005b), which was
published in April 2002, and has the following
features.

It addresses the first mile-last mile connection in wireless metropolitan area networks.
It focuses on the efficient use of bandwidth
between 10 and 66 GHz.

It enables interoperability among devices

so carriers can use products from multiple
vendors. This warrants the availability of
lower cost equipment.
It defines mechanisms that provide for
differentiated quality of service (QoS) to
support the different needs of different
applications. The standard accommodates
voice, video, and other data transmissions
by using appropriate features.
It supports adaptive modulation, which effectively balances different data rates and
link quality. The modulation method may be
adjusted almost instantaneously for optimal
data transfer. Adaptive modulation allows
efficient use of bandwidth and fits a broader
customer base.

The WiMAX technical working group has

developed a set of system profiles, standards for
protocol-implementation conformance, and test
suites (http://www.wimaxforum.org).
One particular technology for WiMAX is non
line of sight (NLOS) networking (Shrick, 2002).
NLOS networks provide high-speed wireless
Internet access to residential and office facilities.

265

Leveraging Pervasive and Ubiquitous Service Computing

NLOS uses self-configuring end points that connect to a PC (personal computer). The end point
has small attached antennas and can be mounted
anywhere without the need to be oriented like
satellite antennas. Two major vendors are Navini
Networks and Nokia.
With the wireless MAN technology, enterprises can quickly set up a network to provide
wireless access to people in a certain area. It is
very useful in situations such as an off-site working session or meeting.

Wireless NANs
Wireless neighborhood area networks are community-owned networks that provide wireless
broadband Internet access to users in public areas (Schwartz, 2001). To set up a wireless NAN,
community group members lend out access to the
Internet by linking wireless LAN connections
to high-speed digital subscriber lines or cable
modems. These wireless LAN connections create
network access points that transmit data for up to
a 1-kilometer radius. Anyone possessing a laptop
or PDA device equipped with a wireless network
card can connect to the Internet via one of these
community-established access points.
Wireless NANs have been established in more
than 25 cities across the United States. Community-based networks differ from mobile ISPs
(Internet service providers) such as MobileStar
and Wayport that offer subscribers wireless access to the Internet from hotels, airports, and
coffee shops. Wireless NANs extend access to
consumers in indoor as well as outdoor areas,
and the access is typically offered at no charge.
For instance, NYC Wireless (http://www.nycwireless.net) provides Internet access to outdoor
public areas in New York City. In addition, this
organization is negotiating with Amtrak to bring
wireless Internet access to Penn Station.
Enterprises could leverage the existing wireless NANs and equip employees with the right

266

devices and security mechanisms in order to use

these wireless networks to securely connect to
the corporate network.

Wireless WANs
Wireless wide area networks are commonly
known as cellular networks. They refer to the
wireless networks used by cell phones.
People characterize the evolution of wireless
WAN technology by generation. First generation
(1G) started in the late 1970s and was characterized by analog systems. The second generation of
wireless technology (2G) started in the 1990s. It
is characterized by digital systems with multiple
standards and is what most people use today.
2.5G and 3G are expected to be widely available
1 to 3 years from now. 4G is being developed in
research labs and is expected to launch as early
as 2006.
Wireless WAN originally only offered voice
channels. Starting from 2G, people have used
modems to transmit data information over the
voice network. More recent generations offer
both voice and data channels on the same cellular network.
One of the major differentiating factors among
the wireless generations is the data transmission
speed in which the wireless device can communicate with the Internet. The table below is a comparison of the data transmission rates of the 2G,
2.5G, 3G, and 4G technologies (3Gtoday, 2005).
Both 2G and 2.5G include different technologies
with different data transmission rates. Global
Systems for Mobile Communications (GSM)
and Code Division Multiple Access (CDMA) are
2G technologies. General Packet Radio Service
(GPRS), CDMA 1x, and Enhanced Data for GSM
Environment (EDGE) are 2.5G technologies.
In the United States, cellular carriers Verizon
and Sprint use CDMA technology. Cingular uses
GSM, GPRS, and EDGE technologies. Both Verizon and Sprint have rolled out their CDMA 1x

Leveraging Pervasive and Ubiquitous Service Computing

Table 3. Data transmission speed of wireless wide area networks

Technology
2G: GSM
2G: CDMA
2.5G: GPRS
2.5G: CDMA 1x
2.5G: EDGE
3G
4G

Maximum
9.6 Kbps
14.4 Kbps
115 Kbps
144 Kbps
384 Kbps
2 Mbps
20 Mbps

services, which is 2.5G. Cingular has rolled out

GRPS service and is starting to roll out EDGE
service in selected markets.
Wireless WANs are available wherever cell
phones can be used. For now, they are the most
pervasive wireless networks. By subscribing to a
service plan, an enterprise users laptop computer
or other mobile device can connect to the Internet
through the service providers cellular towers.

Ultrawideband (UWB)
Traditional radio-frequency technologies send
and receive information on particular frequencies,
usually licensed from the government. Ultrawideband technology sends signals across the entire
radio spectrum in a series of rapid bursts.
Ultrawideband wireless technology can transmit data at over 50 Mbps. A handheld device
using this technology consumes 0.05 milliwatts
of power as compared to hundreds of milliwatts
for todays cell phones. Ultrawideband signals
appear to be background noise for receivers of
other radio signals. Therefore it does not interfere
with other radio signals. Ultrawideband is ideal
for delivering very high-speed wireless-network
data exchange rates (up to 800 Mbps) across
relatively short distances (less than 10 meters)
with a low-power source.

Initial

< 28 Kbps
32 Kbps
64 Kbps
< 128 Kbps
TBD

Typical

28-56 Kbps
32-64 Kbps
64-128 Kbps
128-384 Kbps
TBD

Another feature of ultrawideband signals is that

they can penetrate walls. Therefore, this technology would allow a wireless device to communicate
with a receiver in a different room. This feature
can also be used to detect buried bodies, people
in a building, or metal objects in concrete.

Mesh Radio and Mess Networks

Mesh radio is a wireless network technology that
operates in the 28-GHz range of the radio spectrum and provides high-speed, high-bandwidth
connectivity to the Internet (Fox, 2001). A mesh
radio network consists of antennas connected in
a web-like pattern to a fiber-optic backbone. A
single antenna attached to the roof of a building
could provide Internet access to all of the subscribers residing in the building. Each node on
the network has a small, low-power, directional
antenna that is capable of routing traffic for other
nodes within a 2.8-kilometer radius. In contrast to
other wireless networks, mesh radio avoids many
of the line-of-sight issues between the base station
and each node on the network. Consequently, the
configuration of mesh radio reduces the chance
of encountering physical obstructions that could
impede access to the network.
Mesh radio networks are being developed in
two different ways. CALY Networks has devel-

267

Leveraging Pervasive and Ubiquitous Service Computing

oped a system that utilizes the Internet protocol

(IP) as its communication mechanism, while
Radiant Networks has created a system that communicates using the asynchronous transfer mode
(ATM). Providers of mesh radio services include
British Telecommunications, TradeWinds Communications (http://www.tnsconnects.com), and
Nsight Teleservices (http://www.nsighttel.com).
Features of mesh radio include the following:

Provides upload and download data rates of

up to 25 Mbps
Supports up to 600 subscribers per square
kilometer without degradation of service
Provides cost-effective access to broadband
services in rural communities or urban
areas
Increases network capacity and resilience
as the customer base grows

Different from the mesh radio technology,

mesh networks enable wireless devices to work
as a peer-to-peer network, using the handsets
themselves instead of the radio towers to transmit
data (Blackwell, 2002). Each handset would be
capable of transmitting data at rates from 6 Mbps
to 18 Mbps. This technology can be used for a
group of users or devices communicating with
each other in a peer-to-peer mode without needing
an established wireless network. The technology
was developed by Mesh Networks Inc., which has
been acquired by Motorola.

Sensor Networks
Motes (also called sensor networks or Smart Dusts;
Culler & Hong, 2004) are small sensing and communication devices. They can be used as wireless
sensors replacing smoke detectors, thermostats,
lighting-level controls, personal-entry switches,
and so forth. Motes are built using currently

268

available technology and are inexpensive enough

to be deployed in mass quantities. Depending on
the sensors and the capacity of the power supply, a mote can be as big as 8 cubic centimeters
(the size of a matchbox) or as small as one cubic
millimeter.
Motes are the result of a joint effort between
Defense Advanced Research Projects Agency
(DARPA) and the University of California, Berkeley, research labs. Most initial applications are
positioned to helping the military for tasks such
as surveillance of war zones, the monitoring of
transportation, and the detection of missiles and/or
biological weapons. Commercial mote sensors are
available from Crossbow Technology.
A mote is typically made up of the following:

A scanner that can scan and measure information on temperature, light intensity,
vibrations, velocity, or pressure changes
A microcontroller that determines tasks
performed by the mote and controls power
across the mote to conserve energy
A power supply that can be small solar cells
or large off-the-shelf batteries
TinyOS, an open-source software platform
for the motes. TinyOS enables motes to selforganize themselves into wireless network
sensors.
TinyDB, a small database that stores the
information on a mote. With the help of
TinyOS, the mote can process the data and
send filtered information to a receiver.

These motes enable enterprises to constantly

collect important information and send the information to the appropriate server for processing
so that the appropriate response can be initiated
when necessary. The motes become the generator
of pervasive information that reflects the status
of business processes or environmental conditions.

Leveraging Pervasive and Ubiquitous Service Computing

Pervasive Devices

1.

Pervasive devices come in different forms and

shapes. Compared to a networked computer,
some pervasive devices, such as landline or cell
phones, are more widely available. Other devices
are simply more portable and thus can be easily
carried around. Yet other devices are embedded
in the environment and are able to deliver specialized information. In terms of their functions,
some are for accessing the Internet, some are just
for entering information while the user is on the
go, and others are for storing large amounts of
information and can be easily carried around.

2.

3.

4.

5.

Traditional Telephones, Pagers, and

Cell Phones
Traditional landline telephone has been the
most pervasive communication device around
the world. Voice markup languages such as
VoiceXML (voice extensible markup language;
Rubio, 2004), together with supporting technologies such as the voice browser and voice gateway,
has made the traditional telephone yet another
device for connecting the user to the Internet.
With speech recognition, users can choose to use
touch tone or simply say what they need. Figure
1 shows how a telephone can be used to connect
to the Internet.

The user dials the number from any phone

(landline or mobile).
The call is routed to the corresponding voice
gateway, which maps the phone number to a
particular application hosted at the enterprise
network.
The voice gateway knows the URL (uniform
resource locator) of the application. It uses an
HTTP (hypertext transfer protocol) request
to fetch the first dialog of the application.
The enterprise Web server and application
server return the dialog to the gateway in
the form of a VoiceXML document.
The gateway interprets the VoiceXML
document, plays the greeting, and asks the
user for input. Now the user can use touch
tone or speech to provide input. Based on
the user input and the application logic as
described in the VoiceXML file, the voice
gateway decides what dialog to fetch next
from the enterprise network.

Pagers allow users to receive alerts with a

limited amount of text. With two-way pagers,
users can also reply with a text message.
With cell phones (not smart phones), besides
the same communication capabilities of a landline telephone, most users can use short-message
service (SMS) to send and receive text messages.
This is good for near-real-time conversational
communications.

Figure 1. Voice gateway connects the phone network with the data network

phone
network

2
5

Voice
gateway

internet

enterprise
network

269

Leveraging Pervasive and Ubiquitous Service Computing

Smart Phones, Wireless PDAs, and

Blackberry Devices
Smart phones are cells phones that have both voice
and data capabilities. Such a cell phone comes
with a mini Web browser and thus can be used
to access Internet content. However, since the
smart phones typically have rather small screens,
they can only access pages specifically designed
for small screens and coded in a special markup
language such as Wireless Markup Language
(WML). Some smart phones are equipped with
a computing platform such as the Java Virtual
Machine that can run applications written in
J2ME (Java 2 Micro Edition).
Wireless PDAs typically have larger screens
than cell phones and can directly access HTML
(hypertext markup language) pages. Some wireless PDAs can also be used to make phone calls,
and are referred to as PDA phones. Since many
people prefer to carry only one of such mobile
device around, there is a competition between
PDA phones and smart phones, a war in which
the smart phones seem to be winning.
ViewSonic (http://www.viewsonic.com) made
a super-sized PDA, called the ViewPad, that offers
a regular 800x600-pixel screen. The ViewPad
can be a very useful mobile device when regular
screen size is a necessity while light weight and
zero-boot-up time are also desired.
Blackberry devices made by Research in
Motion (http://www.rim.com) has been a big success for enterprise users as they provide a very
convenient way for reading and typing e-mails
while being away from the office.

Laptop or Tablet PCs with Wireless

Access
When size and weight are not inhibitive, mobile
users may choose to carry a laptop or tablet PC
while on the go. These mobile PCs use wireless
cards to connect to either a wireless LAN or
wireless WAN. Many such laptops now have

270

built-in wireless LAN cards, and have slots for

users to insert a wireless WAN card such as the
AirCard made by Sierra Wireless (http://www.
sierrawireless.com). An enterprise also needs to
be prepared to provide support to mobile users in
order to help them connect to the Internet through
Wi-Fi hot spots (Hamblen, 2005).
Wireless LAN is often available at a corporation campus, or at public hot spots such as many
airports and Starbucks Cafs. Wireless WAN is
available wherever cellular service is available
for the specific provider that the wireless card is
registered with.

IP Phones
IP phones are telephones that use a TCP/IP (transmission-control protocol/Internet protocol) network for transmitting voice information. Since IP
phones are attached to the data network, makers of
such devices often make the screens larger so that
the phones can also be used to access data. What
makes IP phones pervasive devices is that a user
who is away from his or her own desk can come
to any IP phone on the same corporate network,
log in to the phone, and make the phone work as
his or her own phone. The reason is for this is
that an IP phone is identified on the network by
an IP address. The mapping between a telephone
number and an IP address can be easily changed
to make the phone belong to a different user.
In terms of the information-access capability,
Cisco (http://www.cisco.com) makes IP phones
that can access information encoded in a special
XML format. Example applications on the phone
include retrieving stock quotes, flight departure
and arrival information, news, and so forth.
Pingtel (http://www.pingtel.com) developed
a phone that runs a Java Virtual Machine. This
makes the phone almost as powerful as a computer.
Mitel (http://www.mitel.com) made an IP
phone that allows a user to dock a PDA. With
this capability, users can go to any such IP phone,

Leveraging Pervasive and Ubiquitous Service Computing

dock their PDA into the phone, and immediately

have their address books on the PDA available to
the telephone. Users can also have their personal
preferences transferred from the PDA to the phone
and start to use the phone the way they prefer. In
addition, users can benefit from new applications
on the PDA, such as portable voice mail and dialing by the address book.
The Mitel IP phone seamlessly blends the wired
and wireless world for the user so that they are
no longer dealing with two separate communication tools. It also provides users with location
transparency within the network.

Orbs (Ambient Devices)

Orbs are simple devices that convey information
at a glance in a manner that is easy to observe and
comprehend (Feder, 2003). Orbs only present a
visual indication of the data, not detailed information or actual numbers. Orbs come in different
forms. One common orb is a simple globe that
changes color and intensity. Other forms include
the following:

Wall panels that adjust color or blink

Pens, watch bezels, and fobs that change
color
Water tubes that vary the bubble rate
Pinwheels that change speed

Orbs operate via wireless pager networks under the command of a server. This server gathers
pertinent information from sources, including the
Web, condenses it to a simple value, and periodically sends the information to the orbs.
Orbs are currently available from several
retailers. The wireless service costs about $5 per
month per device. Ambient Devices (http://www.
ambientdevices.com) sells orbs and provides the
communications service.
The information displayed by orbs is configurable. There are currently available data feeds for
stock-market movement and weather forecasts.

Input Technologies: Dictation, Anoto

Pen, and Projection Keyboard
Two natural ways for mobile users to input information are speech and handwriting.
Speech input can be at two levels: question-andanswer vs. dictation. Question-and-answer speech
input is useful for entering structured information where the answers can be predefined using
a grammar. Dictation technology allows users to
speak freely and tries to recognize what the user
has said. Diction technology typically requires a
training phase to tune the speech recognizer to
each particular speaker in order to achieve high
recognition accuracy. Leading dictation products
are Dragon NaturallySpeaking from ScanSoft
(http://www.scansoft.com) and ViaVoice from
IBM (http://www.ibm.com).
The Swedish company Anoto (http://www.
anoto.com) invented a technology for pen-based
input (McCarthy, 2000). It consists of a digital pen
that feels like a regular ballpoint pen, a special
paper with patterns of dots printed on it, and a
wireless technology such as Bluetooth that sends
handwritten information stored in the pen to a
computer. As the user writes, the pen not only
records what has been written, but also the order
in which the user writes it. Anoto has partnered
with companies such as Logitech (http://www.
logitech.com) and Nokia (http://www.nokia.com)
to bring this technology to end users.
For users who want to use a keyboard without
carrying one, Canesta (http://www.canesta.com)
developed the projection keyboard, in which the
image of a keyboard is projected on a surface. By
typing on the projection keyboard, information is
entered into the associated PDA device.

Application Scenarios
From an enterprises perspective, the following
applications areas are where pervasive computing
brings business value.

271

Leveraging Pervasive and Ubiquitous Service Computing

Allow employees to stay in touch with phone

calls, voice mail, e-mail, and so forth while
being away from the office.
Give employees access to information or
transactions via mobile devices while on
the road.
Provide employees with access to the corporate network from anywhere on the Internet
(i.e., remote access).
Send location-based information to employees and customers.
Monitor device status, perimeter security,
and so forth using a wireless sensor network.

Communication: Unified
Communication and Instant
Communication
With cell phones and pagers, it is not very hard
to keep mobile users in touch. But some pervasive communication technologies have reached a
higher level. Let us look at two such technologies:
unified communication and instant communication.
Unified communications refers to technologies
that allow users access to all their phone calls,
voice mails, e-mails, faxes, and instant messages
as long as they have access to either a phone or
a computer. With a computer, a software phone
allows the user to make or receive phone calls.
Voice-mail messages can be forwarded to the
e-mail box as audio files and played on the computer. Fax can be delivered to the e-mail box as
images. With a phone, a user can listen to e-mail
messages that the system would read using the
text-to-speech technology. A user can request a
fax to be forwarded to a nearby fax machine.
Unified communications services are offered by most traditional telecommunications
technology providers such as Cisco, Avaya, and
Nortel.

272

Instant communication refers to the ability of

reaching someone instantly via a wearable communication device. Vocera (http://www.vocera.
com) offers a system that uses 802.11b wireless
local area networks to allow mobile users to instantly communicate with one another. Each user
only needs to have a small wearable device to stay
connected. To reach someone, the user would only
need to speak a name, a job role, a group, or a
location to the system, and the system will take
care of the rest. By combining a small wearable
device and the speech-recognition capability,
Vocera offers a highly usable solution for mobile
communication within an organization.
The functions and features of Vocera include
the following:

Instant communication via a small wearable

device and speech commands
Hands-free communication. Except for
pressing the button to start and stop a conversation, a users hands are free during the
communication.
Flexibility in how to specify the recipients.
A user can use a name, role, group, or location to tell the system whom to contact.
The option of having a conversation or
leaving a message, for both one-to-one and
group communications
Call controls such as call transfer, blocking,
or screening
Outside calling through the private branch
exchange (PBX). The Vocera server can
be connected to the PBX to allow users of
Vocera to contact people outside the organization.

The Vocera technology has been well received

in organizations such as hospitals where users
hands are often busy when they need to communicate with others.

Leveraging Pervasive and Ubiquitous Service Computing

Mobile Access to Information and

Applications
Organizations can benefit significantly by allowing mobile access to information and applications.
Here are a few examples.

Sales-Force Automation
Salespeople are often on the road. It is important for them to have access to critical business
information anywhere at anytime. Pervasive access to information increases their productivity
by using their downtime during travel to review
information about clients and prospects, about
the new products and services they are going to
sell, or to recap what has just happened during
a sales event when everything is still fresh in
their memory. Being able to use smart phones or
wireless PDAs to conduct these activities is much
more convenient for salespeople as opposed to
having to carry a laptop PC.

Dashboard or Project-Portfolio
Management
For busy executives, it is very valuable for them to
be able to keep up to date on the dashboard while
they are away from the office and to take actions
when necessary. It is also very helpful for them
to be able to look at the portfolio of projects they
are watching, update information they have just
received during a meeting or conversation, and
take notes or actions about a specific project.

Facility Management and Other On-Site

Service Applications
Mobile access to information can significantly
boost the productivity of on-site service people
such as facility- or PC-support staff. With mobile
access, they can retrieve ticket information on the
spot, update the ticket as soon as they are done

with the work, and get the next work order without
having to come back to the office. Mobile access
also reduces the amount of bookkeeping, which
requires a lot of manual intervention, and thus
reduces the chance of human errors.

Remote Access to Corporate

Network
Allowing employees access to the corporate
network from anywhere on the Internet could
certainly bring convenience to employees and
boost productivity. There are two primary fashions
of allowing remote access.
One approach is through a technology called
virtual private network, or VPN. This typically
requires the user to carry a laptop offered by
the employer. Once the user is connected to the
Internet, a secure connection (called a VPN tunnel) is established between the laptop and the
corporate network after both user and device are
authenticated. Then the user will have access to
all the information and applications just as if the
user were in the office.
The other approach does not require the user
to carry a corporate laptop. It simply requires that
the user has access to a Web browser. In this case,
for security reasons, two-factor authentication is
often employed, in which the user not only needs
to provide a user ID and password, but also something else, such as the security code generated by
a hard token. With this approach, an enterprise
can choose which applications to make available
for remote access. Terminal service technology
offered by Citrix (http://www.citrix.com) can be
used to offer browser-based remote access to applications, both Web based and desktop based.

Location-Based Services
A special type of pervasive application is location-based service. With wireless LANs, when a
mobile user is in the vicinity of an access point,

273

Leveraging Pervasive and Ubiquitous Service Computing

location-specific information can be delivered to

the users mobile device. With wireless WANs, a
users location can be determined by the cellular
tower(s) that the users handset is communicating
with, or by the GPS (Global Positioning System)
receiver the user is using. Location-based services
include pushing information about local businesses, sending promotions to the users device
based on the users profile and preferences, and
showing meeting agendas and meeting material
if the user is on the meeting attendee list for the
room at the time.
If location needs to be accurately determined,
an ultrasonic location system called the bat system
can be used. This 3-D location system uses low
power and wireless technology that is relatively
inexpensive. An ultrasonic location system is
based on the principle of trilateration: position
finding by the measurement of distances. A short
pulse of ultrasound is emitted from a transmitter
or bat that is attached to a person or object to be
located. On the ceiling are receivers mounted at
known points. These receivers can measure the
pulse and length of travel.
An ultrasonic location system is composed of
three main components.

Bats: Small ultrasonic transmitters worn by

an individual or on an object to be located
Receivers: Ultrasonic signal detectors
mounted in the ceiling
Central controller: Coordinator of the bats
and receivers

To locate a bat, the central controller will

send the bats ID via a 433-MHz bidirectional
radio signal. The bat will detect its ID through
the embedded receiver and transmit an ultrasonic
signal containing a 48-bit code to the receiver in
the ceiling. The central controller will measure
the elapsed time that it took for the pulse to reach
the receiver. The system developed at the AT&T
Cambridge facility can provide an accuracy of 3
centimeters.

274

Overall, location-based services are still in

research mode. Once the technology becomes
mature and killer apps are identified, there
could be an explosive adoption.

User-Interaction Models
In the context of pervasive computing, it is usually
inconvenient, if not impossible, for the user to
enter text using a regular keyboard. Sometimes,
it is also inconvenient for the user to read text.
Therefore, other input and output mechanisms
have to be employed.
Nontraditional input mechanisms include
speech recognition, gesture, touch screen, eye
gazing, software keyboard, and projection keyboard. Among these, a combination of speechrecognition and pen-based touch-screen input is
most natural for most situations. This is also what
PDAs and tablet PCs typically offer.
Nontraditional output mechanisms include
converting text to speech and using sound,
blinking, and vibration to convey information
(as in ambient computing described earlier in
this chapter).
Multimodal interaction allows a user to choose
among different modes of input and output. For
mobile users, speech is typically the most convenient way for input, while visual means may still
be the most powerful way of seeing the output
(especially when the output includes pictures or
diagrams).
Kirusa (http://www.kirusa.com) has developed
technologies to support multiple levels of multimodal interaction. SMS multimodality allows users to ask a question in voice and have the answers
delivered to their mobile devices in the form of
an SMS message. Sequential multimodality allows users to use the interaction mode deemed
most appropriate for each step of the process.
Simultaneous multimodality lets users combine
different input and output modes at the same time.
For example, for driving directions, a user can

Leveraging Pervasive and Ubiquitous Service Computing

say I need directions from here to there, while

pointing to the start and end points.
Both IBM and Microsoft have developed
technologies that will support multimodal
interaction. IBMs solution is based on the
XHTML+VoiceXML (or simply X+V) specification. Microsofts solution is based on the speech
application language tags (SALT) specification,
which defines speech input and output tags that
can be inserted into traditional Web pages using
XHTML.
Besides deciding on what interaction mode
to support, much effort is needed to apply usercentered design in order to deliver a good use
experience for mobile users (Holtzblatt, 2005).

A Service-Oriented Architecture to
Support Pervasive Computing
For an enterprise to leverage pervasive computing,
instead of deploying various point solutions, the
better way is to build an architecture that is well
positioned to support pervasive devices and usage.
In order to provide mobile users with maximum
access to enterprise information and applications
with customized interaction methods and work
flow, and at the same time minimize the extra cost
in supporting pervasive access, a service-oriented
architecture should be established.
The following picture shows a service-oriented
architecture that supports pervasive computing.
Let us look at this architecture from the top to
the bottom.

Users access applications from different

devices. Some devices, such as the regular
telephone, have only the voice channel.
Some, such as the Blackberry devices, only
have the visual display. Others may have
both voice and visual channels. The size
of the visual display ranges from 1 inch
for cell phones, several inches for the PDA
and Blackberry, and 15 or more inches for
laptops.

The user devices may be on different network

connections, ranging from wireless LAN
and wireless WAN to telephone networks.
Users access applications through the applications end points. This could be a URL,
a phone number, or a start screen stored on
the end users device.
The pervasive layer sits between the application end points and the SOA layer to
provide services to specifically support the
mobile users.
The device identification engine uses a
unique ID to identify the device the user
is using. This requires the augmentation
of some of the communication protocols to include a universally unique ID
(such as the radio frequency identifier,
or RFID) of the device that is initiating
the request. With this ID, the system
can uniquely identify the device and
thus have knowledge of its capabilities,
the associated user, and so forth. The
ID information is also passed to the
security service in the SOA layer to help
decide whether the user is authorized
to access the application.
The access-control engine uses information about the device and the
communication channel it is coming
from to determine the best way to
communicate with the device: voice
only, visual only, SMS, or some type
of multimodal interaction.
Based on the desired interaction mode
with the user, the content-transformation engine either calls the appropriate
version of the application or dynamically transforms the information into
the appropriate markup language:
HTML, WML, VoiceXML, X+V, and
so forth, using the eXtensible Stylesheet
Language transformation (XSLT) technology.

275

Leveraging Pervasive and Ubiquitous Service Computing

The location-determination service

uses mechanisms built into the networks to determine the geographic
location of the user, and then decides
whether the information should be tailored based on the location and whether
additional location-based information
should be pushed to the user.
The session-persistence engine uses
the device ID and user-identity information to keep the user in the same

session while the user is roaming

from one network to another, or from
disconnected mode to connected mode
again during a short period of time.
For smart-client applications, where
data may be temporarily stored on the
device when connection is lost, the session-persistence layer would also take
care of synchronizing the data on the
device with data on the server.

Figure 2. A service-oriented architecture that supports pervasive computing

Blackberry
User

Smart
Phone
User

Regular
Phone
User

Wireless
PDA
User

WebTV
User

Laptop
User

Business Application End Points

Pervasive Layer
Device
Identifi.
Engine

Access
Control
Engine

Content
Transform.
Engine

Location
Determin.
Engine

Session
Persist.
Engine

Business Application Composition

SOA Layer
Logging
Services

Security
Services

Mgmt and
Control
Services

Other
Technical
Services

Enterprise Service Bus

Net-centric
Services

276

CICS
Services

Stored
Procedure
Services

Thirdparty
Services

Other
Business
Services

Leveraging Pervasive and Ubiquitous Service Computing

The business-application composition layer

uses information received from the pervasive
layer to determine how to integrate the business services together to best fit the need of
this mobile user.
The SOA layer provides the business services
and technical services that are integrated
together through the enterprise service bus.
The business services can be built using
Net-centric technologies such as Java or
Microsoft .NET, or they can be built based
on existing legacy business functions such
as customer information control system
(CICS) transaction and stored procedures.
They can also be based on business functions built using third-party tools or existing
in-house business modules developed in C
or C++, COBOL, and so forth.

After the establishment of such an architecture

(which can be gradually built across multiple
projects), when building an application that supports pervasive access, business services are either
reused or built. When an existing business service
is reused, the project team needs to go through the
services specification to verify that the service
will work well with the pervasive layer to support
all the pervasive channels that the application is
going to support. If not, the first thing the project
team should try to modify is the pervasive layer.
If the issue really lays in the fact that the business
service is not a pure service, that is, the service
is tied to access methods, then the service needs to
be either modified or wrapped in order to support
the new requirements. When such a modification
occurs, the service needs to be made backward
compatible, and existing applications that use the
service need to be regression tested. Eventually,
the service definition needs to be modified to
reflect the changes.
With this architecture, when a new pervasive
access channel appears, or there is a change in an

existing channel, then the only thing that needs to

be modified is the pervasive channel. All business
services can remain the same.

Future Directions
Moving forward, there needs to be much research
and development work on building a system
infrastructure that can use different sources of
information to judge where the user is, and what
devices and interaction modes are available to the
user during a pervasive session. This will enable
smarter location-based information push to better
serve the user.
A related research topic is how to smoothly
transition an interaction to a new device and interaction mode as the user changes locations and
devices. Some initial work on this subject, referred
to as seamless mobility, is being conducted at
IBM and other organizations.
Another area that deserves much attention is
the proactive delivery of information that users
will need based on their profiles and information
such as activities on their calendars or to-do lists.
This relates to previous research efforts on intelligent personal assistants with integration into the
pervasive computing environment.

References
3Gtoday. (2005). Retrieved November 5, 2005,
from http://www.3gtoday.com
Blackwell, G. (2002, January 25). Mesh networks:
Disruptive technology? Wi-Fi Planet. Retrieved
October 25, 2005, from http://www.wi-fiplanet.
com/columns/article.php/961951
Bluetooth SIG. (2005). The official Bluetooth
wireless info site. Retrieved November 3, 2005,
from http://www.bluetooth.com

277

Leveraging Pervasive and Ubiquitous Service Computing

Culler, D. E., & Hong, W. (Eds.). (2004). Wireless

sensor networks. Communications of the ACM,
47(6), 30-57.
Feder, B. (2003, June 10). Glass that glows and
gives stock information. New York Times, P.C1.
Fox, B. (2001, November 21). Mesh radio can deliver super-fast Internet for all. New Scientist. Retrieved November 15, 2005, from http://www.newscientist.com/news/news.jsp?id=ns99991593
Geier, J. (2002, April 15). Making the choice:
802.11a or 802.11g. Wi-Fi Planet. Retrieved
October 16, 2005, from http://www.wi-fiplanet.
com/tutorials/article.php/1009431
Hamblen, M. (2005). Wi-Fi fails to connect with
mobile users. ComputerWorld, 39(37), 1, 69.
Henderson, T. (2003, February 3). Vocera communication system: Boldly talking over the wireless LAN. NetworkWorld. Retrieved November
12, 2005, from http://www.networkworld.com/
reviews/2003/0203rev.html
Holtzblatt, K. (Ed.). (2005). Designing for the mobile device: Experiences, challenges, and methods.
Communications of the ACM, 48(7), 32-66.

IEEE. (2005b). The IEEE 802.16 Working Group

on Broadband Wireless Access Standards. Retrieved November 22, 2005, from http://www.
wirelessman.org
McCarthy, K. (2000, April 7). Anoto pen will
change the world. The Register. Retrieved September 14, 2005, from http://www.theregister.
co.uk/2000/04/07/anoto_pen_will_change/
Rubio, D. (2004, October 20). VoiceXML promised voice-to-Web convergence. NewsForge.
Retrieved November 23, 2005, from http://www.
newsforge.com/article.pl?sid=04/10/15/1738253
Schrick, B. (2002). Wireless broadband in a box.
IEEE Spectrum. Retrieved November 19, 2005,
from http://www.spectrum.ieee.org/WEBONLY/
publicfeature/jun02/wire.html
Schwartz, E. (2001, September 26). Free wireless networking movement gathers speed.
InfoWorld. Retrieved December 5, 2005, from
http://www.infoworld.com/articles/hn/xml/01/
09/26/010926hnfreewireless.xml
ZigBee Alliance. (2005). Retrieved November 11,
2005, from http://www.zigbee.org/en/index.asp

IEEE. (2005a). IEEE 802.15 Working Groups

for WPAN. Retrieved November 7, 2005, from
http://www.ieee802.org/15/

This work was previously published in Enterprise Service Computing: From Concept to Deployment, edited by R. Qiu, pp.
261-284, copyright 2007 by IGI Publishing, formerly known as Idea Group Publishing (an imprint of IGI Global).

278

279

Chapter 1.21

Service Provision for Pervasive

Computing Environments
Emerson Loureiro
Federal University of Campina Grande, Brazil
Frederico Bublitz
Federal University of Campina Grande, Brazil
Loreno Oliveira
Federal University of Campina Grande, Brazil
Nadia Barbosa
Federal University of Campina Grande, Brazil
Angelo Perkusich
Federal University of Campina Grande, Brazil
Hyggo Almeida
Federal University of Campina Grande, Brazil
Glauber Ferreira
Federal University of Campina Grande, Brazil

INTRODUCTION
The fast development on microelectronics has
promoted the increase on the computational power
of hardware components. On the other hand, we
are facing a significant improvement on energy

consumption as well as the reduction of the physical size of such components. These improvements
and the emergence of wireless networking technologies are enabling the development of small
and powered mobile devices. Due to this scenario,
the so-called pervasive computing paradigm, in-

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Service Provision for Pervasive Computing Environments

troduced by Mark Weiser in 1991 (Weiser, 1991)

is becoming a reality. Such a paradigm envisions
a world where environments are inhabited by
computing devices, all of them seamlessly integrated into peoples lives, and effectively helping
to carry on their daily tasks.
Among others, one major characteristic of
Weisers vision is that each device in an environment becomes a potential client or provider of
resources. Not surprisingly, pervasive computing
environments are becoming dynamic repositories
of computational resources, all of them available
to mobile users from the palm of their hands.
However, devices can unpredictably join and
leave such environments. Thus, resources can
be dynamically made available or unavailable.
Such a scenario has a great impact on the way that
resources are found and used. In the case of static
environments, such as the Web, it is reasonable to
look up and access resources, such as Web pages,
knowing the address of their providers beforehand.
On the other hand, for dynamic environments,
such as the pervasive computing ones, this is
not a reasonable approach. This is due to the fact
that one cannot guarantee that the provider of a
resource will be available at any moment, because
it may have left the environment or simply turned
off. A better approach would be to discover these
resources based on their descriptions, or any other
feature that does not require the client to know
the specific address of their providers.
To this end, some of the current pervasive computing solutions, like Wings (Loureiro, Bublitz,
Oliveira, Barbosa, Perkusich, Almeida, & Ferreira, 2006), Green (Sivaharan, Blair, & Coulson,
2005), RUNES (Costa, Coulson, Mascolo, Picco,
& Zachariadis, 2005), and Scooby (Robinson,
Wakeman, & Owen, 2004), are making use of
a novel approach from the branch of distributed
applications, the service-oriented computing
paradigm (Papazoglou, 2003; Huhns & Singh,
2005). This is due to the fact that such a paradigm
provides a crucial element for pervasive computing systems, the ability for dynamically binding

280

to remote resources (Bellur & Narenda, 2005),

which enables mobile devices to find needed
services on demand.
However, pervasive environments may be
structured in different ways. They can range
from wired networks to completely wireless ones,
where communication among the devices is performed in an ad hoc way. Such a characteristic
indicates that the way services are provisioned in
a pervasive computing environment should fit in
its organization, in order to enhance the access
to the services available.
Considering the above discussion, in this
article we provide a review on service provision
and its applicability in pervasive computing. More
precisely, we will list the existing service provision approaches and discuss the characteristics
and problems associated with each one, as well as
their usage in pervasive computing environments.
We start by providing introductory concepts of
service-oriented and pervasive computing, respectively in the service-oriented computing and
pervasive computing sections. Next, we present
the service provision techniques available and
how they can be applied for pervasive computing environments. The main current solutions
within this scope will be introduced in the service
oriented technologies section. Some of the future
trends associated with research for service provision in pervasive computing environments will
be presented in the future research trends section.
Finally, in the conclusions section we present the
conclusions of this article.

SERVICE-ORIENTED COMPUTING
The service-oriented computing (SOC) paradigm
has been considered as the next step in distributed
computing (Papazoglou, 2003). In a general way,
this paradigm can be viewed as the development
of applications through the runtime integration of
software pieces named of services (McGovern,
Tyagi, Stevens, & Mathew, 2003). In this process,

Service Provision for Pervasive Computing Environments

Figure 1. General view of a service-oriented architecture

three elements are involved: defining what is

known as a service-oriented architecture (SOA),
a service client, a service provider, and a service
registry. The former is the one who whishes to
use a service. Conversely, service providers are
those which offer services for potential clients.
Finally, the service registry is where providers
advertise or announce their services (through
service advertisements), enabling clients to dynamically discover them. By dynamic, we mean
that clients are capable of discovering services at
runtime, thus providing a high degree of flexibility
for applications. Once clients have discovered
a service, they are able to bind to it; that is, to
create a link with the service, in order to use it
(through a proxy to the real service). This process
of advertising, discovering, binding, and using a
service is illustrated in Figure 1.
In open environments, the dynamic discovery
of services implies that they can be used by heterogeneous clients. Within this scope, heterogeneity is concerned with features like the operating
system running in each client and the hardware
platform it has been built on. As a consequence
of such heterogeneity, for enabling an application
to flexibly integrate services, they should present
the following features (Papazoglou, 2003):

Loose Coupling: A service must not require

from the clients any knowledge about its
internal implementation.
Implementation Neutrality: The usage
of services must not rely on any specific
programming language, operating system,
or hardware platform.
Dynamically Discoverable: Services
should be discovered at runtime.

PERVASIVE COMPUTING
The field of pervasive computing has its origins at
the Xerox Palo Alto Research Center. The pioneer
work that has been led there has culminated in
the novel article of Mark Weiser in 1991 (Weiser,
1991), where he describes the fist ideas of pervasive
computing. Weisers vision is at the same time
revolutionary and simple: a world where computing is embedded in everyday objects, like cars,
televisions, and air conditionings, all seamlessly
integrated into our lives and performing tasks for
us (Turban, Rainer, & Potter, 2005). When Weiser
talked about seamless integration, he meant that
applications running in these objects should
act proactively on our behalf. They should, for

281

Service Provision for Pervasive Computing Environments

example, present us with relevant information,

based on what we want/need and the resources
(e.g., a printer) available in the environment we
are immersed.

SERVICE PROVISION
APPROACHES IN PERVASIVE
COMPUTING
When provisioning services in a pervasive environment, one aspect to be considered is the way it
is organized; that is, whether the environment is
based on a wired network infrastructure, whether
it is formed in an ad hoc way, or both. This is
necessary for dealing with the particularities of
each environment, and within this scope, we can
say that there are two major ways of performing
service provision (Nickull, 2005): the push-based
and the pull-based approach. In the next sections
we will outline each of these approaches as well
as describe how they fit into pervasive computing
environments.

Push-Based Service Provision

In this approach, a provider advertises its services
directly to potential clients. In other words, it sends
service advertisements to all network hosts, either
they are interested on the service or not. Such
advertisements, when received by a host, will be
kept in a local service registry. Therefore, once a
client wants to discover some service, it will then
look up in such a registry for the services that
match its needs. An example of the push-based
service provision is illustrated in Figure 2. Note
that the provider, host A, sends the advertisement of a service directly to hosts B and C, as
illustrated in Step 1. When this advertisement is
received, it will be stored on a local service registry at each host (Step 2). Then, once a host, in
our example host C, wants to discover a service,
it then inquires this registry, as illustrated in Step
3. Finally, considering that a relevant service has

282

been found, it is possible to ask its provider (host

A) for a proxy to the service (Step 4), enabling
host C to use it.
Using this approach, one major problem to
be pointed out is about the validity of the service
advertisements. It is concerned with the fact that
a service provider can leave the network, but the
advertisements associated with its services can
still be available. One approach which could be
used for solving this problem is to require the
provider to explicitly notify the network hosts
about its leaving, and consequently its services
will be no longer available. The problem is that it
is not always possible to do that. For example, if
the provider is a mobile device, it may be suddenly
run out of energy. To deal with this, providers
could be aware of the energy level of the device,
in order to notify the network hosts that within
some minutes it may not be accessible anymore.
However, other factors can be involved in the
leaving of a provider from the network. It can be
just turned off by its user, or leave the network
coverage area. Keeping track of all these factors
is a task that certainly overloads the provider. A
more reasonable solution would be requiring both
providers and clients to cooperate for renewing
the service advertisement. Therefore, as long as
the advertisements are renewed, it is possible,
but not guaranteed, that the service is available.
On the other hand, when the advertisement has
not been renewed within a time interval, then
the service is probable, but also not guaranteed,
to be unavailable, either because its provider has
left the network or because the service has been
unadvertised.
One interesting point to notice is that the
push-based service provision does not require any
physical infrastructure. This means that such an
approach is well suited in decentralized and/or
infrastructure-less pervasive environments. One
problem, in the scope of pervasive computing,
is that the advertisement task can consume a lot
of bandwidth if many devices are provisioning
services in the environment. Therefore, in envi-

Service Provision for Pervasive Computing Environments

ronments with very limited wireless links this is

certainly a major problem. On the other hand, as
services are searched locally, the discovery process does not involve costs of communication.

Pull-Based Service Provision

In the pull-based approach, in order to discover
services clients must inquiry remote registries
for the needed services. This can be performed in
two ways; either using centralized or distributed
registries.

Centralized Provision
The centralized service provision consists in scattering service registries in specific servers (i.e.,
registry servers) of the network. Therefore, for
advertising a service, the provider must initially
find which of these servers are available in the

network. After that, it has to determine in which

of them the service will be advertised (instead,
the provider could advertise the service in all the
available servers). When a client wants to discover
services, it must also find the registry servers
available in the network, and then discover the
services advertised in them. It is important to
notice that, once the registry servers are found,
unless they become unreachable, clients and
providers do not need to discover them anymore.
Jini and Web Services are some of the current
technologies that support centralized service
provision. In Figure 3 we illustrate an example of
such an approach. In such a figure, services are
advertised by hosts A and B (Step 1), in a single
registry server, host C (we are considering that
the clients have already found the registry server).
After that, each advertisement is stored in the
service registry maintained by host C (Step 2).
Also considering that host D has already found

Figure 2. Example of a push-based service provision

283

Service Provision for Pervasive Computing Environments

Figure 3. Example of a centralized service provision

the registry server, it is then able to inquiry such

server for the advertised services (Step 3). When
a relevant service is found (Step 4), host D can
interact with its provider, in this case host A, to
retrieve the proxy for the service (Step 5).
A problem with this approach is that, if all
these servers are off the network, services can
not be discovered, even when their providers
are available. In these cases a possible solution
could be the election of new servers from the
moment that is detected that the others are no
longer available. Furthermore, the centralized
service provision raises the same problem of the
pull-based one concerned with the validity of the
service advertisements. In this way, the same, or
at least similar, solutions can be applied here.

284

As the centralized service provision requires

the existence of registry servers, it is not well
suited for highly dynamic pervasive environments. In these cases, as nodes join and leave
the environment all the time, there would be too
many changes in the current registry servers,
which would in turn degrade the provisioning
of services. On the other hand, this approach
is very useful for environments equipped with
wired network. In such environments, services
can be deployed in the wired network and thus
be accessed by mobile clients through wireless
links. In environments populated with lots of
mobile clients, this is certainly a good choice,
as the bandwidth available in the wired network
could support a great number of accesses.

Service Provision for Pervasive Computing Environments

Distributed Provision
In the distributed service provision, services are
advertised in registries located in each host. Undoubtedly, in this approach the advertising task is
easier to be performed than in the other ones, as it
does not involve sending advertisements to central
servers or directly to the other hosts. However,
service discovery is more complicated, as it must
be performed by inquiring each available host
for the needed services. As no centralizer hosts
are necessary for advertising services, discovery
is possible whenever a client and a provider are
present in the network. An example of the distributed service provision approach is illustrated
in Figure 4. Initially, each host advertises its
services (Step 1). Once a client needs to perform
service discovery, in our example host A, it asks

every host in the network for the needed service

(Step 2). It is important to note the possibility of
redirecting the service discovery request to hosts
that are not reachable from the client. In Figure 4
this is performed by host B when it redirects the
request of host A to C. When a host has a service
matching the clients needs, it sends a notification
(Step 3). From this notification, the client can then
retrieve a proxy to the service (Step 4).
The major problem with this approach is associated with the protocols for performing service
discovery. As any node in the network is a potential
service provider, the discovery protocol must be
well designed, in order to cover all hosts of the
network. If any host is missing in the discovery
process, it is possible that a relevant service may
not be found. In ad hoc pervasive environments,
one possible solution to this problem is first to dis-

Figure 4. Example of a distributed service provision

285

Service Provision for Pervasive Computing Environments

cover the hosts in the vicinity, and then to retrieve

the relevant services they provide. However this
solution only performs well in small networks,
where the available hosts can be discovered from
any other one. Another point to be considered is
about the network congestion that such service
discovery protocols can cause. As the search
should include all hosts in the network, the protocol
must apply techniques for avoiding flooding the
network. Obviously, this is not a serious problem
in wired networks, but considering the wireless
ones, it must be strictly taken into account. A
good usage scenario of the distributed provision
approach is a decentralized pervasive environment where the edge of the network is formed by
mobile clients and its core is populated by service
providers connected through a wired network.

SERVICE ORIENTED
TECHNOLOGIES
In this section we present the main technologies
related to the provision of services in pervasive
computing environments.

Universal Plug and Play (UPnP)

The Universal Plug and Play (Richard, 2000)
is an open architecture, which uses standard
Internet protocols for pervasive peer-to-peer
network connectivity (http://www.upnp.org). The
UPnP protocol defines a set of steps, addressing,
discovery, description, control, eventing, and
presenting, which enables the automatic network
configuration and service provision. Through the
addressing, devices get their network address,
which is performed by a dynamic host configuration protocol (DHCP). The discovery step
consists of notifying the other hosts, through the
push-based approach, about the services, and embedded devices, that a joining host provides. The
discovery can also be performed in a distributed
pull-based fashion. The next step, description,

286

is about the description of a device, stored as an

XML document. In such a document, it is kept,
among other information, the description of the
services that the device provides. Therefore, by
discovering a device, it is possible to retrieve the
services it provides, and then, through the control
step, invoke their actions. Changes in a service
can be notified to interested devices through the
eventing step. Finally, through the presenting step,
it is possible to load a specific URL, specified in
the device description, in order to display a user
interface for controlling a device.

Jini
Jini is a service-oriented Java technology based on
a centralized pull-based approach (Waldo, 1999).
Therefore, service advertisements are stored in
central servers, which are named lookup servers.
Jini uses the RMI (http://java.sun.com/products/
jdk/rmi - Remote Method Invocation) protocol
for all interactions involved in the advertisement,
discovery, and invocation of services. When a
client discovers and binds to a service, it is incorporated to the client by downloading the code
of a proxy to the required service, named remote
control object.
The Jini platform uses the concept of lease
for controlling the access to the services. A lease
is a sort of warrant that a client has for using a
service during a specific period of time. When
the lease expires the client needs to renew it
with the provider if it wishes to continue using
the service.

Bluetooth
Bluetooth is a standard for wireless communication among small devices within short distances
(Johansson, Kazantzidis, Kapoor, & Gerla, 2001),
defining higher-level protocols for both host and
service discovery (http://www.bluetooth.org). The
discovery of services in the Bluetooth standard is
defined by the service discovery protocol (SDP),

Service Provision for Pervasive Computing Environments

Figure 5. Service discovery in Bluetooth

which enables to enumerate the devices in the

vicinity and retrieve the services they provide.
Bluetooth uses a distributed pull-based approach for service advertising and discovery. To
this end, each device maintains its own service
discovery database (SDDB), which is a registry
where its services are advertised. Therefore, a
Bluetooth device performs service discovery by
querying the SDDBs of the devices around. These
advertising and discovery processes are illustrated
in Figure 5. Notice that, initially, all devices advertise their services on their respective SDDBs
(1). Next, a client searches for all the Bluetooth
devices on the range of its wireless interface (2).
For each device found, the client sends a query
about the availability of an interested service (3).
The devices answer these queries by informing
whether they offer the needed service or not (4).
Once localized a device providing the desired
service, the client can connect directly to such
device and finally use the service (5).

FUTURE RESEARCH TRENDS

Although the first service-oriented pervasive computing solutions have been developed, much work
has to be done yet. For example, the matching of
the users needs and the services functionalities
should be enhanced to improve the interaction
of the user with the pervasive application. Still,
problems remain in the context of service continuity. Solutions to this problem would enable the
user to use a service continuously, as he/she walks
through different environments. This is important
because, sometimes, the service a client was using is not available in the new environment, and
thus, some mechanism should allow it to use a
similar one without, or at least with a minimum,
of interruption.
These, and possibly other problems related to
the provision of services in pervasive environments, must certainly be completely solved so
that we can enjoy the full potential of merging
service-oriented and pervasive computing.

287

Service Provision for Pervasive Computing Environments

CONCLUSION
The service-oriented paradigm has proved to be
an important element in pervasive computing
systems, in order to provide anytime and anywhere
access to services. Its dynamic binding feature
enables to build applications powered with on-demand extensibility and adaptability, two important
elements of any pervasive system.
Given this trend, in this chapter we have
tried to present an overview of service provision
in pervasive computing environments. More
precisely, we have showed an introduction to the
main characteristics, challenges, and solutions
concerning the way that services are advertised,
discovered, and used in pervasive environments.
Although we presented concepts at an introductory level, we believe they may serve as a good
source of knowledge, helping both students and
researchers involved with these fields.

REFERENCES
Bellur, U., & Narendra, N. C. (2005).
Towards
service orientation in pervasive computing systems. In International Conference on Information
Technology: Coding and Computing (Vol. II, pp.
289-295).
Las Vegas, NV.
Costa, P., Coulson, G., Mascolo, C., Picco, G. P.,
& Zachariadis, S. (2005).
The RUNES
middleware: A reconfigurable component-based
approach to networked embedded systems. In
Proceedings of the 16th IEEE International Symposium on Personal Indoor and Mobile Radio
Communications. Berlin, Germany: IEEE Communications Society.
Huhns, M. N., & Singh, M. P. (2005). Service
oriented computing: Key concepts and principles.
IEEE Internet Computing, 9(1), 75-81.
Johansson, P., Kazantzidis, M., Kapoor, R., &
Gerla, M. (2001). Bluetooth: An enabler for

288

personal area networking. IEEE Network, 15(5),

28-37.
Loureiro, E., Bublitz, F., Oliveira, L., Perkusich,
A., Almeida, H., & Ferreira, G. (2006). A flexible
middleware for service provision over heterogeneous pervasive networks. In Proceedings of
the 4th International Workshop on Middleware
for Mobile and Distributed Computing. Niagara
Falls, NY: IEEE Computer Society.
McGovern, J., Tyagi, S., Stevens, M., & Mathew,
S. (2003).
Service oriented architecture. In J.
McGovern, S. Tyagi, M. Stevens, & S. Mathew
(Eds.), Java Web services architecture (pp. 35-63).
San Francisco: Morgan Kaufmann.
Nickull, D. (2005). Service oriented architecture
(White Paper). San Jose, CA, USA: Adobe
Systems Incorporated. Retrieved March 28, 2006,
from: http://www.adobe.com/enterprise/pdfs/Services_Oriented_Architecture_from_Adobe.pdf
Papazoglou, M. P. (2003). Service-oriented computing: Concepts, characteristics, and directions.
In Proceedings of the Fourth International Conference on Web Information Systems Engineering
(pp. 3-12). Rome: IEEE Computer Society
Robinson, J., Wakeman, I., & Owen, T. (2004).
Scooby: Middleware for service composition in
pervasive computing. In Proceedings of the 2nd
Workshop on Middleware for Pervasive and Ad
hoc Computing. Toronto, Canada.
Richard, G.G. (2000). Service advertisement and
discovery: Enabling universal devicecooperation.
IEEE Internet Computing, 4(5), 18-26.
Satyanarayanan, M. (2001). Pervasive computing:
Vision and challenges. IEEE Personal Communication, 8(4), 10-17.
Sivaharan, T., Blair, G., & Coulson, G. (2005,
October). GREEN: A configurable and reconfigurable publish-subscribe middleware for pervasive
computing. In Proceedings of the International

Service Provision for Pervasive Computing Environments

Symposium on Distributed Objects and Applications (Vol. 3760, pp. 732-749). Agia Napa, Cyprus:
Springer Verlag.
Turban, E., Rainer, R. K., & Potter, R. (2005).
Mobile, wireless, and pervasive computing. Information Technology for Management: Transforming Organizations in the Digital Economy (pp.
167-206). New York: John Wiley & Sons.
Waldo, J. (1999). The Jini architecture for network-centric computing. Communications of the
ACM, 42(7), 76-82.
Weiser, M. (1991). The computer for the 21st century. Scientific American, 265(3), 94-104.

KEY TERMS

Service: A software entity that can be integrated to a remote distributed application.

Service-Oriented Computing: The newest paradigm for distributed computing, where
applications should be built by dynamically
integrating services.
Service Advertisement: The element used for
publishing and discovering a service.
Service Client: The one wishing to use a
service.
Service Provider: The one that offers services.
Service Registry: The place where services
are published.

Pervasive Computing: The vision conceived

by Mark Weiser which consists of world where
computing will be embedded in every day objects.
This work was previously published in the Encyclopedia of Mobile Computing and Commerce, edited by D. Taniar, pp. 877-884,
copyright 2007 by Information Science Reference, formerly known as Idea Group Reference (an imprint of IGI Global).

289

290

Chapter 1.22

Preparing Participants for

Computer Mediated
Communication
Robert Jordan
U.S. Bureau of Labor Statistics, USA

Abstract
Computer mediated communication (CMC)
provides a way of incorporating participant interaction into online environments. Use of such
features as discussion forums and chats enhance
collaborative work and learning. For many,
however, CMC may be an unfamiliar medium.
To ensure a successful CMC event, it is essential
to adequately prepare participants for CMC. A
proposed four step model prepares participants for
CMC. The four steps include conducting a needs
and population analysis, providing an orientation
before the event and shortly after the event begins,
and providing continuing support.

Introduction
Computer mediated communication (CMC) provides interaction during online events, whether
synchronous or asynchronous. The modera-

tor-to-participant and participant-to-participant

interaction present in CMC offers participants an
environment that closely replicates face-to-face
interaction with which they may be more comfortable. The advantages of CMC are diminished
if participants are not adequately prepared for its
deployment and use. In addition to examining
why participants may be uncomfortable with
CMC-based interaction, this paper proposes
several strategies for effectively preparing participants for CMC. These strategies are based on
experiences delivering online training across the
country to an audience that was unfamiliar with
CMC-based interaction.

Background
Moore and Kearsley (2005) cite one common
misconception that participants new to online
learning hold is that learning online is less demanding than face-to-face learning. In fact, the

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Preparing Participants for Computer Mediated Communication

opposite is generally true. Effective participation

in CMC environments requires participants to
be highly motivated and less passive than what
may be possible in a face-to-face environment.
They must be prepared for what they will experience in CMC. If not addressed at the outset, this
misconception may result in participants falling
behind and never becoming comfortable and
engaged in CMC.
Sherry (2000) cites Fishman in highlighting
several potential constraints on CMC-based learning. These constraints include the participants
lack of experience with using computers, cultural
influences, and apprehension about electronic
communication (Sherry, 2000). Hiltz and Wellman (1997) describe a sense of pressure on new
CMC participants and an uncomfortable feeling
of being lost that many of them experience after
they first venture into a CMC environment. In
addition, new participants are often apprehensive
about using technology to communicate and may
be more guarded in what they post in discussion
forums or chat rooms when compared to what
they might otherwise communicate in a face-toface environment (Puetz, 2000; Sherry, 2000).
They may need time to find their online voice
(Sherry, 2000). Finding ones own voice, knowing
what to say and when to interject it in an online
environment is critical in CMC and the earlier
that participants are comfortable with this, the
more successful the CMC event will likely be.
As features such as instant text messaging and
chat are more widely incorporated in everyday
life, participants should be more comfortable
with CMC.
Many universities incorporate instant messaging in their online education. One such example
was the use of MSN Messenger in a mathematics
course at the University of Southern Queensland
(Loch & McDonald, 2007). Students in the mathematics course were comfortable using many of
the features of the instant messaging application,
including the electronic ink feature (Loch &

McDonald, 2007). Increased organizational use

of instant messaging and its improving functionality will likely result in its increased use in
CMC events.
While a lack of comfort in communicating
online is the primary challenge for CMC, surveys have indicated several other factors that
may result in discomfort with CMC (Lewis
& Romiszowski, 1996). Some participants are
uncomfortable with the nature of asynchronous
communication and may be frustrated with the
time lags between postings and responses in online discussion forums (Lewis & Romiszowski,
1996; Sherry, 2000). For some participants, the
delay in back-and-forth communication lacks
the spontaneity and directness of face-to-face
communication. Some participants are reluctant
to express candid personal views in discussion
forums or chats, since they fear that their postings can easily be printed out or saved (Lewis &
Romiszowski, 1996). Apprehension about leaving
a paper trail is understandable. What might be
said and forgotten in a face-to-face environment
may be misinterpreted and taken out of context
in a written form. Participants may become
frustrated when others post comments in discussion forums and in chat sessions that reflect their
own ideas before they get a chance to comment
(Lewis & Romiszowski, 1996). In a face-to-face
environment, participants may be able to better
gauge when they should interject comments.
Those who post late in discussion environments
are at a disadvantage in expressing what might
appear to others to be original ideas. This situation
occurs in face-to-face environments as well, as
anyone who is familiar with the overeager hand
raiser can attest. It might seem to be more clearly
evident in CMC due to the text-based method of
communicating. The give-and-take of face-to-face
verbal interplay is replaced with a written record
of who stated what and when.
Technology continues to evolve and may mitigate some of these challenges. For instance, some

291

Preparing Participants for Computer Mediated Communication

discussion board applications now allow users to

easily post audio or even video postings and responses, thus reducing the emphasis on text-based
communication. In addition, most CMC tools
allow participants to subscribe to discussion
board forums so they can receive instant updates
of activity on the boards either via e-mail or really
simple syndication (RSS) site feeds.
Another concern expressed about CMC is that
online discussion board or chat postings, made
without the benefit of non-verbal visual cues, may
be misinterpreted and more contentious than what
might otherwise occur in a face-to-face environment. There are social communication mores that
are evolving in online environments, such as the
wider usage of emoticons, which are alleviating
this concern. Many available Web conferencing
and other CMC tools now formally incorporate
the use of emoticons as non-verbal cues in communication.
Moore and Kearsley (2005) cite four reasons
for participant anxiety in online environments.
1.
2.
3.

4.

Participants are apprehensive about CMC,

particularly those new to it
Some participants express concern about the
quality of online events, such as courses
Access to technology-based resources necessary for participating may be a challenge
for some participants
Some participants worry about the personal
resources and expenditures that are required
to fully participate in CMC

Technological challenges remain. Participants

who can not easily access CMC applications are
at an immediate disadvantage from which they
may never fully recover. For example, some tools
require downloadable plug-ins and players, which
may pose security problems if the participant is
using a computer supplied by their workplace or
public library. Bandwidth availability must be
considered as well if the tools used are rich in

292

multimedia. As use of Web-based applications

expands, these challenges should diminish. For
the present, however, developers of CMC events
are wise to assess these potential challenges.

Strategies for Increasing

Participant Comfort with
CMC
Lorraine Sherry (2000) suggests five keys to
adoption of CMC by participants:
1.
2.
3.
4.
5.

They must believe there is a clear benefit to

engaging in CMC
They must overcome any fear they have of
technology
They must feel social and cultural compatibility in the CMC environment
They must feel comfortable with the CMC
events scaffolding
They must find their online voice

Sherry (2000) cites Engestrom in highlighting

the importance of cultural effects. Ultimately,
CMC is a social culture that includes those who
participate in it, the online tools and conventions
that foster it, and the tangible and intangible
results that accrue from it (Sherry, 2000). For
CMC to be successful, participants must believe
that they have a stake in the process and feel a
connection with fellow participants as could exist
in a face-to-face interaction. As online interaction
expands, sustaining online communities should
become easier, but resistance remains strong
among those who view online communities as
remote or impersonal.
There are several strategies for preparing
participants for online participation. Sherry
(2000) cites Wilson and Ryder who suggest
several strategies. These include demonstrating
the technology prior to a CMC event, providing
online consultation with participants, employ-

Preparing Participants for Computer Mediated Communication

ing moderated discussion forums, establishing a

model for netiquette, and providing desk aids
and tip sheets for participants (Chou, 2001; Sherry,
2000). Such tactics reassure CMC participants
and reinforce the idea the idea that they are part
of a larger community.
Gilly Salmon (2003) recommends an online
orientation session highlighting tips for communicating before online courses using CMC.
Orientation might take the form of a face-to-face
demonstration of how CMC works or in a Web
conference, in which the system and tools to be
used in the CMC event can be demonstrated and
remote control passed to participants so they can
gain comfort in using the tools. In addition, moderators might decide to incorporate features such
as online seminars, virtual office hours, and participant journals as ways to make participants more
comfortable with CMC (Chou, 2001). Including
a chat session creates an interactive environment
that more closely resembles face-to-face interaction. Emerging Web conferencing tools provide a
convenient and effective way to build participant
confidence in CMC technology before they enter a
CMC event. Many of the Web conferencing tools
currently available provide polling tools that allow
moderators to assess the feelings of participants
toward CMC before the event occurs and allow
the moderator to make adjustments if needed.
Web conferencing applications, such as WebEx,
Adobe Connect, Live Meeting, or Centra among
others, are generally more effective than the chat
tools currently often integrated in asynchronous
online environments.
Developers of CMC-based events should integrate a variety of learning styles into their events
by providing for a variety of participant activities
(Salmon, 2003). The advantage to incorporating
various learning styles is that it promotes the
concept that there is something to accommodate
all participants in the CMC environment. An
auditory learner, for example, might appreciate
an audio or video clip introducing the event. A

visual learner might appreciate well-designed

tables or simulations and screen captures with
voice-over narration. Tactile learners might find
online collaborative exercises and projects engaging. In any CMC event, moderators should strive
to keep the entire audience interested in not only
the communication but also in the content on the
site. An interested participant is more inclined to
contribute.
CMC moderators should engage new participants quickly to make them comfortable with
communicating online. This particularly is true for
inexperienced participants, especially those who
may be skeptical or fearful about participating in
CMC. The optimal solution is that participants are
comfortable with logging into an event. Preparation should begin well before the event and continue as the event unfolds. The next section will
describe a specific plan for preparing participants
for CMC, with an emphasis on preparing the most
inexperienced participants and increasing their
comfort with what will be expected of them in a
CMC environment.

A Model for Preparing

Participants for CMC
A four-step model assists in preparing participants
for CMC events. The four steps are as follows:
1.
2.
3.
4.

Analysis of needs and population

Participant CMC orientation (before
event)
Online introduction to CMC
Continuing support

Analysis of Needs and Population

As in many development processes, the analysis
phase is essential. Developers must analyze the
needs of the specific population that will be
served to determine the appropriate levels of the

293

Preparing Participants for Computer Mediated Communication

orientation and continuing support necessary to

ensure success of the CMC event. Performing a
thorough analysis in advance of the event will
provide information to the developers that they
can use in building their event. The model for
providing analysis discussed here is based on two
models that are often incorporated in traditional
instructional system design models:
1.
2.

Needs assessment
Population analysis

Needs assessment reveals the attitudes of the

targeted audience as well as their needs (Hodell,
2000). For online training using CMC, this step
would likely be undertaken as a part of the instructional systems design process, but is should
be completed for any significant CMC event.
While the focus of the needs analysis will often
be focused more on the subject matter content
that will be discussed during the event, the information is of value in assessing how CMC can
be effectively used to deliver the subject matter.
For instance, if the needs assessment reveals a
desire for participant and moderator interaction
revolving around specific topic areas, the event
can be designed with discussion boards or synchronous chat sessions. What developers should
be looking to obtain from the needs analysis is
how CMC will be employed during the event to
meet the interests of the audience and build upon
their needs.
Population analysis assesses the targeted audience (Hodell, 2000; Puetz, 2000). This is important
to the CMC developer and moderator in that it will
determine the direction of the remaining steps in
the four-step model highlighted above. Questions
in the population analysis survey should be designed to assess the comfort level of the targeted
audience with the methods commonly used in
CMC (Puetz, 2000). Hodell (2000) recommends
building a population profile of the targeted audience from the results of the survey. Surveys might
include the following questions:

294

1.
2.
3.

4.

5.

How much time is spent online during an

average work week?
How much time is spent online outside of
work during an average week?
Are CMC features such as instant messaging,
chat rooms, and discussion boards commonly used in the audiences workplace?
How familiar and comfortable is the targeted audience with these features and do
they use them frequently at home and in the
workplace?
What is the audiences perception of online
environments versus face-to-face environments?

Taken together, the needs assessment and

population analysis provide an important first step
in the process of preparing participants for CMC.
Without completing the analysis step, a CMC event
developer will be at a disadvantage.

Participant CMC Orientation

(Before Event)
The level of participant orientation necessary
before a CMC event should be determined once
the results of the assessment in step one are complete. Orientation should begin one to two weeks
before the event. The objective of the orientation
is to provide participants will a clear idea of what
to expect when they first logon to the event and
familiarize them with the features that they will
be expected to use to communicate with fellow
participants.
Two orientation methods are suggested:
1.
2.

Asynchronous and synchronous online

orientation
Desk aids

Online orientation may be either asynchronous or synchronous. Asynchronous orientation

provides participants with a virtual tour of the
features of the technology that will be employed

Preparing Participants for Computer Mediated Communication

during the event. The CMC technology that will

be used in the event may also have a tutorial that
can be provided to participants. There are many
screen capture applications available in which
screen actions are recorded with or without voiceover narration. Recorded actions should include
a demonstration of the features that will be used
during the event.
Synchronous orientations sessions provide the
additional benefit of participant interaction. Commercial Web conferencing tools allow moderators
to demonstrate the features of the system being
used to participants. Web conferencing also allows
for participant questions and reactions which may
be useful in refining the event before it occurs.
Desk aids that participants can refer to throughout the course are also invaluable. Participants can
refer to the desk aid until they become familiar with
the features of the CMC system that will be used
in the event. The desk aid might incorporate screen
captures that show participants how the screens
look when they log into the CMC event. Other
functions such as logging in, posting in discussion
forums, and participating in chat sessions can be
included in the desk aid. Policies and restrictions
may be included in the desk aid as well. The desk
aid may be distributed in electronic format so
participants can print it out as needed.

Online Introduction to CMC

Participant preparation models exist, including
a five-step model for online learning developed
by Gilly Salmon (2003). Salmons model is applied as the CMC event begins. The first two
steps of Salmons five-step model directly relate
to initial orientation. These two steps are access
and motivation and online socialization (Salmon,
2003). In the first week, moderators must ensure
that participants access the system successfully.
If there are challenges initially in accessing the
event, participants may drop out of the event or
never become fully engaged with the event. This

is particularly true for those participants who are

not comfortable with technology or the demands
that are made of CMC participants. To ensure full
participant access, moderators should monitor
participation and contact participants who do not
log in shortly after the event begins. Moderators
should also provide an e-mail address or a telephone number so participants can contact them if
they experience difficulties accessing the event.
Online socialization early in the event is
desirable. It is suggested that a discussion forum
be established for participant introductions. The
moderator should encourage participants to post
early during the event. Participants should provide information on themselves, their reasons
for participating in the event, and perhaps post a
photograph. Building online socialization early in
the event will promote communication throughout
the event.
Another suggestion for early in the event is to
provide participants with information on proper
techniques for communicating in an online environment. Articles on netiquette may be attached
or linked to from within the CMC Web site for
participant access.

Continuing Support
Participants should be supported throughout the
course of the event. It is suggested that a discussion
forum be established for questions that the participants might have. Moderators should be vigilant
in monitoring participation and be prepared to
intervene if participants fail to contribute. It is
possible that a lack of participation may indicate
that participants are experiencing technical or
other difficulties.

Case Study
The techniques described above were used in
delivering an online training based on CMC

295

Preparing Participants for Computer Mediated Communication

that was delivered over a four week period. The

application used for this training was an open
source courseware management system called
the modular object oriented developmental learning environment, or Moodle. Moodle provides
functions such as discussion forums and chat
rooms commonly used in CMC. Well before the
development team selected Moodle as the tool to
deliver the training, work had begun to prepare
participants for CMC.

Analysis of Needs and Population

Analysis of the population of potential learners
and their needs began before the training was set
to begin. The population was comprised of approximately 170 participants from nearly every
U.S. state, the District of Columbia, the Virgin
Islands, and Puerto Rico. A survey was developed to assess the readiness of this population
for CMC and their needs, with a focus on what
features of CMC they wanted to see included in
the online training.
The survey showed the training developers that
the majority of survey respondents frequently used
the Internet in their work and were comfortable
with computer-based technology. In addition, a
number of the respondents had participated in
online learning events with CMC features. The
respondents of the survey indicated a definite
interest interaction with online moderators as
well as with fellow learners.
The conclusion from this front-end analysis
was that the population was comfortable with
using the Internet and had some experience with
online learning and CMC, but by no means had
mastered the functions of CMC. The audience
did express interest in CMC features that would
enable interaction with the course moderators
and fellow participants. These results assisted
the online training developers in approaching
the orientation.

Participant CMC Orientation (Before

Event)
A week before the event began, participants were
provided with a training plan and an orientation
desk aid. The desk aid covered navigating to the
training Web site, establishing a Moodle account,
and signing into a specific section of the training.
Also included was information on participating
in discussion forums and chat sessions. Screen
captures from the Moodle site with detailed
explanations were included in the desk aid. Web
conferencing might have been used as well, if one
had been available to the course developers.

Online Introduction to CMC

In the first weeks activity, participants were
provided a discussion forum entitled Ask the
Moderators, where they were encouraged to
ask questions related to the course. Moodle
incorporates an internal messaging system in
which participants sent direct messages to the
event moderators. The moderators work e-mail
addresses were also provided. Finally, a link to
a Wikipedia entry on netiquette was posted in
the first weeks activities.
As a result of the preparation the great majority
of participants experienced no difficulty logging
into the Moodle Web site. In the few instances
where participants experienced difficulties logging in, the difficulties were resolved quickly, in
part because they knew how to quickly contact
the moderators. Participants indicated in the
evaluation sheets that the desk aid helped them
to log in and know what to do once they were in
the Web site. They also appreciated knowing the
moderators could be contacted easily from within
the Web site.

Continuing Support
The Ask the Moderators discussion forum was
included in all four weeks of activities. The mod-

296

Preparing Participants for Computer Mediated Communication

erators made use of a Moodle feature referred to as

a new forum discussion board to post reminders
and tips for enhancing the experience.

Future Trends
As more organizations look to contain costs,
it is expected that even more of them will turn
to some form of CMC. As CMC draws more
participants, there is potential for problems if
these new users are not fully prepared for what
is required in a CMC environment. With more
people using Internet technology, including CMC,
it can be expected that less intensive preparation
may be required in the future. The nature of the
preparation may also change with the evolution
of technology and in the way CMC is delivered.
For instance, audio-based discussion boards now
exist allowing participants to record their postings
and responses to posts, rather than responding
by typing text.
Another future trend is the emergence of more
sophisticated modes of online communication.
One popular social online community is called
Second Life, a virtual reality gaming environment
that allows participants to create alternate identities and lifestyles through avatars. Community
members can buy and sell virtual products, services, and even real estate using virtual dollars
that can be exchanged for real money (Newitz,
2006). The success of these virtual communities
serves as a model for creating and sustaining an
online community. Companies such as Wal-Mart,
American Express, and Intel have been studying
interaction in Second Life in conjunction with
their corporate training objectives (Newitz, 2006).
Ohio University has been incorporating Second
Life into its formal online learning programs (Dewitt, 2007). Virtual communities such as Second
Life provide a more life-like form of interaction
than what is generally seen today in CMC.
The closer developers can get to replicating
face-to-face interaction, the more likely that CMC

will be accepted by a broader audience. Virtual

reality applications, such as those provided in
Second Life, provide a glimpse into the future
of CMC can become.

Conclusion
Organizations are turning to CMC as a way of
meeting, collaborating, and learning. More people
are being introduced to CMC, including users who
are unfamiliar with the style of communication or
the technology involved. Participant acceptance
of CMC will grow if they are prepared for what
is encountered in a CMC environment.
A four-step process may be employed to prepare participants for CMC. These steps include
conducting a needs and population analysis on
the target audience, providing a participant orientation before the CMC event begins and during
the first week, and providing continuing support throughout the event. A systematic process
ensures that CMC developers and moderators
will understand their audiences strengths and
weaknesses and will enable them to develop an
effective deployment strategy. Communication
techniques found in CMC offer participants an
environment that more closely replicates faceto-face interaction with which they may be most
comfortable. Preparing participants for CMC is
critical in making participants comfortable with
communicating online.

References
Chou, C. (2001, Summer-Fall). Formative evaluation of synchronous CMC systems for a learnercentered online course. Journal of Interactive
Learning Research. Retrieved July, 31 2006, from
Infotrac College Edition.
Dewitt, D. (2007, January 29). Virtual-reality
software creates parallel campus, enhances educa-

297

Preparing Participants for Computer Mediated Communication

tion. The Athens News. Retrieved March 7, 2007

from http://www.athensnews.com/issue/article.
php3?story_id27194
Hiltz, S., & Wellman, B. (1997, September).
Asynchronous learning networks as a virtual
classroom. Communications of the ACM, 40(9).
Retrieved July, 31 2006, from Infotrac College
Edition.
Hodell, C. (2000). ISD from the ground up. United
States: ASTD.
Lewis, J., & Romiszowski, A. (1996, November).
Networking and the learning organization: Networking issues and scenarios for the 21st century.
Journal of Instructional Science and Technology,
1(4). Retrieved October 13, 2004, from http://www.
usq.edu.au/electpub/e-jist/docs/old/vol1no4/article1.htm
Loch, B. & McDonald, C. (2007). Synchronous
chat and electronic ink for distance support in
mathematics. Innovate 3(3). Retrieved March 7,
2007, from http://www.innovateonline.info/index.
php?view=article&id=374
Moore, M., & Kearsley, G. (2005). Distance
education: A systems view (2nd ed.). Belmont:
Thomson Wadsworth.
Newitz, A. (2006, September). Your second
life is ready. Popular Science. Retrieved November 13, 2006, from http://www.popsci.
com/popsci/technology7ba1af8f3812d010vgnvcm1000004eecbccdrcrd.html
Puetz, L. (2000, March). Twelve learning interventions that combat technophobia. Learning
Circuits. Retrieved November 24, 2006, from
http://www.learningcircuits.org/2000/mar2000/
mar2000_elearn.html

Sherry, L. (2000). The nature and purpose of

online conversations: A brief synthesis of current
research. International Journal of Educational
Telecommunications, 6(1), 19-52. Retrieved June
3, 2004, from http://www.rmcdenver.com/TEN/
dialogue.htm

Key Terms
Asynchronous: Online communication that
does occur at the same time.
Avatar: A computer-generated representation
of an online participant commonly found in virtual
reality environments.
Computer Mediated Communication
(CMC): Design that incorporates synchronous
and asynchronous online communication that
promotes interaction among participants. Such
features include discussion forums and online
chat features.
Needs Assessment: An analysis of the needs
and desires of a target audience. Methods of
conducting a needs assessment include survey
questionnaires, focus groups, and personal interviews.
Open Source: Programs that are developed
with an open license and distributed without
cost. Upgrades to programs and support are often
provided by the user community.
Population Analysis: An analysis of a targeted
audience focusing on its attributes, abilities, and
feelings.
Synchronous: Online communication that
occurs among participants simultaneously.

Salmon, G. (2003). E-moderating: The key to

teaching & learning online (2nd ed.). London:
RoutledgeFalmer.
This work was previously published in the Handbook of Research on Computer Mediated Communication, edited by S. Kelsey
and K. St.Amant, pp. 25-33, copyright 2008 by Information Science Reference, formerly known as Idea Group Reference (an
imprint of IGI Global).
298

299

Chapter 1.23

Computer-Mediated
Communication Research
J. D. Wallace
Lubbock Christian University, USA

Abstract
This chapter asks What is meant by computermediated communication research? Numerous
databases were examined concerning business,
education, psychology, sociology, and social sciences from 1966 through 2005. A survey of the literature produced close to two thousand scholarly
journal articles, and bibliometric techniques were
used to establish core areas. Specifically, journals,
authors, and concepts were identified. Then, more
prevalent features within the dataset were targeted,
and a fine-grained analysis was conducted on
research-affiliated terms and concepts clustering
around those terms. What was found was an area
of scholarly communication, heavily popularized
in education-related journals. Likewise, topics
under investigation tended to be education and
Internet affiliated. The distribution of first authors
was overwhelming populated by one time authorship. The most prominent research methodology
emerging was case studies. Other specific research
methodologies tended to be textually related, such
as content and discourse analysis. This study was

significant for two reasons. First, it documented

CMCs literature historical emergence through a
longitudinal analysis. Second, it identified descriptive boundaries concerning authors, journals, and
concepts that were prevalent in the literature.

Introduction
Computer-mediated communication (CMC) involves a wide number of characteristics involving
human communication. It also includes systems,
methods, and techniques that are typical of online
environments. Therefore, one would rightfully
expect definitional difficulties both technological
and methodological. Wallace (1999) extensively
surveyed the literature concerning CMC and
found relatively few definitions. While differences abounded in the definitions found, the one
constant was the use of the computer as an intermediary device. The centrality of the computer
and communication layers human characteristics
and technological issues.

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Computer-Mediated Communication Research

Levinson (1990) suggests that in order to understand a device or a technique, not only should
we take a microscopic view through research
and examination, but we should also take a more
macroscopic view. A survey of the scholarly
communication might help provide a different
perspective. Hopefully, it would reveal some of
the larger areas of inquiry concerning online
research in general, and computer-mediated communication specifically. This macroscopic survey
would enable researchers and scholars to more
efficiently coordinate their own activities with
outlets and concepts that have the most pressing
need for their contributions. It has the additional
benefit of documenting CMCs developmental
features that can be compared with future research
or differing methodologies.
Similar to other such studies, the purpose of
this chapter is provide an overview of the CMC
scholarly literature, and to identify its component features in providing a tangible means of
identification (Dick & Blazek, 1995, p. 291).
Likewise, it is not to determine the magnitude
that CMC occupies as a discipline, field, specialty,
or subspecialty area. For purposes of literary
description, the term field is not a cataloguing
designate, but rather a convenient moniker under
which CMC scholarship resides. CMC is often
described in the literature as a field. However,
designates of specialty, or subfield are probably
more accurate.
Simply put, the statement of the problem is:
what are trends in computer-mediated communication research? Definitions and descriptions
of current literature on the subject reflect views
that are selective and often disparate. Rather than
revisit debatable definitional issues, an arguably
more objective approach will be used as the focus
of this inquiry. Specifically, what authors, journals,
concepts, and research issues possibly populate
the CMC domain?
Certainly, a number of conceptual problems
would be introduced with any kind of predictive
examination (Hargittai, 2004). Therefore, ex-

300

ploratory and descriptive procedures seem more

appropriate than postulating hypotheses. With
this in mind, the original question concerning
CMC has, as one possible answer, a bibliometric
analysis into the nature of the field. Bibliometrics
is the . . . mathematical and statistical analysis
of patterns that arise in the publication and use
of documents (Diodato, 1994, p. ix).
Library and information science have long
used bibliometrics for this kind of analysis. They
have a body of literature supporting their validity
and reliability. Moreover, bibliometric procedures
provide the means upon which relationships of
theoretical inquiry can be based. Borgman &
Rice (1992) state that: Bibliometric data are particularly useful for studying longitudinal trends
in scholarly disciplines because of the massive
datasets that can be utilized. Virtually no other
method provides as comprehensive coverage of a
topic in scholarly communication (p. 400).
Journal articles appear to be reasonable and
available artifacts for identifying this area. This is
done for three reasons. First, their affiliation with
the bibliometric theory of problematic network
analysis provides support for their representation
of scholarly activity. (Coutial, 1994; Courtial,
Callon, & Sigogneau, 1984). This theory views
scholarly communication and literature as a series
of problematisations that reflected the underlying
intellectual discourse. Journal articles are seen as
the primary artifact from which to extract elements
of that discourse. Second, analyses of journals do
not consume the financial resources inherent in
more exhaustive treatments. Third, their online
availability makes this current analysis easier to
replicate and contrast when used as a benchmark
upon subsequent research.
Research questions. One advantage of surveying the field is to let the field define itself as it
exists in the extant literature. In this regard, almost
40 years of archival data was examined by combining the results of a previous survey (Wallace,
1999) with this 2005 examination. Hopefully, this
will offer a more comprehensive look at how CMC

Computer-Mediated Communication Research

research is viewed through a database analytic

lens. As search engines replace annotated bibliographies and other indexical instruments, the
nature of the scholarly profile produced through
this lens becomes more critical.
In regard to the above literature framework,
this chapter poses four research questions concerning the computer-mediated communication
research. These questions are general in nature,
and align themselves theoretically with artifacts,
producers, and research concepts. The conceptual
issues are somewhat more complex than general
identification and demarcation that led to the
creation of a fourth research question. Simply
put, the research questions posed are:

RQ1: What are the primary journals of CMC

literature, both past and present?
RQ2: Who are the producers of CMC literature, both past and present?
RQ3: What are the CMC concepts being
examined in scholarly journals?
RQ4: How do the current research concepts
of CMC relate to other topics in the field?

Research question 1: Journals. Research

question 1 addresses the scholarly journals representing literature concerning computer-mediated communication. Specifically, it will identify
these artifacts by the analysis of frequency data
and an identification of core journals across several disciplines. Common databases were set up
both in the previous and the current study so that
frequency and other component features could
be identified.
One advantage of using journals over other
artifacts, such as citations, is that they are believed
to contribute to minimizing threats to validity
(Cronin & Overfelt, 1994; Glaenzel & Moed, 2002;
Hyland; 2003; Schloegl & Stock, 2004). Cites
often go to secondary sources (obfuscating the
real influence), and authors often do not hold the
opinions that citation patterns attributed to them.
Once frequency counts are established, a number

of other component features can be derived, such

as prolific and core journals. This may provide
indicators of publication patterns that can be used
to target future scholarship.
Research question 2: Authors. Research
question 2 looks at producers of CMCs scholarly communication. It will be addressed by the
analysis of frequency data concerning authorship.
Frequency of author publication rate will give a
glimpse into the most prolific authors, and has
been indicative of some of the major theoretical
contributions of bibliometrics (Cronin & Overfelt,
1994; Nicholls, 1989; Rice, Chapin, Pressman,
Park, & Funkhouser., 1996).
Research question 3: Concepts. Research
question 3 identifies the more prominent areas
of focus in regard to CMC research. It will be
addressed by the analysis of indexical key words
from the master database. Indexical key words
were chosen as concepts because they have demonstrated useful in identifying research trends in
other studies (Callon, Courtial, & Laville, 1991;
Cambrosio, Limoges, Courtial, & Laville, 1993;
Stegmann & Grohmann, 2005). Key words stand
for indicators of scholarly problematisations that
are being articulated in the literature (Courtial,
1994; Callon et al., 1991)
Research question 4: Relationship of research
concepts. Research question 4 investigates how the
research concepts of CMC relate to topics in the
field. It will be addressed through coword analysis.
This is generally done by analyzing the proximity
that key words and other concepts have with one
another (Diodato, 1994; Larsen & Levine, 2005;
Stegmann & Grohmann, 2005).

Method
In order to produce a literature survey that spans
almost 40 years, two studies were combined. A
more recent analysis of CMC research extended
a previous CMC literature study (Wallace, 1999).
The previous studys data was collected in June

301

Computer-Mediated Communication Research

of 1998, for 1997 and prior, utilizing the indexes

of ABI/INFORM, ERIC, Psychlit, Social Science
Index, and Sociofile. The more recent studys
data was collected in August of 2005. EBSCO
was used as a common vender in the 2005 study.
Academic Search Complete, Communication &
Mass Media, Business Search Premier, ERIC,
PsychInfo, and PsychArticles were selected as
databases. The selected databases are considered
prominent resources in the areas of business,
education, psychology, and the social sciences.
Furthermore, these databases are noted for their
strong research affiliations, and have an intuitive link with CMC. A key word search of the
respective databases was done for the variable
computer-mediated communication.. Both
hyphenated and unhyphenated variations of the
term computer-mediated were used to insure
breadth of coverage.
Each of the databases had idiosyncratic characteristics that made identical survey techniques,
across all databases, impossible. However, all
methodologies adhered to the following guidelines. Databases were examined to see if there were
any inherent journal markers that would enable
the extraction of journal articles. Books, book
reviews, conference papers, and other nonjournal
materials were eliminated through either filters
or inspection. Extraneous information and fields
were eliminated. Documents were then placed into
a common file. Procedures were similar for both
studies, and a relatively exhaustive description is
available in Wallace (1999).

Data Analysis
Journals. The data file was examined in terms
of journal frequency. From this, it could be determined the identity of CMC journals and how
they were positioned in terms of the overall literature as defined by this chapter. Subsequently,
Bradford-type partitions were derived to identify
the core journals.

302

Bradford partitions are where journals are

ranked from most to least prolific in terms of
number of articles produced concerning the
subject under scrutiny. They are then divided
into zones that have roughly the same number
of articles. Predictably, the journals that produce
the most articles have the least number of journals
in their zone. The zone that has the least journals is used to identify the core journals for a
subject area. Bradford recommended the use of
three zones (Diodato, 1994). The previous study
serendipitously arrived at three zones through a
somewhat more complex procedure. These zones
had a relatively equal number of journal articles
with the core zone containing a relatively few
prolific journals. For comparative purposes, the
more current study utilized three zones as well.
Authors. The 2005 analysis of authors was
also restricted by comparative constraints. The
previous study used the straight count method of
authorship. The straight count method identifies
all the first authors from the journal database.
Therefore, both studies used straight count to
extract authors for analysis.
After authors were extracted from the master database, they were then rank ordered, and
standardized. Frequency and cumulative data
was then calculated. The CMC authors were then
analyzed from an authors x number of contributions format.
Concepts. Frequency of problematisations
overall were calculated and analyzed for segmentation. This is not only critical for this research
question, it is imperative for RQ4. Both Hinzes
(1994) 10 journals or more and Rice et al.s (1996)
top 200 key words were considered exemplars
in this regard. The decision was made to view the
distribution in terms of which kind of demarcation
was most efficient while capturing the majority
of the distribution. Fifty percent was considered
the benchmark on which to base the comparisons.
This is an area where the literature distribution
must be known to be able choose the more efficient

Computer-Mediated Communication Research

method. If neither of the previously mentioned

methods surpassed the 50% benchmark, then
the literature would be deemed too disparate to
determine more prominent terms.
Once segmentation was determined, then the
core key words were longitudinally examined
for their visibility. The previous study examined
key words concerning 1997 and prior. The more
recent study surveyed the literature from 1997
through 2005. The overlap was intentional as many
databases literatures are not well represented in
their last year of analysis (Egghe, 1990). Indicators of this droop were present in the previous
1997 survey.
Relationship of research concepts. The above
concepts were then scanned for more familiar research affiliated terminology. Terms that appeared
methodological, or had the term research as a
component-attached term were extracted. Establishing relationships was done through cluster
analysis. A number of clustering procedures have
been used in the analyses of emerging fields. The
complete linkage method (furthest neighbor) was
considered an appropriate procedure because of
its rigor over the single linkage method and its
comparability to the previous study (Aldenderfer
& Blashfield, 1984; Hinze, 1994, Wallace, 1999).
The lack of knowledge of current granularity of
findings also makes this procedure well suited.
(McCain, 1989; Spasser, 1997). As with other studies, similarity between concepts was measured
using a Pearson relation procedure. (e.g., Hinze,
1994; McCain, 1989; Spasser, 1997).

Results
The 1997 CMC study generated 611 unique article
references for 1997 and prior. Four hundred and
fifty-nine, or about 75%, tended to be education
related as indicated by their ERIC affiliation.
The current examination started from a more
mature 1997 dataset through 2005. It generated

1,326 unique article references. Nine hundred and

fifty-three, or about 72%, tended to be education
related as indicated by their ERIC affiliation.
The following sections survey journal publication trends for CMC, identify some of the
more common terminology within that survey,
and then narrow that terminology to researchaffiliated terms and concepts clustering around
those terms.

Artifacts: Survey of Core Journals

A Bradfordian-type analysis was done both in
the more current and the previous study (Table
1). In the 1997 study, journals had to produce
six or more articles total to be considered core.
In the 2005 study, journals had to produce 11 or
more articles.
While roughly the same number of journals
were considered core in both studies, only six
journals were recognized as core in both studies.
These were
American Journal of Distance Education,
Computers & Education, Computers and Composition, Educational Media International, Information Society, Internet Research, Journal of Educational Computing Research, and TechTrends.
A cursory look at journal affiliation reveals that
the journals are largely education affiliated. Two
notable exceptions are Information Society and
Internet Research (see Table 2).

Table 1. Core journal distribution

Early
Journals
1
22*
2
54
3
153
*6 or more articles

Zones

Early
Recent
Recent
Articles Journals Articles
195
25*
430
196
101
427
195
360
426
*11 or more articles

303

Computer-Mediated Communication Research

Table 2. Survey of core journals for CMC

1966-1997
Behaviour and Information Technology
Canadian Journal of Educational Communication
Communication Education
Communication Research
Computers in Human Behavior
Educational Technology
Educom Review
Human Communication Research
Interpersonal Computing and Technology
Journal of Communication
Journal of Computer-Mediated Communication
Learning and Leading with Technology
Organization Science

Productivity of these core journals ranged, in

the earlier research, from a high of 16 to a low
of 6. The more recent research frequency was
appreciably higher ranging from a high of 34 to
a minimum threshold level of 11.
To give an idea how this distribution might
be comparable to other literature descriptions,
in terms of longevity, it was tracked over time.
While no journals appear till 1984, databases were
searched in previous years. The most far-ranging
search occurred in ERIC, which started in 1966.
Records were also examined from ABI/INFORM,
Psychlit, Sociofile, and Social Science Index in
1986, 1971, 1971, and 1983, respectively. All databases were producing articles by 1986

Producers: Survey of Core Authors

Three or more references were used to demarcate
core authors in both studies. This formulation has
the benefit of fitting within the general contention

304

1997-2005
British Journal of Educational Technology
Business Communication Quarterly
CALICO Journal
CyberPsychology & Behavior
Distance Education
Distance Education Report
Indian Journal of Open Learning
Instructional Science
Internet and Higher Education
Journal of Adolescent & Adult Literacy
Journal of Computer Assisted Learning
Journal of Educational Technology Systems
Journal of Instruction Delivery Systems
Journal of the American Society for Information
Science
New Media & Society
Quarterly Review of Distance Education
Small Group Research

that authors producing one or two articles have

the vast majority of the distribution and should
be excluded from consideration (e.g., 94% by
Burnham, Shearer, & Wall 1992; 83% by Keenan,
1988). On the other hand, because the production
numbers were low in general, using more than
three was considered too severe a threshold for
the respective studies.
The CMC literature examined was heavily
weighted toward authors producing only one
article. The early 1997 research had 506 different
first authors producing 611 different references.
Close to 96% of the authors had one or two publications. They accounted for 87% of the articles
produced. The 21 core authors were published in 54
different journals producing 80 different articles.
The 2005 survey had a similar distribution, with
1,169 authors producing 1,326 different articles.
However, the core authors produced a somewhat
smaller 8.5% of the articles. Specifically, there
were 31 authors that produced 113 articles in

Computer-Mediated Communication Research

Table 3. Survey of CMC first authors

1966-1997 Authors
Adrianson, Lillemor
Phillips, Gerald M.
Baym, Nancy K.
Rice, Ronald E.
Collis, Betty
Riel, Margaret
Dyrli, Odvard Egil
Rojo, Alejandra
Harris, Judith B.
Schrum, Lynne
Hiltz, Starr Roxanne
Snyder, Herbert
Lea, Martin
Valacich, Joseph S.
Mantovani, Giuseppe
Walther, Joseph B.
Matheson, Kimberly
Weinberg, Nancy
McMurdo, George
Zack, Michael H
Olaniran, Bolanle A

71 different publications. Only Joe Walther was

included in both previous and recent sets of core
authors (Table 3).

Concepts: Survey of Key Terms

First, to contextualize the conceptual survey
concerning CMC, it is useful to understand the
scope of the current and previous examination.
The previous study had 1,787 unique problematisations (indexical terms) contributing to a total
dataset of 6,898 terms. The 2005 study had 2,700
unique problematisations contributing to 12,935
total terms.
The overall datasets were then examined
for more dominant concepts. Both Hinzes and
Rices methods detailed earlier surpassed the
50% threshold. The previous study used Hinzes
(1994) demarcation, and identified 116 core terms.
They accounted for more than 3,684, or 53%, of
the total indexical terms mentioned. However,
Hinzes demarcation was primarily used because
the previous studys emphasis was on the totality of the conceptual base. Since the 2005 study
was designed specifically to identify research
affiliations within that base, Rice et al.s (1996)

1997-2005 Authors
Abrams, Zsuzsanna Ittzes
Li, Qing
Arbaugh, J. B
MacDonald, Lucy
Baron, Naomi S
Riva, Giuseppe
Belz, Julie A
Rourke, Liam
Benbunan-Fich, Raquel
Saba, Farhad
Caverly, David C.
Savicki, Victor
Fahy, Patrick J
Selwyn, Neil
Flanagin, Andrew J.
Trentin, Guglielmo
Guguen, Nicolas
Tu, Chih-Hsiung
Hampton, Keith N.
Vrooman, Steven S.
Haythornthwaite, Caroline
Walther, Joseph B.
Herring, Susan C
Warnick, Barbara
Johnson, E. Marcia
Wellman, Barry
Kling, Rob
Wilson, E. Vance
Kock, Ned
Wolfe, Joanna
Lee, Lina

more liberal top 200 terms was utilized. This

accounted for 7,827 or roughly 60% of the entire
conceptual set.
While a complete discussion of the core terminology would be unwieldy at best, a physical
inspection of the two studies indicated about 30%
of the totality of terms is included in the 21 most
prominent descriptors. With the admitted concession that computer-mediated communication is the
search term predicating the study, these terms still
provide an overview of some of the more dominate
themes that have emerged in the CMC literature.
Table 4 details both studies with the count and
rank of terminology that occurred.
Thirteen of the 21 terms appeared in both
studies. As would be expected, even with rank
differentiation, the more recent study has substantially more contributions. Online systems had the
highest overall percentage increase. Decreases
were only observed concerning the terms electronic mail and computer networks.
There were eight terms that were not reciprocal (Table 5). The overwhelming number of these
fell in the bottom half of the rankings where
predominance was not as stable.

305

Computer-Mediated Communication Research

Table 4. Common core terms for CMC

Term

Difference

Computer-Mediated Communication
Higher Education
Internet
Distance Education
Computer-Assisted Instruction
Foreign Countries
Educational Technology
Computer Uses in Education
Online Systems
Electronic Mail
Information Technology
Computer Networks
Teleconferencing

210%
250%
206%
322%
198%
268%
317%
219%
332%
78%
139%
51%
111%

Freq
2005
960
510
287
245
212
153
152
151
146
107
71
70
69

Rank
2005
1
2
3
4
5
7
8
9
10
12
19
20
21

Freq
1997
457
204
139
76
107
57
48
69
44
137
51
136
62

Rank
1997
1
2
3
8
6
14
18
10
19
4
17
5
12

Table 5. Unique core terms for CMC

World Wide Web

Telematics

2005

Freq
2005
175
142

Rank
1997
2005
6
Telecommunications
11
Information Networks

Freq
1997
100
72

Rank
1997
7
9

Second Language Learning

Student Attitudes
Teaching Methods
Second Language Instruction
Online Courses
Interaction

93
90
83
82
81
78

13
14
15
16
17
18

67
57
53
52
42
42

11
13
15
16
20
21

Interpersonal Communication
Computer Applications
Adulthood
Computers
Experimental Theoretical
Group Dynamics

Table 6. Survey of research terms

Concept
Case Studies
Communication Research
Comparative Analysis
Content Analysis
Discourse Analysis
Educational Research
Evaluation Criteria
Evaluation Methods
Literature Reviews
Pilot Projects
Research
Research Methodology
Surveys
Use Studies

306

1997
29
38
21
17
14
5
3
7
11
9
5
12
14
21

2005
37
36
32
14
28
23
11
31
15
11
14
11
17
14

Computer-Mediated Communication Research

The more exhaustive 2005 analysis was used

to identify research-affiliated terms to be inlaid
within respective conceptual maps of the two studies. The research-affiliated concepts are listed for
both the 1997 and 2005 research studies in Table 6.
The highest number of scholarly articles was connected with the term communication research.
The top-ranked term was communication
research for the 1997 survey and case studies
for the 2005 survey. As would be expected, most
concepts were identified at higher levels in the
larger, more recent study. However, there were
some exceptions to this trend. These included
communication research, research methodology,
and use studies.

Conceptual Map of CMC

Research Methodologies
Research question 4 was explored in two ways.
First was the identification of prominent problematisations within each of the previously identified
research affiliated terms. This was done by setting
a somewhat arbitrary four-occurrence threshold
that had to be surpassed. What follows is a more
lexical description of those relationships. Clearly,
terms that are more numerous are not necessarily strongly related, and terms strongly related
are not necessarily more numerous. Below are
topical areas that were most dominant when the
respective research problematisations were used
as filters. Because of the magnitude of higher
education occurrences, it was included in almost
all categories.
Communication research. Both studies included higher education, Internet, communication
behavior, and group dynamics as areas of foci.
Additionally, the 2005 study included interpersonal relationships, student attitudes, and World
Wide Web. The 1997 conceptual set was relatively
extensive including the following terms:

Adulthood
Communication

Communication Social Aspects

Computer Applications
Computer Assisted Instruction
CMC Systems
Computer Networks
Computers
Decision Making
Electronic Mail
Group Decision Making
Interpersonal Relations
Organizational Communication
Social Interaction
Telecommunications

Collaborative learning
Distance education
Foreign countries
Instructional effectiveness
Interaction
Internet
Online systems
Research methodology
Student attitudes
Student evaluation

Case studies. Both studies included the indexical terms higher education, Internet, distance
education, foreign countries, computer-assisted
instruction, and electronic mail. The 2005 study
also had World Wide Web, instructional effectiveness, online systems, information technology, college students, literacy, and teacher role. The 1997
study was more writing oriented with computer
networks, teaching methods, writing research,
student attitudes, collaborative writing, technical
writing, and writing instruction.
Evaluation methods. Only higher education
and computer-assisted instruction appeared
prominently in both studies when evaluations
methods was used as a filter. However, no other
terms, except for the previously mentioned higher
education, reached a level of predominance in the
1997 study. The 2005 conceptual set was relatively
extensive including the following terms:

307

Computer-Mediated Communication Research

Teaching methods
Teleconferencing
World Wide Web

Comparative analysis. Dominant concepts in

both studies included higher education, distance
education, and computer-assisted instruction.
The 2005 study also had face-to-face communication, interaction, instructional effectiveness,
interpersonal communication, nontraditional
education, second-language instruction, and second-language learning. The 1997 study had more
dominant terms including computer networks,
futures of society, teleconferencing, electronic
mail, and experimental theoretical.
Discourse analysis. Higher education and
interpersonal communication was the only dominant concept in both studies. The 2005 dominant
concepts included communication behavior,
Internet, language research, second-language
instruction, second-language learning, sex differences, and written language. Only two additional
terms were dominant in the 1997 study. These
were computer-assisted instruction and computer
networks.
Educational research. The term educational
research did not have sufficient mentions to be
considered in the dominant term analysis for the
1997 study. The dominant terms for the 2005
study included higher education, computer-assisted instruction, cooperative learning, distance
education, educational technology, Internet,
nontraditional education, online systems, and
World Wide Web.
Surveys. There were no dominant terms included in both studies other than higher education.
The 2005 study featured student attitudes, distance education, and World Wide Web, while the
1997 study featured Internet, computer networks,
electronic mail, teleconferencing, information
networks, and scholarly communication.
Literature reviews. Distance education was
a dominant term in both studies concerning
literature reviews. The 2005 study also included

308

community and information technology. The

1997 study additionally had higher education,
computer networks, electronic mail, and teleconferencing.
Research. When used as a filter, the singular
term, research, did not have sufficient mentions of
any terminology to be included in this analysis.
Use studies. Higher education, Internet, and
electronic mail were dominant in both studies.
Additionally, the 2005 study featured the term
World Wide Web. The 1997 studys dominant
concepts were more ranging, including communication thought transfer, computer networks,
foreign countries, online systems, tables data,
telecommunications, and teleconferencing.
Content analysis. Only higher education was
identified for both studies. The only additional
dominant term for the 2005 study was Internet.
Terms for the 1997 study included electronic mail,
interpersonal communication, research needs,
and tables data.
Pilot project. There were no dominant mentions of terminology in both studies except for
higher education. Additionally, the 2005 study
included foreign countries and educational technology. The 1997 study had electronic mail.
Evaluation criteria. Evaluation criteria did
not have sufficient enough mentions for dominant
term analysis in the 1997 study. The dominant affiliated terms with the 2005 study included higher
education, computer-assisted instruction, distance
education, Internet, and online systems.
Research methodology. There was no overlap
in dominant terms. The 2005 study had distance
education and interaction, and the 1997 study
had higher education, computer networks, and
electronic mail.
The second way research question 4 was
explored was to examine strength of affiliation
through cluster analysis. This was thought to
provide some clues as to which core terms were
most closely related to particular research problematisations. Some of the linkages were clearly
methodological (Table 7). However, not all terms

Computer-Mediated Communication Research

were included in both analyses due to the number

of occurrences when research term filters were
applied. Table 7 identifies clusters of research
and affiliated co-occurring terms that appeared
in at least one of the studies. Clustered terms

tend to be conceptual sets that journal articles

focus upon.
Aside from terms that were affiliated across
both studies, there were terms that only appeared

Table 7. Longitudinal research term comparison

2005 - research clusters

1997- research clusters

Case studies

Case studies

Literacy

group discussion

computer literacy

graduate study

Tutoring

writing instruction

Communication research

Communication research

communication behavior

Organizational communication

Discourse analysis

interpersonal communication

classroom communication

Computers

Comparative analysis

Discourse analysis

face to face communication

microcomputers

interpersonal communication

man machine systems

Computer attitudes

Content analysis

Computer literacy

Research methodology
Evaluation methods*

Comparative analysis
learner controlled instruction

Survey
electronic mail

Content analysis

ethics

Research methodology
research needs

Use studies
computer use

Survey

undergraduates

Teleconferencing

technology utilization

foreign countries
Use studies

Literature reviews
Community

Literature reviews
tables data
Questionnaires.
*=not enough for 1997 analysis

309

Computer-Mediated Communication Research

in the 2005 analysis. These relationship clusters

included:

Educational Research: Distance education,

nontraditional education, computer-assisted
instruction, and educational practices
Evaluation Criteria: Feedback, course
content, and course development
Pilot Project: Information services, library
services, reference services, user needs,
information, and electronic libraries
Research: Communication

Discussion
This chapter analyzed what is meant by computer-mediated communication research? While
a number of suitable answers exist for this question, it chose to let the field define itself as it
exists in the extant literature concerning business,
education, psychology, and the social sciences. In
this regard, almost 40 years of archival data was
surveyed. Wallaces study of 1997 and prior literature was combined with this 2005 examination
to offer a look at how CMC research is viewed
through a database analytic lens. Because of the
interdisciplinary nature of computer-mediated
communication (CMC), a parallel analysis of
multiple databases from different perspectives was
used (Ingwersen & Christensen, 1997; McLaughlin, 1994; Spasser, 1997). It should be noted that
as database sophistication increases regarding
education, business, and social life in general,
this kind of lens may become more crucial.
The descriptive nature of this study necessarily dictated a balance between rigor and latitude.
Several limitations must be considered in this
respect. These include generalization, design
limitations, and theoretical assumptions. First,
operationalization of the domain should clearly
impact the use of the findings. Therefore, results
from this analysis do not claim to identify characteristics of CMC beyond the domain examined.

310

Second, the selection of cluster analysis over other

classifying procedures has a number of inherent
limitations. This was considered a necessary
condition given the emerging nature of the field.
Emergence (or lack of emergence) of clusters will
have to be weighed against the nonrandom nature
of their partitioning (Aldenderfer & Blashfield,
1984). Third, the design was somewhat restricted
in order to have a longitudinal comparison. The
above limitations were addressed in the traditional
way for descriptive studies. Detailed explanations,
including supporting literature, were provided
for choices made.

Conclusion
This survey of computer-mediated communication literature revealed three interesting trends.
The first trend is a paradoxical turbulence and
stability common in literature surveys. This pattern was somewhat present for articles, journals,
authors, concepts, and research affiliations. The
total articles doubled while the number of core
journals remained relatively constant. Both the
overall production of articles and the production
of core journals increased by 217%. Despite this
increase, the number of core journals producing
those articles only advanced by 3, from 22 to
25.
The total number of core authors producing
more than two articles also had relatively little
growth, while there was a virtual turnover in
actual author names. Joe Walther was the only
author to emerge in both surveys. Core authorship increased by an anemic 30%, from 21 to 31.
When considering total articles produced, core
author production actually shrank from 13% in
the 1997 survey, to 8.5 % in the 2005 survey.
The top 21 terms in both studies accounted for
30% of the total indexical terms. Eight of those
terms were not reciprocal. Most of the unique
terms can be attributed to shifts in the more turbulent bottom half of the distribution. However,

Computer-Mediated Communication Research

telecommunications and information networks

were ranked seventh and ninth, respectively, in
the 1997 survey, and were not prominent in the
2005 study. One interesting note is that while
World Wide Web was not a core terminology
in the 1997 study, it was identified as the number
one emerging concept. In the 2005 survey, it was
firmly established as sixth in overall mentions.
Communication research and discourse analysis
were the only research affiliated terms that fell
within the top 21 terms of both surveys.
The majority of the previously identified research terminologies emerged in both studies, but
were not related to the same issues. Nine out of
the 14 research affiliated terms emerged in both
surveys. Serendipitously, the only two terms that
were related to each other were also research affiliated. These were content analysis and research
methodology. The terms educational research,
literature reviews, research, pilot projects, and
evaluation criteria were exclusive in that they
did not overlap with any of the other dominant
research terms.
The second trend is that literature tended to
have an online and higher education focus. Five
hundred and ten articles were higher-education
related. Internet was the second most populous
descriptor, with 287 articles. Furthermore, World
Wide Web had 176 articles, and online systems
had the highest percentage increase with 146
articles. Besides articles, both surveys indicated
in excess of 70% of the journals were related to
education. Core journals were also highly education and Internet related.
Higher education was also connected with
almost all research terminology for both surveys.
The only exceptions were the terms research,
research methodology, and literature review for
the 2005 survey. The 1997 study had higher education absent for educational research, evaluation
criteria, and the singular term, research. While
the 1997 survey results could be attributed to low
production numbers, this was not the case with the
2005 results. Other problematisations with similar

production numbers managed to have at least four

articles with a higher education affiliation.
The third trend emerged through a post hoc
analysis of the 2005 dataset that was not conducted
in the 1997 study. Journals were examined to
see if they affiliated with any of the dominant
research terminology. Most journals were affiliated with a single prominent research term.
Eight were affiliated with two terms, and only the
journal Internet Research involved three terms
that included comparative analysis, evaluation
criteria, and use studies. The appendix describes
the research affiliation as indicated by key word
term, journal that it was published in, and the
count of articles in a journal concerning a particular research term.

Future research
Clearly, this study detailed CMC research as an
area that tended to be education and Internet
affiliated. Furthermore, the computer-mediated
communication literature prominently used a
number of textual analysis techniques, such as
content and discourse analysis. Noticeably absent
were articles focused on possible experimental
techniques, ethnographies, and focus groups.
This does not mean that these were not tools used
in the literature, merely that they were not the
focus of the more prominently presented research
articles (Schneider & Foot, 2004). Surveys also
had a surprisingly diminutive presence. Certainly,
there are both specific and extensive treatments
of online survey methodology (e.g., Andrews,
Nonnecke, & Preece 2003; Katz & Rice, 2002).
However, the current examination suggests a need
for this and other volumes to specifically localize
methodological issues relevant to computer-mediated and online communication.
With the exception of Joe Walther, core authors had completely overturned. While Nicholls
concedes the robustness of the straight count, an
exhaustive identification of authorship might more

311

Computer-Mediated Communication Research

fully answer questions concerning ranking issues,

shifting authorships, and omissions (Cronin &
Overfelt, 1994; Nicholls, 1989).
One general consideration is that indexed
literature does not always reflect the centrality
of artifacts and producers. The current analysis
focused on work being produced without discrimination in regard to usage. Identification of
CMCs literature usage patterns might provide
complementary information to the current study.
Furthermore, it would help shed light on whether
theoretical or pragmatic issues are driving CMCs
scholarly communication.
Another important consideration is that it
is unlikely that people engaged in CMC are
overwhelmingly involved in education-related
phenomena (Katz & Rice 2002; Papacharissi &
Rubin, 2000). If the study of CMC is to more accurately reflect current usage patterns, it should be
aggressively broadened in other areas. Conceptual
linkages exposed by this analysis are formative
at best. Future studies should rigorously examine
both the linkages and the conclusions in regard
to their longevity and stability.
However, while conceding a number of caveats, this study does identify component features
of past and more recent computer-mediated
communication research. These resources are
readily available online through many library
services and outside vendors. Therefore, CMC
and online researchers can use these features
to identify research redundancies, and opportunities to consider, and hopefully more efficiently
advance, the scholarship of the field.

tional Journal of Human Computer Interaction,

16(2), 185-210.

References

Dick, J., & Blazek, J. (1995). The nature of the

discipline: A bibliometric study of communication
with some comparisons to library reference/information work. In J. B. Whitlatch (Ed.), Library
users and reference services (pp.289-304). New
York: Haworth Press.

Aldenderfer, M. S., & Blashfield, R. G. (1984).

Cluster analysis. Newbury Park, CA: Sage
Andrews, D., Nonnecke. B., & Preece, J. (2003).
Electronic survey methodology: A case study in
reaching hard-to-involve Internet users. Interna-

312

Borgman, C. L., & Rice, R. E. (1992). The convergence of information science and communication:
A bibliometric analysis. Journal of the American
Society for Information Science, 43(6), 397-411.
Burnham, J. F, Shearer, B. S., & Wall, B. C. (1992).
Combining new technologies for effective collection development: A bibliometric study using
CD-ROM and a database management program.
Bulletin of the Medical Library Association, 80,
150-156.
Callon, M., Courtial, J. P., & Laville, F. (1991). Coword analysis as a tool for describing the network
of interactions between basic and technological
research: The case of polymer chemistry. Scientometrics, 22(1), 155-205.
Cambrosio, A., Limoges, C., Courtial, J. P., &
Laville, F. (1993). Historical scientometrics?
Mapping over 70 years of biological safety research with co-word analysis. Scientometrics,
27(2), 119-14.
Courtial, J. P. (1994). A co-word analysis of scientometrics. Scientometrics, 31(3), 251-260.
Courtial, J. P., Callon, M., & Sigogneau, M. (1984).
Is indexing trustworthy? Classification of articles
through co-word analyses. Journal of Information
Science, 9, 47-56.
Cronin, B., & Overfelt, K. (1994). Citation-based
auditing of academic performance. Journal of
the American Society for Information Science,
45(2), 61-72.

Computer-Mediated Communication Research

Diodato, V. (1994). Dictionary of bibliometrics.

New York: Haworth Press.
Egghe, L. (1990). A note on different Bradford
multipliers. Journal of the American Society for
Information Science, 41(3), 204-209.
Glaenzel, W., & Moed, H. F. (2002).
Journal impact factors measures in bibliometric research.
Scientometrics, 53(2) 171-193.
Hargittai, E. (2004). Internet access and use in
context. New Media & Society, 6(1), 137-143.
Hinze, S. (1994). Bibliographical cartography of an
emerging interdisciplinary discipline; The case of
bioelectronics. Scientometrics, 29(3), 353-376.
Hyland, K. (2003). Self-citation and self-reference: Credibility and promotion in academic
publication. Journal of the American Society
for Information Science and Technology, 54(3),
251-259.
Ingwersen, P., & Christensen, F. H. (1997) Data
set isolation for bibliometric online analyses of
research publications: Fundamental methodological issues. Journal of the American Society for
Information Science, 48(3), 205-217.
Katz, J. E., & Rice, R. E. (2002). Social consequences of Internet use: Access, involvement and
interaction. Cambridge, MA: MIT Press.
Keenen, M. (1988). Report on the 1987 membership survey. Journal of Finance, 43, 767-777.
Larsen, T. J., & Levine, L. (2005). Searching for
management information systems: Coherence
and change in the discipline. Information Systems
Journal, 15 (4), 357-381.
Levinson, P. (1990). Computer conferencing in the
context of the evolution of media. In L. M. Harasim
(Ed.), Online education: Perspectives on a new
environment (pp. 3-14). New York: Praeger.

McCain, K. W. (1989). Mapping authors in intellectual space. Communication Research, 16(5),

667-681.
McCain, K. W., & Whitney, P. J. (1994). Contrasting assessments of interdisciplinarity in emerging specialties. Knowledge: Creation, Diffusion,
Utilization, (3), 285-306.
McLaughlin, M. (1994). Announcing the Journal of Computing Mediated Communication.
Retrieved January 30, 1999, from http://sunsite.
unc.edu/cmc/mag/1994/jul/jcmc.html
Nicholls, P. (1989). Bibliometric modeling processes and the empirical validity of Lotkas law.
Journal of the American Society for Information
Science, 40(6), 379-385.
Papacharisi, Z., & Rubin, A. M. (2000). Predictors of Internet use. Journal of Broadcasting and
Electronic Media, 44(2), 175-196.
Rice, R. E., Chapin, J., Pressman, R., Park, S.
& Funkhouser, E. (1996). Whats in a name?
Bibliometric analysis of 40 years of the Journal
of Broadcasting (& Electronic Media). Journal of
Broadcasting & Electronic Media, 40, 511-539.
Schloegl, C., & Stock, W. G. (2004). Impact
and relevance of LIS journals: A scientometric
analysis of international and German-language
LIS journalsCitation analysis vs. reader survey.
Journal of the American Society for Information
Science and Technology, 55(13), 1155-1168.
Schneider, S. M., & Foot, K. A. (2004).
The Web as an
object of study. New Media & Society, 6(1), 114-122.
Spasser, M. A. (1997). Mapping the terrain of
pharmacy: Co-classification analysis of the international pharmaceutical abstracts database.
Scientometrics, 39(1), 77-97.
Stegmann, J., & Grohmann, G. (2005). Hypothesis generation guided by co-word clustering.
Scientometric, 56(1), 111-135.

313

Computer-Mediated Communication Research

Wallace, J. D. (1999). An examination of computer-mediated communications scholarly communication. Unpublished doctoral dissertation,
University of Oklahoma.

key terms
Artifacts: Artifacts are any number of forms
of scholarly communication. Conceptually, they
could range from working papers to books.
Bibliographic Coupling: Bibliographic coupling is where two documents each have citations
to one or more of the same publication, but do not
have to necessarily cite each other.
Bibliometrics: The mathematical and statistical analysis of patterns that arise in the publication
and use of documents
Bradford Partitions: These partitions are
used in library and information science to establish core journals. The process ranks journals
from most to least prolific in terms of number of
articles produced concerning a subject. They are
then divided into three or more zones that have
roughly the same number of articles. The zone

314

that has the least journals is used to identify the

core journals for a subject area.
CMC: An acronym standing for computermediated communication or computer mediated
communication.
Co-Citation Analyses: The analysis of journals, articles, or authors that are cited together in
an article or articles.
Concepts: Concepts are terms or words used
by the producers themselves. This could include
words in titles and text, or assigned terminology
such as key words.
Core Authors: Core authors are generally
established through a well-articulated benchmark
that exceeds two publications.
Core Concepts: Core concepts are the relatively few concepts that account for a large amount
of problematisations under study.
Core Journals: Core journals are generally
established through a well-established benchmark
such as Bradford partitions.
Co-Word Analyses: The analyses of the cooccurrence of two or more words in one document
or in different documents.

Computer-Mediated Communication Research

Appendix: 1997-2005
Journal
Case studies

Count

Journal

Count

Evaluation methods

Computers and Composition

American Journal of Distance Education

Educational Media International

Instructional Science

Internet and Higher Education

Quarterly Review of Distance Education

Journal of Computer Assisted Learning.

Educational Technology & Society

Communication research

Computers & Education

Journal of Educational

Human Communication Research

Computing Research

Computers and Composition

Distance Education

Public Relations Review

Performance Improvement.

International Review of Research in Open and

Distance Learning

Literature reviews

Business Communication Quarterly

ARIST

Journal of Computer-Mediated Communication

Evaluation methods

Written Communication

Internet Research

Comparative analysis

6
2

Pilot project

Internet Research

Computers & Education

Educational Media International

Research
Content analysis

Journal of Language & Social Psychology

Educational Computing Research

Instructional Science

Discourse analysis

Survey
Educational Media International

American Journal of Distance Education

Computers and Composition

CALICO Journal

Use studies

Research on Language and Social Interaction

Internet Research

Written Communication

Research methodology
The Journal of the American Society for
Information Science

Educational research
British Journal of Educational Technology

Internet and Higher Education

Journal of Adolescent & Adult Literacy

Quarterly Review of Distance Education

Community College Journal

This work was previously published in the Handbook of Research on Electronic Surveys and Measurements, edited by R.
Reynolds, R. Woods, and J. Baker, pp. 207-222, copyright 2007 by Information Science Reference, formerly known as Idea
Group Reference (an imprint of IGI Global).

315

316

Chapter 1.24

Computer-Mediated
Communication in Virtual
Learning Communities
Lisa Link
Flensburg University of Applied Sciences, Germany
Daniela Wagner
University of Hildesheim, Germany

THE MEANING OF CMC

Computer-mediated communication (CMC) can
be simply defined as communication that takes
place between human beings via the instrumentality of computers (Herring, 1996, p. 1) and the
technology used in CMC as media that facilitate
the exchange of semantic content, transmitted
through telecommunication networks, processed
through one or more computers, between individuals and among groups (Rice, 1984, p. 438).
As a concomitant result of the widespread use of
computers and the Internet in the curriculum of
higher education as well as the inclusion of various forms of online learning, CMC has become
a common communication modality in teaching
and learning contexts. Although all forms of
CMC can be meaningfully integrated into learning contexts, the majority of CMC at present is

nevertheless text based and hence the focus of

this contribution.
CMC plays a pivotal role in the context of
e-learning models in higher education that offer
students new forms of learning that would not be
possible in a traditional classroom environment.
Cooperative and collaborative learning, problemoriented and project-based learning, as well as
authentic learning in which real-world scenarios
are simulated, are some examples. In the context
of these types of e-learning models, virtual learning communities as well as smaller virtual teams
are often an essential component. The learning
objectives of these models are manifold, but they
generally include offering students the opportunity to acquire, practice and enhance important
key competencies required in their future professional work. Hence, communication and media
skills as well as the ability to work cooperatively

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Computer-Mediated Communication in Virtual Learning Communities

in a virtual team are just as important as the actual

subject matter of the course itself.

CHARACTERISTICS OF CMC
CMC can be classified into two major groups:
asynchronous and synchronous CMC. The main
difference between these two types is temporal:
asynchronous CMC is time independent, that is, it
does not require that the communication partners
be simultaneously online, whereas synchronous
CMC takes place in real time or quasi real time,
requiring the telepresence of the communication
partners. E-mail, mailing lists, and discussion
forums are examples of asynchronous forms.
Chat rooms and shared whiteboards represent
synchronous forms of CMC.
A further classification of CMC is whether
it represents a one-to-one (1:1), one-to-many (1:
n) or many-to-many (n:n) communication form.
Depending on their use, the different types of
CMC can fall into more than one category, for
example, e-mail and chat can represent both 1:1
and n:n communication. A topic of interest in this
context is the double function CMC can have: It
can be used for individual communication, but also
for mass communication. This goes along with a
double function that is very interesting in a learning setting, for example, in higher education. Email messages and discussion forum postings can
simultaneously fulfill two successive functions:
(1) interpersonal communication between two or
more participants and subsequently (2) serve as
an information pool for other participants. Chats
that have a protocol option can also be used as
an information pool for passive students. Fritsch
(1998) coined the term witness learning to describe
the indirect learning possibilities of learners who
do not actively take part in interactions, but learn
from witnessing the interactions of others. In
virtual learning environments, participants have
ranked witnessing (i.e., reading) the interactions

of others high among the things they have learned

from (Fritsch, 1998; Link, 2002).
As text-based CMC is not realized face-to-face
but mediated via computers, the communication
partners cannot directly experience nonverbal
signals (e.g., facial expressions or gestures) or
paraverbal signals (e.g., voice volume and tone).
In practice, nonverbal and paraverbal signals are
often compensated for by emoticons, inflectives,
and other expressions created by entering characters on the keyboard.
Many studies, particularly those from a linguistic perspective, have investigated the style
participants use in communicating. The most
often noted phenomenon is no doubt the use of
colloquial language in chat, e-mail, and even in
discussion forums. This phenomenon is often
accompanied by unconventional orthography,
spelling and grammar errors, as well as slang
expressions. Furthermore, the communication is
influenced by a register embracing word creations
taken from the language of computer hackers.
If the model of Koch and sterreicher (1994) is
applied to CMC, it becomes apparent that the
linguistic characteristics of this type of written
communication cannot simply be classified as
totally oral or totally written: Here it is helpful
to differentiate between the concept and medium
of spoken and written language. Considering the
communication style of CMC, it must be regarded
as a hybrid language variety, displaying characteristics of both spoken and written language. As
its usage norms are becoming conventionalised,
some authors venture to propose that this communication style represents a new type of language
which they refer to as netspeak or cybertalk, which
cannot be compared to conventional communication (e.g., Crystal, 2001).
Studies with a focus on work-flow analysis
often examine CMC as to its impact on the effectiveness of communication in professional
contexts as well as on the dependency between
communication style and the roles of the commu-

317

Computer-Mediated Communication in Virtual Learning Communities

nication partners (e.g., communication between

team colleagues vs. communication between
employee and employer).
With regard to the literature and research reports published in the last 10 years, CMC plays
an important role in nearly every community
phenomenon. In social contexts, the style people
use to communicate influences to a high degree
the social relations and team dynamics.

CMC THEORIES
For the effective use of CMC in educational
contexts, a variety of computer-mediated communication theories can provide insights into
selecting appropriate CMC tools as well as understanding their limitations. Prevalent theories can
be categorized into three large groups (Dring,
2003, p. 128):
1.

2.

3.

Media choice (e.g., media richness theory:

Daft & Lengel, 1984, 1986; Rice, 1992; social
presence: Short, Williams, & Christie, 1976;
Rice, 1993)
Media characteristics (e.g., cues-filtered-out
model: Culnan & Markus, 1987; reduced
social cues: Sproull & Kiesler, 1988)
Mediated communication behaviour (e.g.,
social information processing: Walther,
1992; Fulk, Schmitz, & Steinfeld, 1990).

Media choice theories focus on cognisant

decision making about which medium to use for
specific communicative tasks. The media richness
theory (Daft & Lengel, 1984, 1986; Rice, 1992)
is by far the most predominant theory of media
choice. This theory defines a mediums richness
in terms of its capacity to carry information,
feedback, channel, source, and language. Placing
face-to-face communication at the richest end of
the spectrum and numerical computer documents
at the leanest, various communication media are

318

then placed along this scale. In the media richness

model of Reichwald, Mslein, Sachenbacher, &
Englberger (2000, p. 57), which includes most
forms of CMC, the spectrum from rich to lean
media range from (1) face-to-face dialog, (2) video
conference, (3) telephone/telephone conference,
(4) voice mail, (5) computer conference, (6) telefax,
(7) e-mail, (8) letter post and documentation. This
theory contends that in organisations information
is processed in order to reduce uncertainty or
equivocality. Reducing uncertainty (e.g., communicating the postponement of a scheduled
meeting) can be carried out best with lean media
whereas reducing equivocality (e.g., a team meeting about a new project) requires a rich medium.
Thus, the primary assertion of this theory is that
the information richness of the task should correspond to the media richness of the medium.
To date, empirical studies have not been able to
validate this theory.
Media characteristic theories examine the
impact that media characteristics of CMC have
on the communication process. The starting point
is that CMC is typically text based. Accordingly,
this results in reducing communication channels
to text only, which in turn filters out physical
and social cues. The cues-filtered-out approach
(Culnan & Markus, 1987) propounds that since
text-based CMC cannot communicate nonverbal
or paraverbal information, very little social and
sociodemographic information about the communication participants is transmitted. Additional
research has found that the lack of social cues can
be liberating, for example, in decreasing inhibition
(Kiesler, Siegal, & McGuire, 1984).
Mediated communication behaviour theories
focus on how participants behave during CMC,
whether they exchange social information, form
relational bonds, or use special Internet language.
Social information processing (Walther, 1992)
maintains that with time, CMC participants will
develop communication skills to compensate for
limitations of the medium. In contrast to the other

Computer-Mediated Communication in Virtual Learning Communities

CMC theories, social information procession

views the Internet as a new social environment
in which over time communication partners find
new ways in which to integrate nonverbal social
information into their online behaviour.

THE ROLE OF CMC IN VIRTUAL

LEARNING COMMUNITIES IN
HIGHER EDUCATION
In view of the substantial use of CMC in many
professional fields, especially in those dealing
with research and development projects and
distributed teamwork, higher education should
prepare students to work effectively with CMC
tools and familiarise them with the characteristics
of CMC situations as well as with the impact these
can have on work efficiency.
Even if many people know how to use an e-mail
program or a chat tool or how to place a message
in a Web-based bulletin board, this does not mean
that they are automatically aware of the advantages and disadvantages of CMC. In this respect,
there is still a significant lack of information and
need for training which should be addressed in
higher education. The boom in e-learning research
projects has afforded most institutes of higher
education the technical infrastructure necessary
to offer students learning opportunities in virtual
learning communities.
The following concept is just one example of
how students can achieve the above-mentioned
key competencies in virtual learning communities
in higher education. The concept has been developed and successfully implemented in the project
MEUM, a joint pilot project of the Flensburg
University of Applied Sciences and the University
of Hildesheim in Germany. Both institutes offer
specialised degree programmes in translation
and have teamed up to design and implement a
distributed telelearning programme in which students can learn about and gain firsthand practical
experience in Internet-based cooperative work

in the field of translation. The students work in

virtual teams on a concrete translation project.
Here, we define a virtual team as a geographically
distributed team, whose members communicate
mainly via CMC. A detailed description of the
MEUM project can be found in Wagner and
Link (2003).
The blended learning courses offered in the
context of MEUM are characterised by three
phases:
1.

2.

3.

Orientation phase in which the students

receive initial information about the course
didactics, objectives, and schedule and are
organised into teams. This phase also includes an introduction to the Internet-based
learning platform in which both learning
modules and CMC tools (e-mail, chat, and
discussion forums) are integrated;
Teamwork phase during which the virtual
teams work autonomously to prepare and
carry out their translation job; and
Team assessment and reflection phase in
which the students reflect on their team and
work process as well as on the impact the
CMC and the overall virtuality had on their
work.

In these courses, students are offered information on the theory and characteristics of CMC,
for example, in Web-based learning modules. The
virtual teams are free to choose which communication tools to use for the various tasks and phases
of their virtual teamwork. This has resulted in all
CMC tools being utilised. This combination of
theory, task-based teamwork, and joint reflection
phase has been rapidly accepted by the students
and the reflection phase, in particular, is seen by
them as a vital component of this concept.
For a successful transfer of this concept to
other courses, it is important to consider the
competencies required of students and instructors
in a virtual learning community.

319

Computer-Mediated Communication in Virtual Learning Communities

Competencies of the instructors: Virtual

learning communities require instructors to
take on the supportive role of mentors or tutors
in the background, functioning, for example, as
consultants on demand. In general, they prepare
the introduction of the virtual community and
during the course monitor the activities of the
communities from behind the scenes. In a preparatory phase, the instructor decides whether
face-to-face meetings should be integrated, and if
so, whether the team members should meet their
team colleagues at the start. Prior to the course,
students should be asked to give information
about their media and professional skills as well
as their prior experience with (virtual) teamwork.
Instructors must also decide how teams should
be formed: should students be given the task of
building their own teams or is it advantageous to
compose comparable teams according to student
competency profiles?
As our own experience in virtual learning communities has shown, it is, on the one hand, very
important to explain the role of the instructors in
detail to the students and, on the other hand, for
instructors to remain aware of their own role as
tutors or mentors. By monitoring the work phase in
the learning communities, instructors can decide
when it might be necessary to intervene in the
teamwork. This presupposes a didactic model in
which the tutor/instructor has a consulting function in the background.
Competencies of the students: The following
can be considered the requisite competencies for
members in a virtual learning community as well
as the fundamental skills and abilities that students
should have at the end of the course.
The members of a virtual learning community
are aware of the characteristics of and differences between the CMC tools chat, e-mail, and
discussion forums. They have knowledge of the
standard communication conventions for CMC
(e.g., netiquette and chatiquette). The members
are familiar with emoticons and other strategies

320

for compensating for the lack of nonverbal and

paraverbal signals in CMC, as well with moderation techniques, for example, in chats with more
than two participants. The virtual community
members know the importance of feedback for
asynchronous communication, for example,
a confirmation that an e-mail has arrived or a
response to a note in a discussion forum to encourage the communication. They understand
the role of small talk in CMC and the difficulties
that go along with the mixture of task-oriented
and relationship-oriented communication. It is
important that community members recognise
when communication results in misinterpretation
and are able to identify the problems that CMC
can produce in the learning community.
In the wake of globalisation, CMC and distributed virtual teams will become more and more
commonplace in professional and educational
contexts. At present we are still in an experimental
phase that calls for more empirical studies into the
efficient and effective use of these phenomena.
At the same time, future development of virtual
learning communities should be extended across
cultural and discipline boundaries. A new challenge in virtual learning communities will thus be
intercultural and interdisciplinary cooperation.

CONCLUSION
In this article we presented a definition and classification of CMC. CMC is a phenomenon that is
studied by a wide variety of disciplines: linguists,
social psychologists, and computer scientists
have proposed approaches to help understand
the particularities and impacts of CMC. In addition to a discussion of selected CMC theories,
we presented a sensibilization concept for CMC
in higher education with the aim of helping students and instructors attain the key competencies
required of members of a virtual (learning) community. The dissemination of distributed virtual

Computer-Mediated Communication in Virtual Learning Communities

teamwork in professional fields requires new skill

and competency requirements, which in turn
must be addressed in educational training. Giving
students the opportunity to learn how to utilise
information and communication technology as
well as to develop suitable strategies in CMC is
a new challenge for higher education.

REFERENCES
Crystal, D. (2001). Language and the Internet.
Cambridge, UK: Cambridge University Press.
Culnan, M. J., & Markus, M. L. (1987). Information technologies. In F. M. Jablin, L. L. Putnam,
K. H. Roberts, & L. W. Porter (Eds.), Handbook
of organizational communication: An interdisciplinary perspective (pp. 420-443). Newbury
Park, CA: Sage.
Daft, R., & Lengel, R. (1984). Information richness: A new approach to managerial behavior
and organization design. In B. Shaw & L. L.
Cummings (Eds.), Research in organizational
behavior, Vol. 6 (pp. 191-233). Greenwich, CT:
JAI Press.
Daft, R., & Lengel, R. (1986). Organizational
information requirements, media richness, and
structural design. Management Science, 32,
554-570.
Dring, N. (2003). Sozialpsychologie des Internet.
Gttingen, Germany: Hogrefe.
Fritsch, H. (1998). Witness-learning. Pedagogical
implications for net-based teaching and learning.
In M. Hauff (Ed.), [email protected]?
Entwicklung GestaltungEvaluation neuer
Medien (pp. 123-152). Mnster, Germany: Waxmann.
Fulk, J., Schmitz, J., & Steinfeld, C. (1990). A
social influence model of technology in use. In
J. Fulk & C. Steinfeld (Eds.), Organizations and

communication technology (pp. 117-140). Newbury Park, CA: Sage.

Herring, S. (1996). Introduction. In S. Herring
(Ed.), Computer-mediated communication: Linguistic, social and cross-cultural perspectives
(pp. 1-10). Amsterdam: John Benjamins.
Kiesler, S., Siegel, J., & McGuire, T. W. (1984).
Social psychological aspects of computer-mediated communication. American Psychologist,
39, 1123-1134.
Koch, P., & sterreicher, W. (1994). Schriftlichkeit
und Sprache. In H. Gnther, & O. Ludwig (Eds.),
Schrift und Schriftlichkeit. Ein Interdisziplinres
Handbuch Internationaler Forschung. Halbbd.1
(pp. 587-604). Berlin: de Gruyter.
Link, L. (2002). Die Bedeutung von Kommunikationsmedien undFormen in Internetbasierten
Fernlehrmodulen. In G. Bachmann, O. Haefeli,
& M. Kindt (Eds.), Campus 2002: Die virtuelle
Hochschule in der Konsolidierungsphase (pp.
408-416). Mnster, Germany: Waxmann.
Reichwald, R., Mslein, K., Sachenbacher, H., &
Englberger, H. (2000). Telekooperation. Verteilte
Arbeits- und Organisationsformen. Heidelberg,
Germany: Springer.
Rice, R. E. (1984). The new media: Communication, research and technology. Beverly Hills,
CA: Sage.
Rice, R. E. (1992). Task analyzability, use of new
media and effectiveness: A multi-site exploration
of media richness. Organizational Science, 3,
475-500.
Rice, R. E. (1993). Media appropriateness. Using
social presence theory to compare traditional
and new organizational media. Communication
Research, 19(4), 451-84.
Short, J., Williams, E., & Christie, B. (1976).
The social psychology of telecommunication.
London: Wiley.

321

Computer-Mediated Communication in Virtual Learning Communities

Sproull, L., & Kiesler, S. (1988). Reducing social

context cues: Electronic mail in organizational
communication. In I. Greif (Ed.), Computersupported cooperative work (pp. 683-712). San
Mateo, CA: Morgan Kaufmann.
Wagner, D., & Link, L. (2003). Real-world
scenarios in educational training. Establishing
a virtual translation network for students. In
M. Auer & U. Auer (Eds.), Proceedings of the
International Workshop Interactive Computer
Aided Learning ICL2003. Learning Objects and
Reusability of Content. Kassel, Germany: Kassel
University Press.
Walther, J. B. (1992). Interpersonal effects in computer-mediated interaction: A relational perspective. Communication Research, 19(1), 5290.

Key Terms
Blended Learning: Learning design that
combines various activities such as face-to-face
meetings, Internet-based learning modules, and
virtual learning communities.

Computer-Mediated Communication
(CMC): Communication between humans using
the computer as a medium.
Emoticons: A combination of punctuation
marks and other special characters from the
keyboard used to convey the tone of a computermediated communication message. For example,
the combination :-) depicts smiling.
Learning Platform: Software systems that
are used to deliver and support online teaching
and learning. Learning platforms manage access to
the platform and to learning materials and usually
include various communication tools.
Netiquette: Standard rules of courtesy and
correct behaviour on the Internet.
Witness Learning: A term coined by Dr.
Helmut Fritsch, senior researcher at the FernUniversitt in Hagen, Germany, that refers to the
indirect learning possibilities of learners who do
not actively take part in interactions, but learn
from witnessing the interactions of others, for
example, in online discussion forums.

Chatiquette: Standard rules of courtesy and

correct behaviour for online chat.

This work was previously published in the Encyclopedia of Virtual Communities and Technologies, edited by S. Dasgupta, pp. 4953, copyright 2006 by Information Science Reference, formerly known as Idea Group Reference (an imprint of IGI Global).

322

323

Chapter 1.25

Human-Computer Interaction
and Security
Kai Richter
Computer Graphics Centre (ZGDV), Germany
Volker Roth
OGM Laboratory LLC, USA

INTRODUCTION
Historically, computer security has its roots in the
military domain with its hierarchical structures
and clear and normative rules that are expected to
be obeyed (Adams & Sasse, 1999). The technical
expertise necessary to administer most security
tools stems back to the time where security was
the matter of trained system administrators and
expert users. A considerable amount of money and
expertise is invested by companies and institutions to set up and maintain powerful security
infrastructures. However, in many cases, it is
the users behavior that enables security breaches
rather than shortcomings of the technology. This
has led to the notion of the user as the weakest
link in the chain (Schneier, 2000), implying that
the user was to blame instead of technology. The
engineers attitude toward the fallible human and
the ignorance of the fact that technologys primary

goal was to serve human turned out to be hard to

overcome (Sasse, Brostoff, & Weirich, 2001).

BACKGROUND
With the spreading of online work and networked
collaboration, the economic damage caused by
security-related problems has increased considerably (Sacha, Brostoff, & Sasse, 2000). Also, the
increasing application of personal computers,
personal networks, and mobile devices with their
support of individual security configuration can
be seen as one reason for the increasing problems
with security (e.g., virus attacks from personal
notebooks, leaks in the network due to personal
wireless LANs, etc.) (Kent, 1997). During the
past decade, the security research community
has begun to acknowledge the importance of the
human factor and has started to take research on

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Human-Computer Interaction and Security

human-computer interaction into consideration.

The attitude has changed from blaming the user
as a source of error toward a more user-centered
approach trying to persuade and convince the
user that security is worth the effort (Ackerman,
Cranor, & Reagle, 1999; Adams & Sasse, 1999;
Markotten, 2002; Smetters & Grinter, 2002;
Whitten & Tygar, 1999; Yee, 2002).
In the following section, current research
results concerning the implications of user attitude and compliance toward security systems are
introduced and discussed. In the subsequent three
sections, security-related issues from the main
application areas, such as authentication, email
security, and system security, are discussed. Before the concluding remarks, an outlook on future
challenges in the security of distributed contextaware computing environments is given.

USER ATTITUDE
The security of a system cannot be determined
only by its technical aspects but also by the attitude of the users of such a system. Dourish
et al. (2003) distinguish between theoretical
security (e.g., what is technologically possible)
and effective security (e.g., what is practically
achievable). Theoretical security to their terms
can be considered as the upper bound of effective
security. In order to improve effective security, the
everyday usage of security has to be improved. In
two field studies, Weirich and Sasse (2001) and
Dourish et al. (2003) explored users attitudes to
security in working practice. The findings of both
studies can be summarized under the following
categories: perception of security, perception of
threat, attitude toward security-related issues, and
the social context of security.
Perception of security frequently is very inaccurate. Security mechanisms often are perceived
as holistic tools that provide protection against
threats, without any detailed knowledge about the

324

actual scope. Therefore, specialized tools often

are considered as insufficient, as they do not offer general protection. On the other hand, people
might feel protected by a tool that does not address
the relevant issue and thus remain unprotected
(e.g., firewall protects against e-mail virus).
Perception of threats also reveals clear misconceptions. None of the users asked considered
themselves as really endangered by attacks. As
potential victims, other persons in their organization or other organizations were identified, such
as leading personnel, people with important information, or high-profile institutions. Only a few
of them realized the fact that they, even though
not being the target, could be used as a stepping
stone for an attack. The general attitude was that
no one could do anything with the information
on my computer or with my e-mails.
Potential attackers mainly were expected to
be hackers or computer kids, with no explicit
malevolent intentions but rather seeking fun.
Notorious and disturbing but not really dangerous offenders, such as vandals, spammers, and
marketers, were perceived as a frequent threat,
while on the other hand, substantially dangerous
attackers such as criminals were expected mainly
in the context of online banking.
The attitude toward security technology
was rather reserved. Generally, several studies
reported three major types of attitudes toward
security: privacy fundamentalists, privacy pragmatists, and privacy unconcerned (Ackerman et
al., 1999). Users experiences played a considerable role in their attitude, as experienced users
more often considered security as a hindrance
and tried to circumvent it in a pragmatic fashion
in order to reach their work objectives. Weirich
and Sasse (2001) report that none of the users
absolutely obeyed the prescribed rules, but all
were convinced that they would do the best they
could for security.
Additionally, users individual practices are
often in disagreement with security technology.

Human-Computer Interaction and Security

People use legal statements in e-mail footers or

cryptic e-mails, not giving explicit information
but using contextual cues instead. In conjunction
with such subsidiary methods and the fact that
people often seem to switch to the telephone when
talking about important things (Grinter & Palen,
2002) indicates the poor perception users have of
security technology.
The feeling of futility was reported with
respect to the need for constantly upgrading
security mechanisms in a rather evolutionary
struggle (i.e., if somebody really wants to break
in, he or she will). As a result, personal accountability was not too high, as users believed that
in a situation where someone misused his or her
account, personal credibility would weigh more
than computer-generated evidence, in spite of the
fact that the fallibility of passwords is generally
agreed.
The social context has been reported to play
an important role in day-by-day security, as users
are not permanently vigilant and aware of possible threats but rather considered with getting
their work done. Therefore, it is no wonder that
users try to delegate responsibility to technical
systems (encryption, firewalls, etc.), colleagues
and friends (the friend as expert), an organization
(they know what they do), or institutions (the bank
cares for secure transfers). Most people have a
strong belief in the security of their companys
infrastructure. Delegation brings security out of
the focus of the user and results in security unawareness, as security is not a part of the working
procedure anymore.
Whenever no clear guidelines are available,
people often base their practice on the judgments of others, making the system vulnerable
to social engineering methods (Mitnick, Simon,
& Wozniak, 2002). In some cases, collaboration
appears to make it necessary or socially opportune
to disclose ones password to others for practical
reasons, technical reasons, or as a consequence
of social behavior, since sharing a secret can be

interpreted as a sign of trust. Such sharing is a

significant problem, as it is used in social engineering in order to obtain passwords and to gain
access to systems.
Dourish et al. (2003) came to the conclusion
that where security research has typically focused on theoretical and technical capabilities
and opportunities, for end users carrying out
their work on computer systems, the problems
are more prosaic (p. 12). The authors make the
following recommendations for the improvement
of security mechanisms in the system and in the
organizational context:

Users should be able to access security settings easily and as an integral part of the
actions, not in the separated fashion as it is
today; therefore, security issues should be
integrated in the development of applications
(Brostoff & Sasse, 2001; Gerd tom Markotten, 2002).
It is necessary that people can monitor and
understand the potential consequences of
their actions (Irvine & Levin, 2000) and that
they understand the security mechanisms
employed by the organization.
Security should be embedded into working
practice and organizational arrangement,
and visible and accessible in everyday physical and social environment (Ackerman &
Cranor, 1999).
Security should be part of the positive values
in an organization. So-called social marketing could be used to establish a security
culture in a company.
The personal responsibility and the danger
of personal embarrassment could increase
the feeling of personal liability.
The importance of security-aware acting
should be made clear by emphasizing the
relevance to the organizations reputation
and financial dangers.

325

Human-Computer Interaction and Security

As has been shown, the design and implementation of security mechanisms are closely
interlinked to the psychological and sociological
aspects of the users attitude and compliance
toward the system. Any security system is in
danger of becoming inefficient or even obsolete if
it fails to provide adequate support and motivate
users for its proper usage. The following sections
discuss these findings in the context of the main
application domains of computer security.

AUTHENTICATION
Information technology extends our ability to
communicate, to store and retrieve information,
and to process information. With this technology
comes the need to control access to its applications for reasons of privacy and confidentiality,
national security, or auditing and billing, to name
a few. Access control in an IT system typically
involves the identification of a subject, his or her
subsequent authentication, and, upon success, his
or her authorization to the IT system.
The crucial authentication step generally is
carried out based on something the subject knows,
has, or is. By far the most widespread means of
authentication is based on what a subject has (e.g.,
a key). Keys unlock doors and provide access to
cars, apartments, and contents of a chest in the
attic. Keys are genuinely usablefour-year-olds
can handle them. In the world of IT, something
the subject knows (e.g., a password or a secret
personal identification number [PIN]) is the
prominent mechanism.
The exclusiveness of access to an IT system
protected by a password rests on the security of
the password against guessing, leaving aside other
technical means by which it may or may not be
broken. From an information theoretic standpoint,
a uniformly and randomly chosen sequence of
letters and other symbols principally provides
the greatest security. However, such a random
sequence of unrelated symbols also is hard to

326

remember, a relation that is rooted in the limitation of humans cognitive capabilities.

As a remedy, a variety of strategies were
invented to construct passwords that humans
can memorize more easily without substantially
sacrificing the security of a password (e.g., passwords based on mnemonic phrases). For instance,
Yan, et al. (2000) conducted a study with 400
students on the effect of three forms of advice on
choosing passwords. They found, for example,
that passwords based on pass phrases were remembered as easily as naively chosen passwords,
while being as hard to crack as randomly chosen
passwords. Insight into human cognition also has
led to the investigation of alternatives such as
cognitive passwords (Zviran & Haga, 1990), word
associations (Smith, 1987), pass phrases (Spector
& Ginzberg, 1994), images (Dhamija & Perrig,
2000), or pass faces (Brostoff & Sasse, 2000).
Authentication in public places, as is the
case with automatic teller machines (ATM), has
turned out to be vulnerable to attacks, where
criminals obtain a users PIN by using cameras
or other methods of observation in so-called
shoulder-surfing attacks (Colville, 2003). In order
to obscure the numbers entered by the user and
thus hamper the recording of the necessary PIN,
several techniques have been proposed (Hopper
& Blum, 2001; Wilfong, 1997). Recently, Roth, et
al. (2004) suggested variants of cognitive trapdoor
games to protect users against shoulder-surfing
attacks. In this approach, the buttons on a PIN
pad are colored either black or white, and the user
has to decide whether the number to be entered
is in the black or white group. As the colors are
changing randomly, the user has to enter the same
number three to four times to complete an input. By
blurring the response set with so-called shadows,
this method can be made resistant against camera
attacks. Even though this approach is slightly
more complicated than the classical approach,
this technique has proven to be accepted by the
user in an experimental setting.

Human-Computer Interaction and Security

E-MAIL SECURITY
Before the middle of the 1970s, cryptography was
built entirely on symmetric ciphers. This meant
that in order for enciphered communication to
take place, a secret key needed to be exchanged
beforehand over a secure out-of-band channel.
One way of doing that was to send a trusted
courier to the party with whom one intended to
communicate securely. This procedure addressed
two important issues: the secret key exchange
and the implicit authentication of the exchanged
keys. Once established, the keys could be used to
secure communication against passive and active
attacks until the key was expected to become or
became compromised.
When asymmetric cryptography (Diffie &
Hellman, 1976; Rivest, Shamir, & Adleman,
1978) was invented in the 1970s, it tremendously
simplified that task of key exchange, and gave
birth to the concept of digital signatures. Asymmetric cryptography did not solve the problem
of authenticating keys per se. Although we now
can exchange keys securely in the clear, how
could one be certain that a key actually belonged
to the alleged sender? Toward a solution to this
problem, Loren Kohnfelder (1978) invented the
public key certificate, which is a public key and
an identity, signed together in a clever way with
the private key of a key introducer whom the
communicating parties need to trust. This idea
gave rise to the notion of a public key infrastructure (PKI). Some existing models of public key
infrastructures are the OpenPGP Web of Trust
model (RFC 2440) and the increasingly complex
ITU Recommendation X.509-based PKIX model
(RFC 3280) (Davis, 1996; Ellison, 1996, 1997;
Ellison & Schneier, 2000).
In applications such as electronic mail, building trust in certificates, exchanging keys, and
managing keys account for the majority of the
interactions and decisions that interfere with the
goal-oriented tasks of a user and that the user has
difficulty understanding (Davis, 1996; Gutmann,

2003; Whitten & Tygar, 1999). At the same time,

the majority of users only has limited understanding of the underlying trust models and concepts
(Davis, 1996) and a weak perception of threats
(see previous discussion). Consequently, they
avoid or improperly operate the security software
(Whitten & Tygar, 1999).
In the safe staging approach, security functions may be grouped into stages of increasing
complexity. A user may begin at a low stage
and progress to a higher stage once the user
understands and masters the security functions
at the lower stages. The safe-staging approach
was proposed by Whitten & Tygar (2003), who
also pioneered research on the usability of mail
security by analyzing users performances when
operating PGP (Whitten & Tygar, 1999).

SYSTEM SECURITY
Computer systems progressed from single user
systems and multi-user batch processing systems
to multi-user time-sharing systems, which brought
the requirement to sharing the system resources
and at the same time to tightly control the resource
allocation as well as the information flow within
the system. The principal approach to solving this
is to establish a verified supervisor software also
called the reference monitor (Anderson, 1972),
which controls all security-relevant aspects in
the system.
However, the Internet tremendously accelerated the production and distribution of software,
some of which may be of dubious origin. Additionally, the increasing amounts of so-called malware
that thrives on security flaws and programming
errors lead to a situation where the granularity
of access control in multi-user resource-sharing
systems is no longer sufficient to cope with the
imminent threats. Rather than separating user
domains, applications themselves increasingly
must be separated, even if they run on behalf of
the same user. A flaw in a Web browser should

327

Human-Computer Interaction and Security

not lead to a potential compromise of other applications and application data such as the users
e-mail client or word processor. Despite efforts
to provide solutions to such problems (Goldberg
et al., 1996) as well as the availability of off-theshelf environments in different flavors of Unix,
fine-grained application separation has not yet
been included as a standard feature of a COTS
operating system.
Even if such separation were available, malicious software may delude the user into believing,
for example, that a graphical user interface (GUI)
component of the malware belongs to a different
trusted application. One means of achieving this
is to mimic the visual appearance and responses
of the genuine application. One typical example
would be a fake login screen or window. Assurance that a certain GUI component actually
belongs to a particular application or the operating system component requires a trusted path
between the user and the system. For instance,
a secure attention key that cannot be intercepted
by the malware may switch to a secure login
window. While this functionally is available in
some COTS operating systems, current GUIs still
provide ample opportunity for disguise, a problem
that also is eminent on the Web (Felten, Balfanz,
Dean, & Wallach, 1997). One approach to solving
this problem for GUIs is to appropriately mark
windows so that they can be associated with their
parent application (Yee, 2002). One instance of a
research prototype windowing system designed
with such threats in mind is the EROS Trusted
Window System (Shapiro, Vanderburgh, Northup,
& Chizmadia, 2003).

FUTURE TRENDS
Mobile computing and the emergence of contextaware services progressively are integrating into
new and powerful services that hold the promise
of making life easier and safer. Contextual data

328

will help the user to configure and select the

services the user needs and even might elicit
support proactively. Far beyond that, Ambient
Intelligence (IST Advisory Group, 2003) is an
emerging vision of dynamically communicating
and cooperating appliances and devices in order to
provide an intelligent surrounding for tomorrows
citizens. Radio frequency identification transmitters (RFID) already have been discussed with
respect to their implications on privacy (Weis,
2004). Certainly, one persons contextual awareness is another persons lack of privacy (Hudson
& Smith, 1996). In the future, the development of
powerful and usable security concepts for applying
personal information to the context and vice versa
is one of the greatest challenges for todays security engineers and human-computer interaction
research (Ackerman, Darell, & Weitzner, 2001).
To accomplish this task seems crucial for future
acceptance and for chances of such technologies
without them becoming a privacy Chernobyl
(Agre, 1999).

CONCLUSION
The view of the user as the weakest link and
potential security danger finally has turned out
to be an obsolescent model. Security engineers
and perhaps, more importantly, those people who
are responsible for IT security have noticed that
working against the user will not do, and instead,
they have decided to work with and for the user.
During the past years, an increasing number of
research has focused on the issue of making security usable, addressing the traditional fields of
authentication, communication, and e-mail and
system security. This article has given a brief
overview of some of the work done so far. In order
to make information technology more secure,
the user is the central instance. The user must
be able to properly use the security mechanisms
provided. Therefore, understanding users needs

Human-Computer Interaction and Security

and identifying the reasons that technology fails

to convince users to employ it is crucial. The first
part of this article summarized some work done
by Dourish, Weirich, and Sasse that provided
important insights. But much work still has to
be done.
Future technology will build even more on
the integration and sharing of heterogeneous
sources of information and services. The tendency
toward distributed and location-based information infrastructures will lead to new security
problems. Feeling safe is an important aspect of
acceptance. The success of tomorrows systems
also will depend on the users feeling safe while
sharing information and using services, which
has already been shown during the first stage of
e-commerce. Therefore, making security usable
is an important aspect of making security safe.

mons.somewhere.com/rre/1999/RRE.notes.and.
recommenda14.html

REFERENCES

Davis, D. (1996). Compliance defects in public

key cryptography. Proceedings of the 6th USENIX
Security Symposium.

Ackerman, M. S., & Cranor, L. (1999). Privacy

critics: UI components to safeguard users privacy.
Proceedings of the ACM SIGCHI Conference on
Human Factors in Computing Systems, Pittsburgh,
Pennsylvania.
Ackerman, M. S., Cranor, L., & Reagle, J. (1999).
Privacy in e-commerce: Examining user scenarios and privacy preferences. Proceedings of
the ACM Conference in Electronic Commerce,
New York.
Ackerman, M. S., Darell, T., & Weitzner, D.
(2001). Privacy in context. Human-Computer
Interaction, 16(2), 167-176.
Adams, A., & Sasse, M. A. (1999). Users are
not the enemy: Why users compromise security
mechanisms and how to take remedial measures.
Communications of the ACM, 42(12), 40-46.
Agre, P. (1999). Red rock eater digestNotes and
recommendations. Retrieved from http://com-

Anderson, J. P. (1972). Computer security technology planning study (No. ESD-TR-73-51). Bedford,
MA: AFSC.
Brostoff, S., & Sasse, M. A. (2000). Are passfaces
more usable than passwords? A field trial investigation. In S. McDonald, Y. Waern, & G. Cockton
(Eds.), People and computers XIVUsability
or else! Proceedings of HCI 2000, Sunderland,
UK.
Brostoff, S., & Sasse, M. A. (2001). Safe and
sound: A safety-critical design approach to security. Proceedings of the New Security Paradigms
Workshop, Cloudcroft, New Mexico.
Colville, J. (2003). ATM scam netted 620,000
Australian. Risks Digest, 22, 85.

Dhamija, R., & Perrig, A. (2000). Deja vu: A user

study using images for authentication. Proceedings of the 9th USENIX Security Symposium,
Denver, Colorado.
Diffie, W., & Hellman, M. E. (1976). New directions in cryptography. IEEE Transactions on
Information Theory, 22(6), 644-654.
Dourish, P., Grinter, R. E., Dalal, B., Delgado de
la Flor, J., & Dalal, M. (2003). Security day-today: User strategies for managing security as
an everyday, practical problem (ISR Technical
Report No. UCI-ISR-05-5). Irvine: University of
California.
Ellison, C. (1996). Establishing identity without
certification authorities. Proceedings of the Sixth
USENIX Security Symposium.
Ellison, C., & Schneier, B. (2000). Ten risks of
PKI: What youre not being told about public

329

Human-Computer Interaction and Security

key infrastructure. Computer Security Journal,

16(1), 1-7.

tion infrastructure. Washington, DC: National

Academy Press.

Felten, E. W., Balfanz, D., Dean, D., & Wallach,

D.S. (1997). Web spoofing: An Internet con game
(technical report no. 540-96). Princeton, NJ:
Princeton University.

Kohnfelder, L. M. (1978). Towards a practical

public-key cryptosystem [B.S. Thesis]. Cambridge, MA: MIT.

Gerd tom Markotten, D. (2002). User-centered

security engineering. Proceedings of the 4th EurOpen/USENIX Conference, Helsinki, Finland.

Mitnick, K. D., Simon, W. L., & Wozniak, S.

(2002). The art of deception: Controlling the human element of security. New York: John Wiley
& Sons.

Goldberg, I., Wagner, D., Thomas, R., & Brewer,

E.A. (1996). A secure environment for untrusted
helper applications. Proceedings of the 6th Usenix
Security Symposium, San Jose, California.

Rivest, R.L., Shamir, A., & Adleman, L. (1978).

A method for obtaining digital signatures and
public-key cryptosystems. Communications of
the ACM, 21(2), 120-126.

Grinter, R., & Palen, L. (2002). Instant messaging

in teen life. Proceedings of the ACM Conference
on Computer-Supported Cooperative Work, New
Orleans, Louisiana.

Roth, V., Richter, K., & Freidinger, R. (2004). A

PIN-entry method resilient against shoulder surfing. Proceedings of the CCS, Washington, DC.

Gutmann, P. (2003). Plug-and-play PKI: A PKI

your mother can use. Proceedings of the 12th USENIX Security Symposium, Washington, DC.
Hopper, N. J., & Blum, M. (2001). Secure human identification protocols. Proceedings of the
ASIACRYPT.

Sasse, M. A. (2003). Computer security: Anatomy

of a usability disaster, and a plan for recovery.
Proceedings of the CHI2003 Workshop on Human-Computer Interaction and Security Systems,
Ft. Lauderdale, Florida.
Sasse, M. A., Brostoff, S., & Weirich, D. (2001).
Transforming the weakest link: A humancomputer interaction approach to usable and
effective security. BT Technology Journal, 19(3),
122-131.

Hudson, S. E., & Smith, I. (1996). Techniques for

addressing fundamental privacy and disruption
tradeoffs in awareness support systems. Proceedings of the ACM Conference on Computer
Supported Cooperative Work, Boston.

Schneier, B. (2000). Secrets and lies. New York:

Wiley & Sons.

Irvine, C., & Levin, T. (2000). Towards quality of secure service in a resource management
system benefit function. Proceedings of the 9th
Heterogeneous Computing Workshop.

Shapiro, J. S., Vanderburgh, J., Northup, E., &

Chizmadia, D. (2003). The EROS trusted window system (technical report no. SRL2003-05).
Baltimore: Johns Hopkins University.

IST Advisory Group. (2003). Ambient intelligence:

From vision to reality (draft). Brussels, Belgium:
The European Commission.

Smetters, D. K., & Grinter, R. E. (2002). Moving

from the design of usable security technologies to
the design of useful secure applications. Proceedings of the New Security Paradigms Workshop,
Virginia Beach.

Kent, S. (1997). More than screen deep: Towards

every citizen interfaces to the nations informa-

330

Human-Computer Interaction and Security

Smith, S. L. (1987). Authenticating users by word

association. Computers & Security, 6, 464-470.
Spector, Y., & Ginzberg, J. (1994). Pass-sentenceA new approach to computer code. Computers & Security, 13, 145-160.
Tom Markotten, D. G. (2002). User-centered security engineering. Proceedings of the 4th NordU
Conference, Helsinki, Finland,
Weirich, D., & Sasse, M. A. (2001). Pretty good
persuasion: A first step towards effective password security for the real world. Proceedings of
the New Security Paradigms Workshop 2001,
Cloudcroft, New Mexico.
Weis, S. A. (2004). RFID privacy workshop: Concerns, consensus, and questions. IEEE Security
and Privacy, 2(2), 34-36.
Whitten, A., & Tygar, J. D. (1999). Why Johnny
cant encrypt: A usability evaluation of PGP 5.0.
Proceedings of the 8th USENIX Security Symposium, Washington, DC.
Whitten, A., & Tygar, J. D. (2003). Safe staging for
computer security. Proceedings of the CHI2003
Workshop on Human-Computer Interaction and
Security Systems, Ft. Lauderdale, Florida.
Wilfong, G. T. (1997). Method and apparatus
for secure PIN entry (patent #5,940,511). Lucent
Technologies.
Yan, J., Blackwell, A., Anderson, R., & Grant, A.
(2000). The memorability and security of passwordsSome empirical results (technical report).
Cambridge, UK: Cambridge University.
Yee, K. -P. (2002). User interaction design for secure systems. Proceedings of the 4th International
Conference on Information and Communications
Security, Singapore.
Zviran, M., & Haga, W. J. (1990). Cognitive passwords: The key to easy access control. Computers
& Security, 9, 723-736.

KEY TERMS
Asymmetric Cryptography: A data encryption system that uses two separate but related
encryption keys. The private key is known only to
its owner, while the public key is made available in
a key repository or as part of a digital certificate.
Asymmetric cryptography is the basis of digital
signature systems.
Public Key Infrastructure (PKI): The public
infrastructure that administers, distributes, and
certifies electronic keys and certificates that are
used to authenticate identity and encrypt information. Generally speaking, PKI is a system of
digital certificates, certification authorities, and
registration authorities that authenticate and verify
the validity of the parties involved in electronic
transactions.
Shoulder Surfing: The practice of observing
persons while entering secret authentication information in order to obtain illegal access to money
or services. This often occurs in the context of
PIN numbers and banking transactions, where
shoulder surfing occurs together with the stealthy
duplication of credit or banking cards.
Social Engineering: The technique of exploiting the weakness of users rather than software by
convincing users to disclose secrets or passwords
by pretending to be authorized staff, network
administrator, or the like.
Spoofing: The technique of obtaining or mimicking a fake identity in the network. This can be
used for pretending to be a trustworthy Web site
and for motivating users (e.g., entering banking
information), pretending to be an authorized
instance that requests the users password, or
making users accept information that is believed
to come from a trusted instance.
Types of Authentication: Authentication
generally can be based on three types of informa-

331

Human-Computer Interaction and Security

tion: by some thing the user has (e.g., bank card,

key, etc.), by something the user knows (e.g.,
password, number, etc.), or by something the user

is (e.g., biometric methods like fingerprints, face

recognition, etc.).

This work was previously published in the Encyclopedia of Human Computer Interaction, edited by C. Ghaoui, pp. 287-294,
copyright 2006 by Information Science Reference, formerly known as Idea Group Reference (an imprint of IGI Global).

332

333

Chapter 1.26

Security for Ubiquitous

Computing
Tobias Straub
Fraunhofer Institute for Secure Information Technology, Germany
Andreas Heinemann
Technical University of Darmstadt, Germany

Abstract
Taking typical ubiquitous computing settings as a
starting point, this chapter motivates the need for
security. The reader will learn what makes security
challenging and what the risks predominant in
ubiquitous computing are. The major part of this
chapter is dedicated to the description of sample
solutions in order to illustrate the wealth of protection mechanisms. A background in IT security
is not required as this chapter is self-contained. A
brief introduction to the subject is given as well
as an overview of cryptographic tools.

Introduction
Mark Weisers vision of ubiquitous computing
(UC) raises many new security issues. Consider a
situation where a large number of UC peers interact
in a spontaneous and autonomous manner, without

previously known communication partners and

without a common security infrastructure. Such
extreme conditions make it difficult to apply established security methods that have been tailored
for classical information technology (IT). Virgil
Gligor emphasizes this by comparing the Internet
clich where processing is free and physically
protected, but communication is not with the new
clich about UC where neither processing nor
communication is free and physically protected
(Gligor, 2005).
This chapter gives a systematic introduction
into the field of security for UC. It is structured
as follows: first (the section Four UC Settings),
we illustrate the diversity of UC systems and applications in order to raise the readers awareness
of security questions. Our discussion focuses on
four representative UC settings. For each of them,
a characteristic application scenario is given. The
diversity of the settings, for example, concerning presumed knowledge about components or

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Security for Ubiquitous Computing

system complexity, has direct implications for

which mechanisms are appropriate to achieve a
desired IT security objective.
The section A Taxonomy of UC Security
starts with an introduction to basic terminology
explaining the objectives of IT security as well
as the threat model. UC security can be regarded
from two different viewpoints: The special characteristics of UC may lead to known and new
security risks, if not addressed appropriately. In
addition, limitations concerning resources and
infrastructure pose a number of challenges in
order to achieve desired security objectives.
A compact overview of cryptographic tools
suitable to enforce those objectives is given in
the section Overview of Cryptographic Tools.
A brief explanation of cryptographic primitives
like ciphers, hash functions, and signatures is
provided for the nonspecialist reader. A discussion
of the potential and limitations of cryptography
in UC concludes the section.
After having learned the particularities and
challenges of UC systems with respect to security,
we look at sample solutions to mitigate these limitations. A selection of elaborated approaches for
secure UC systems is given in the section Sample
Solutions. They tackle the issues of privacy and
availability as well as the establishment of secure
communication.
A brief summary and a list of references for
further reading conclude this chapter.

Four UC Settings
In order to pave the way for the development
of a systematic view on UC characteristics and
limitations in the section A Taxonomy of UC
Security, we sketch four representative settings.
Each setting exhibits one or more security-related
properties. They are termed mobile computing,
ad hoc interaction, smart spaces, and real-time
enterprises.

334

Mobile Computing
Mobile computing supports mobile users with
connectivity and access to services and backend
systems while being on the move. A synonymous
term is nomadic computing, emphasizing the
goal of providing a working environment more
or less equivalent to that of a desktop user. The
widespread availability of cellular networks and
802.11 WiFi allows a field worker to connect
to an arbitrary service on the Internet or to the
companys backend at almost any place and at
any time.
Mobile computing relies on a given infrastructure managed by a provider, for example, a cellular
network company. This fact has implications for
security: In order to access a service, a user needs
to register with a provider. Thus, the user group
is closed and the provider controls access to the
infrastructure. In addition, users are not able to
act in an anonymous manner.
Mobile devices can easily get lost, for example
left behind in the proverbial taxi (see http://www.
laptopical.com/laptops-lost-in-taxi.html). In case
of theft, an attacker might be able to impersonate
the legitimate device owner or learn her private
data like business contacts or personal email. This
physical threat is given whenever mobile devices
are considered.

Scenario 1: The Mobile Salesman

While on the road, a salesman needs to regularly download up-to-date client reports from
his companys databases. His laptop is equipped
with several wireless communication interfaces
which can be used to connect via different service
providers depending on what kind of service/infrastructure is available.
At the clients office, there is a WiFi network the
salesman can access. There are also some networked printers available for guests. However,
it is unclear to what extent the infrastructure can
be trusted.

Security for Ubiquitous Computing

This scenario raises three major issues:

Firstly, the protection of communication (from
the salesmans device to the company backend)
over potentially insecure channels. Secondly,
the secure storage of company internals on his
laptop. Secure tunneling, VPNs (virtual private
networks), and hard disk encryption are standard technologies in this field. The necessary
cryptographic tools are explained in the section
Overview of Cryptographic Tools. However,
the aspect of secure device association, discussed,
for example, byBalfanz, Smetters, Stewart,and
Wong(2002), is something new. Later in this
chapter, we answer the following questions: Is
there a way to securely send a confidential document over the air to a printer located in the office?
Does it help if the salesman selects a printer close
to him equipped with a secondary communication interface?

Ad Hoc Interaction
In contrast to mobile computing, the second setting
does not rely on an infrastructure provider. Instead
of that, UC devices build the infrastructure on
their own by establishing temporary, wireless, and
ad hoc communication links between them. On
the application layer, they expose a spontaneous
interaction behavior. A typical characteristic is
the lack of a central instance allowing or restricting participation. A priori, there are no given or
managed user groups; all devices are free to join.
Plus, users and devices might act anonymously.
Here we illustrate a collaboration scenario
based on spontaneous interaction. This type of
communication is typical for Opportunistic Networks which are discussed in depth in the chapter
Opportunistic Networks.

Scenario 2: Passive Collaboration in

Opportunistic Networks
In an Opportunistic Network, passers-by exchange
information, for example digital advertisements

(Straub& Heinemann, 2004), while being colocated. After an initial configuration, devices
interact autonomously and without users attention. Information dissemination is controlled by
profiles stored on the usersdevices. Such a profile
expresses a users interest in and knowledge about
some pieces of information to share.
The particularities of ad hoc interaction pose
numerous security challenges. On the one hand,
devices do not already know each other when they
start to communicate. On the other hand, personal
data is kept on the devices and exchanged with
strangers. As a consequence, privacy is inherently
at risk if systems are not designed carefully.

Smart Spaces
Smart spaces, which form our third UC setting,
emphasize user-friendliness and user empowerment as well as support for human interactions.
Interaction within a smart space happens in an
unobtrusive way. The use of contextual information (see chapter Context Models and Context
Awareness) plays also an important role here.
Sometimes, it is assumed that users carry some
type of digital identification and/or other devices
with or on them.
Due to the sensing and tracking capabilities
of a smart space, user privacy is at stake. Location privacy is an important field of UC research;
an overview is given by Grlach, Heinemann,
Terpstra,and Mhlhuser (2005). In addition, due
to the volatile nature of smart spaces, concepts
like trust (see Chapter 15) and reputation play
an important part in these kinds of applications.
We take patient monitoring in a hospital as an
example to illustrate a smart space.

Scenario 3: Patient Monitoring

In a hospital, all records of patients are digitally
stored and maintained in a central database.
Records are updated with the results of physical

335

Security for Ubiquitous Computing

examinations or continuous monitoring. Husemannand Nidd (2005) describe a middleware

capable of integrating a wide range of medical
analyzers that have a common wireless interface.
Consider a battery driven heartbeat monitor which
is attached to the body and sends measurements
to the database. The data can be used as well
for a patient surveillance system that triggers an
alarm in case of an anomaly.
This scenario raises a number of security issues: for new patients the heartbeat monitor has
to be unambiguously and securely associated
with their record. The data from the monitor
needs to go to the right record, communication
needs to be protected, and the correctness of the
data must be assured. In addition, after a patient
leaves the hospital, the heartbeat monitor needs
to be detached from the digital record for reuse.
The resurrecting duckling(Stajano& Anderson,
1999) security policy framework helps in this
setting. It is described in the section Sample
Solutions.

Real-Time Enterprises
Real-time enterprises, which are defined in the
preface of this book, are an effort to leverage UC
technology and methods within enterprises. A
driving force behind these efforts is the goal of having immediate access to comprehensive and up-todate information about processes and procedures
within an enterprise. This allows management to
react very flexibly to variances in the market and
to increase customer support and satisfaction. For
example, an enterprise with detailed real-time
information on all production steps, including
delivery statuses of subcontractors, can provide
a customer with very accurate information on
when an order will be delivered.

336

Scenario 4: RFID-Based Warehouse

Management
Radio frequency identification (RFID) offers a
variety of opportunities in tracking goods (see
e.g.,Fleisch& Mattern, 2005). Suppose all goods
stocked at a warehouse are tagged with an RFID
transponder. With the corresponding readers integrated into storage racks, the process of stocktaking can be completely automated and inventory
information is available in real-time.
The use of RFID allows for more efficient
stocking, product locating, and product theft
protection, to name a few. The deployment of
tags and readers within a company gives rise to
security questions, for instance, those of industrial
espionage. A competitor, equipped with a mobile RFID reader, might be able to derive useful
information about a company, by reading their
product tags. RFID tags should also be looked
at from a privacy point of view as theyunder
certain conditionsmay allow the surveillance
and tracking of humans.
RFID privacy is currently also a hot topic in
the context of machine-readable travel documents.
Privacy is discussed with respect to biometric data
like photos and fingerprints stored on the chip.
Another question is whether ID cards can be used
to create movement profiles of their holder. The
reader is referred toKnospeand Pohl (2004) for
an overview of RFID security. We come back to
this issue in the section Sample Solutions.

A Taxonomy of UC Security
The beginning of this section provides a compact
introduction to IT security by explaining the
common objectives and the threats to them. We
then formulate two views on UC security issues.
In section First View: UC Characteristics and

Security for Ubiquitous Computing

Associated Risks," we explain which and how

typical characteristics of UC lead to well-known,
but also new risks that demand appropriate countermeasures. UC limitations concerning resources
and deployed infrastructures are in focus of the
second view in the section Second View: UC
Limitations and Associated Challenges. These
limitations give rise to security challenges that
require novel methods, some of which are presented in the section Sample Solutions.

Basic Terminology and Objectives of

IT Security
Security of IT systems is usually discussed in the
following way: First, assets (and their respective
value) are identified. The notion of assets covers
data as well as hardware. With respect to our
four scenarios, data that has to be protected, for
instance, comprises:

Confidential documents (Scenario1),

An individuals habits and preferences
(Scenario2),
Medical information (Scenario3),
The stock list at a warehouse (Scenario4).

When speaking of the protection of data, we

mean that particular attributes are preserved. In
the information security community the mnemonics CIA or CIAA are often used to refer to the
following fundamental protection objectives:

Confidentiality (C): Refers to the aim of

keeping pieces of information secret from
unauthorized access.
Integrity (I): Is the requirement that data is
safe from changes, be it either accidentally
or deliberately.
Authenticity (A): Concerns itself with the
genuineness of messages or the identity of
entities in a networked system.

These objectives can be achieved by cryptography as we will see below. In this respect, the
next objective is different as it typically requires
noncryptographic efforts as well.

Availability (A): Means the provisioning of

a systems services to its users in a reliable
way.

An attacker may try to prevent legitimate use

in a so-called denial-of-service (DoS) attack. Demanding a large portion of (computational and/or
network) resources is a strategy to slow down the
system. Another flavor of DoS is the blocking of
communication links between legitimate parties.
Designing a UC system for redundancy and diversity is a standard way towards achieving availability (see e.g.,Vogt, 2005). This can be done,
for instance, by using peer-to-peer networks, the
replication of information, or distributed computing facilities. Intrusion-detection systems, which
are a common protection technology for computer
systems in an organization, are starting to attract
attention in the UC community(Robinson, 2005).
Bahl,
Chandra, Padhye, Ravindranath, Singh,
Wolman
(2006) describe a system to fight DoS
attacks on corporate wireless networks and to
detect malicious mobile devices.

Threat Modeling
Having discussed basic terminology and objectives, we now turn to a common abstract network
model to describe security threats: Two parties
exchange messages over a channel to which an
attacker has access, too. This notion captures
any kind of attackers (computer systems, individuals, organizations) and is independent of the
data transport medium itself. There is no need
to differentiate between the transmission and
the storage of data as the latter can be seen as a
special case of the model.

337

Security for Ubiquitous Computing

We follow the security communitys convention in using the names Alice and Bob for the
legitimate actors (instead of simply numbering
them serially) and in calling the attacker Mallory
(malicious). Mallory may change data Alice
sends to Bob, may generate her own messages
under the name of another person, or simply
eavesdrop on their connection. An attack of the
latter kind is called passive while the other two are
called active. A passive attacker can compromise
confidentiality at best, but an active onewho is
also called man-in-the-middle (MITM)targets
at all CIAA goals. Acting as a MITM, Mallory
sits in between the communication link, making
Alice believe she is Bob and spoofs Bob into
believing she is Alice.
Attacks are typically directed toward more
than one of the before-mentioned security objectives. For instance, Mallory may launch a DoS
attack in order to paralyze the systems defense
mechanisms.
The risk that a system may become compromised is proportional to both its vulnerabilities
and the threats acting upon it. Risks have to be
identified and rated in the light of the corresponding assets value. Threat perception can be very
subjective: While one individual cares about data
emitted by her UC device, which might be linked
back to her, another does not. Furthermore, not all
objectives are equally important in practice. For
instance, a bank primarily has a vital interest in
keeping its account datas integrity, but a research
lab emphasizes confidentiality. The choice of algorithms and security also takes into consideration
the (assumed) attackers strategy and resources,
especially computational power.

First View: UC Characteristics and

Associated Risks
In order to put the UC vision to work, most scenarios, including the ones described in the last
section, rely on wireless communication between
nodes. Communication might happen in an ad

338

hoc manner as illustrated in the ad hoc interaction setting; other applications might ask for a
wireless multi-hop communication (see sensor
networks in the chapter Wireless and Mobile
Communication). Wireless communication
makes eavesdropping very easy as radio signal
are usually emitted in all directions. They can
be received by anyone in the senders vicinity
without her noticing it. MITM attacks are feasible
in the case of multi-hop wireless communication and they do not even require the attacker
to be physically close to the victim. In addition,
wireless ad hoc communication bears the risk of
impersonation, that is, an attacker might be able
to steal a peers credential by eavesdropping and
use it to access a certain service.
The pervasive nature of UC introduces even
more risks. Sensor nodes or RFID tags for example, are physically exposed, unmanaged, and
unsupervised. This bears the risks of device
and/or data theft as well as device manipulation.
As a consequence, access to the data stored on
the device must be carefully protected in order to
prevent identity theft(Eckert, 2005). The BlackBerry PDA for instance is a centrally manageable
system that supports a remote kill command to
erase data on a stolen device(Greiner, 2006).
UC devices are often battery-powered, which
allows for a DoS attack called sleep deprivation
torture(Stajano& Anderson, 1999): By constantly sending requests to a device, an attacker
can quickly drain its battery, thus rendering the
device useless. Last but not least, the UC settings
offer the capability of tracing objects or humans.
This feature is useful in many UC applications like
the real-time enterprise setting, but may violate
users privacy. Networked sensors, for instance,
may gather a good deal of personal information
that can be used to build user profiles.
The characteristics in UC and their corresponding risks are summarized in Table 1. In the
section Overview of Cryptographic Tools we
cover appropriate countermeasures.

Security for Ubiquitous Computing

Table 1. First view: Characteristics and risks

characteristics
communication
pervasive nature

wireless
ad hoc
multi-hop
physical exposure
battery-powered
traceability

Second View: UC Limitations and

Associated Challenges
Our second view on UC security concerns resource and infrastructure limitations of UC and
their corresponding challenges. The functionality provided by UC devices varies widely and is
limited by a number of parameters. These include
device capabilitiesconcerning, for example,
memory capacity, energy supply, CPU power,
and the user interfaceor connectivity which is
characterized among others by network coverage,
bandwidth, environmental conditions (like shielding or speed in case of moving devices).
Since a UC device can be less powerful that
a desktop computer, it is important to use appropriate, that is, lightweight, security mechanisms.
Otherwise, wasting memory and energy resources
would achieve the exact opposite effect of keeping
the system safe. Observe that a slow CPU might
take significant time to carry out sophisticated
cryptographic operations which may negatively
affect the user experience. User interface capabilities may also limit the choice of appropriate
methods, too: For example, a mobile device with a
voice only user interface such as described by Aitenbichler, Kangasharju,and Mhlhuser (2004),
demands different authentication methods to a

risks
eavesdropping
impersonation
man-in-the-middle attacks
device/data theft, manipulation
sleep deprivation torture
privacy violation

device that comes with a keyboard. At the bottom

end, very small and feeblish UC devices like passive RFID chips are found. These devices do not
even come with a user interface at all. A limited
user interface makes it difficult to establish what
is called a trusted path. This is a mechanism to
ensure that users can assert that they are interacting with a genuine system or program instead of
one controlled by an attacker (for instance when
asked for a password).
The absence of a centralized authority is typical
for UC systems like the ad hoc interaction setting
presented in section Ad Hoc Interaction. This
makes it hard to build an infrastructure where
UC peers can authenticate each other in order to
communicate securely. Another consequence is
the lack of common support for trust and policy
decision. The following chapter discusses appropriate countermeasures for this issue. Alternative
methods for entity authentication based on outof-band communication schemes are the subject
of section Out-of-Band Channels.
The limitations corresponding to UC security
and their induced challenges are summarized
in Table2. Selected methods that address these
challenges are discussed in section Sample
Solutions.

339

Security for Ubiquitous Computing

Table 2. Second view: Limitations and challenges

Limitations
Infrastructure

lack of centralized authority

Resource

limited CPU power, few/no

memory, limited power supply
user interface limitations

Overview of Cryptographic
Tools
This section provides a compact overview of
cryptographic tools which are suitable to enforce
the security objectives introduced in the section
A Taxonomy of UC Security. It serves as a basis for the following discussion of the solutions
addressing UC characteristics and challenges.
Cryptographic primitives, namely symmetric and
public key cryptosystems, hash functions, and
authentication schemes, are the building blocks of
more complex techniques and protocols. We also
introduce some general rules of cryptography and
reason about security parameters. At the end of this
section we turn to the potentials and limitations
of cryptography in a UC setting. Readers with
previous knowledge in the area of cryptography
may directly proceed to section Potential and
Limitations of Cryptography in UC.

Symmetric Cryptosystems
Messages can be encrypted to prevent disclosure
of their content and ensure confidentiality. Encryption is a transformation that renders plaintext, that
is, the original message in its readable form, into
ciphertext which is unintelligible to an outsider.
The reverse transformation is called decryption.
In order to give Alice and Bob a competitive
edge, there must be some information Mallory

340

Challenges
entity authentication, policy
decision
algorithm implementation,
protocol design
trusted path

does not have. The encryption and decryption

function of all modern cryptosystems (or ciphers)
are parameterized by keys. A key is a short fixedlength piece of data (while the plaintext may be
of arbitrary length). It has to be kept secret from
Mallory as only the correct key allows successful
decryption of the ciphertext.
As early as the 19th century, cryptographer
Kerckhoff claimed the following important law:
A cryptosystems strength should not be based on
the assumption that its algorithm is kept secret,
but only on the attackers uncertainty regarding
the key. An elementary security requirement is
the fact that it should be unfeasible for Mallory to
simply try out all possible keys by a brute-force
attack. This imposes a lower bound on the cardinality of the set of all possible keys and demands
that keys be generated randomly.
The distinction whether the same key is used
for encryption and decryption or not may seem
marginal at first sight, but is in fact crucial. Symmetric cryptosystems require two or more communication partners to know the same shared
key. Modern symmetric cryptosystems use keys
of bitlength 112, 168 (3DES) or 128, 192, 256
(AES). As a rule of thumb, keys shorter than
80 bits should be used with great care. A good
place to look up reliable key lengths based on
cryptographic analysis is the Web site http://www.
keylength.com.

Security for Ubiquitous Computing

On this occasion, we exhibit a problem common to all symmetric ciphers: The use of shared
keys implies a secure key distribution step before
the actual communication takes place. Key distribution is often a problem as it needs out-of-band
mechanisms, that is, an additional communication channel not accessible to the attacker. We
will come back to this issue in the context of UC
in the section Out-of-Band Channels. Another
downside of shared keys is that the risk grows
proportional to the number of group members.
If one device falls into Mallorys hands, she
will be able to read all messages of the group. A
real-world example for the use of shared keys is
802.11 WEP (Wired Equivalent Privacy, see
for exampleEckert, 2006).

Asymmetric Cryptosystems and PKI

The key distribution problem can be avoided
by using asymmetric cryptosystems, which are
synonymously called public key cryptosystems. In
this case, encryption and decryption use different
keys, such that these functions can be effectively
separated. The respective keys are denoted public
key and private key. Such a key pair has to satisfy
certain mathematical conditions depending on the
particular cryptosystem. Bob solely has to keep his
private key secret, but can make the corresponding
public key available to Alice or everybody else in
the worldincluding Malloryas it is infeasible
to compute the private key from the public key.
Public key encryption is thus an example of a
so-called one-way function since the ciphertext
can be computed efficiently, while inversion is
infeasible. More specifically, public key encryption falls into the category of one-way functions
with trapdoor as knowledge of the private key
allows the inversion of the mapping. We will see
below that asymmetry is an important pillar of
modern cryptography.
The RSA cryptosystem, named after R. Rivest,
A. Shamir,and L. Adleman (1978), is the most
prominent example of an asymmetric cryptosys-

tem. Typical key lengths are 1024, 2048, or 4096

bit. Its security is based on the presumed intractability of the factorization of large composite
numbers. The ElGamal cryptosystem in contrast
is based on the problem of computing discrete
logarithms in finite cyclic groups (a typical group
size is 2160 or 2230). A standard construction for
ElGamal is to use a subgroup of the multiplicative
is a prime (typically 1024
group of p, where p
or 2048 bits in length). Elliptic curves (EC) over
finite fields can also serve as a mathematical structure for ElGamal encryption (see e.g.,Hankerson,
Menezes,and Vanstone, 2004) for an introduction
to EC cryptography).
Observe that public key cryptography alleviates the problem of confidential secret key
distribution for the price of authentic public key
distribution in the following way: Alice who wants
to send a private message to Bob needs to have
his genuine public key KB. If Mallory manages
to foist her key KM on Alice, the message would
be encrypted for her and not for Bob. Digital
certificates are a common solution to this problem. A certificate is a statement linking a public
key to an identity (e.g., a users name, a device
address, or a pseudonym). In order to prevent
certificates from being forged they are digitally
signed (see the following section for an explanation
of digital signatures). Certificates form the basis
of a so-called public key infrastructure (PKI),
an environment where the authenticity of public
keys is assured.

Hash Functions
In order to ensure data integrity, some kind of
redundancy has to be added to the payload. A
modification detection code (MDC) is a hash
algorithm, that is, a function that compresses
bitstrings of arbitrary finite length to bitstrings
of fixed length. State-of-the-art algorithms like
RIPEMD-160(Dobbertin, Bosselaers,& Preneel,
1996) or SHA-1(Eastlake& Jones, 2001) produce
outputs with a length of 160 bit, which should be

341

Security for Ubiquitous Computing

considered the lower bound due to security reasons. To be useful for integrity protection, a hash
function has to be 2nd preimage resistant: Given
an input x that hashes to h(x), an attacker must not
be able to find a value y x such that h(y) = h(x).
Such a pair of different inputs that result in the
same hash value is called a collision. Collisions
always exist due to the pigeon-hole principle since
the co-domain is finite while the domain is infinite.
Cryptographic hash functions must be collisionresistant, that is, finding concrete collisions must
be computationally infeasible. This allows us to
protect data x of arbitrary length against modifications by storing h(x), a small piece of information,
in a safe place. A widespread application of this
mechanism is integrity-protected software distribution via download mirror sites on the Internet.
Flipping a single bit in x results in a hash value
that differs in about half of the bits from h(x) if h
is one of the common hash functions.
Message authentication codes (MACs) are
hash functions that are additionally parameterized
with a secret key. Assume that Alice and Bob share
a key k, which for instance has been established in
a key exchange scheme as explained below. Alice
adds a MAC hk(x) to her message x to Bob. From
the pair (x,hk(x)) Bob can tell that the message
came from her since it is infeasible for Mallory
to create a pair (y,hk(y)) without knowing k. Remember that a MDC does not have this property,
since Mallory may change x to y and compute the
(unkeyed) hash function h(y). MACs in turn not
only provide data origin authentication, but also
integrity as Bob could detect a modification x
x because the received hk(x) would not match the
value hk(x) he computed himself. Each MDC h
can be extended to a MAC in the following way:
On input x, compute h(k || p1 || h(k || p2 || x)) where
k is the key, p1,p2 are constant padding strings,
and || denotes concatenation. This generic construction is called HMAC (hash-based MAC),
see Krawczyk, Bellare,and Canetti, 1997 for
implementation and security details.

342

Digital Signatures
MACs are useful for guaranteeing the authenticity
of a communication link. However, they do not
provide a transferable proof of authorship: As the
parties at both ends of the channel share knowledge of the key, a MACed message (x,hk(x))
could originate from either of them. As a matter
of fact, a third party cannot deduce its author.
Transferable proofs of authorship are required to
model digital workflows, for exampleorders in
electronic commerce that must be verifiable by
multiple parties (possibly including a judge in
order to arbitrate in a dispute). Digital signature
schemes are a means to provide such evidence.
They are also implemented with public key
cryptography. Signature verification is a one-way
function with trapdoor, referring to the ability to
create signatures.
The RSA signature scheme is based on the
same mathematics as the RSA cryptosystem.
Signing corresponds to decryption while verification corresponds to encryption. Digital Signature
Algorithm (DSA) and its EC-based variant ECDSA are alternative schemes which are not yet
as prevalent as RSA. EC-DSA is more efficiently
computable than RSA and uses shorter keys for
comparable strength.
Signatures are, in fact, computed over the hash
value h(x), not the message x itself. One reason
is the reduction of computational costs associated with public key cryptography. Secondly, the
one-way property of the hash function prevents
so-called existential forgery of signatures (see
e.g., Buchmann, 2004). However, h must be collision-resistant as in the case where h(x) = h(x), a
signature of x is always a signature of x, too.
Messages that comprise declarations of intent
should be legally binding for the originator. On the
one hand, this implies that the originator commits
himself to the content in a way that he cannot
deny his intent later. On the other hand, no one
other than the legitimate person should be able to

Security for Ubiquitous Computing

make a commitment in his name. This property is

called non-repudiation. Non-repudiation can be
achieved with the help of dedicated hardware, for
examplea smartcard or TPM (Trusted Platform
Module, see e.g., Eckert, 2006) chip that keeps the
signing key under the sole control of its owner.
Observe that deniability, the opposite of nonrepudiation, may indeed be a desirable security
goal sometimes, for instance in order to cope
with censorship in UC networks. In such a setting, MACing should be preferred to signing of
messages.

Hybrid Encryption and Key

Exchange
In practice, asymmetric and symmetric cryptosystems are combined in hybrid encryption in
the following way: Symmetric cryptosystems
typically have a much higher throughput compared to public key systems with a comparable
security level. Therefore Alice first encrypts the
plaintext with a symmetric cryptosystem using
a random, ephemeral session or transaction key.
Then she transmits the symmetric key to Bob in a
confidential way by encrypting it with his public
key (if the message has multiple recipients, they
all obtain the same ciphertext accompanied by
the same symmetric key encrypted with their
individual public key). The mechanism of distributing key material is called key exchange. The
term key agreement is often used synonymously,
but may sometimes emphasize the fact that both
parties contribute to the generation of the key.
The Diffie-Hellman (DH) protocol is the most
famous interactive key exchange scheme. It allows two parties linked by an authentic, but not
necessarily private, channel to agree on a common
secretwhich seems counter-intuitive at first
sight. Note that unauthenticated DH is inherently
prone to MITM attacks. Authenticated DH can
be implemented by having each party digitally
sign its protocol messages. The mathematics and

security properties of DH are closely related to

the ElGamal cryptosystem.
An authentic key agreement protocol like the
DH scheme allows parties to achieve forward
secrecy, that is, to limit the possible impact of a
broken key: If a long-term signing key falls into
Mallorys hand, this does not leak the plaintext of
past communications Mallory may have recorded.
On the other hand, a broken ephemeral symmetric
key only affects the particular messages encrypted
with it. Forward secrecy allows the parties to use
relatively short symmetric keys provided they
are changed frequently. When this trade-off is
made in a UC scenario, the cost reduction for
symmetric encryption due to smaller key sizes
has to be balanced against the overhead for key
exchange.

Potential and Limitations of

Cryptography in UC
Having outlined the general tools of cryptography,
we conclude this section by a discussion of computational costs. Since energy consumption is a
serious issue in UC, we direct the readers attention
to Table3. These figures are the outcomes of experiments with a 206MHz Compaq iPAQ H3670
described by Potlapally, Ravi, Raghunathan,and
Jha (2003). The table shows the setup and current
costs for the families of algorithms we explained
above. Observe that the costs of hashing and symmetric encryption are given in J, whereas mJ is
the unit for digital signatures.
Due to its output size of only 128-bit, we
included the hash function MD5 for comparison
although its use is no longer recommended. The
figures for the AES algorithm refer to the standard operation mode CBC (see e.g.,Buchmann,
2004). AES is a so-called iterative block cipher.
Its computation is organized in multiple rounds
implementing the same subroutines, but with
different round keys. The round keys are derived
from the actual (128, 192, or 256 bit) key. This
process is called key scheduling.

343

Security for Ubiquitous Computing

Table 3. Energy costs

Algorithms

Setup Costs

Hash Functions
MD5
SHA-1
HMAC

Current Costs
Hashing (J/Byte)
0.59
0.75
1.16

Symmetric Encryption
AES (128 bit)
AES (192 bit)
AES (256 bit)

Key Scheduling (J)

7.83
7.87
9.92

Encryption (J/Byte)
1.62
2.08
2.29

Digital Signatures
RSA (1024 bit)
DSA (1024 bit)
EC-DSA (1024 bit)

Key Generation (mJ)

270.13
293.20
226.65

Sign (mJ)
546.5
313.6
134.2

Key Agreement
DH (1024 bit)
EC-DH (163 bit)
DH (512 bit)

Key Generation (mJ)

875.96
276.70
202.56

Key Exchange (mJ)

1046.5
163.5
159.6

The table shows the signature algorithms

RSA, DSA, and EC-DSA with their respective
security parameters. These three combinations are
considered of having comparable strength. While
the costs for the respective setup do not differ too
much, the costs for signing and signature verification vary significantly. Obviously, EC-DSA
is a better choice than DSA as each operation is
cheaper. However, the choice between RSA and
EC-DSA is determined by the expected number
of operations of the UC system. If the number
of signature verifications compared to signature
generations dominates, the RSA signature scheme
is clearly favorable compared to EC-DSA.
However, concerning key exchange, EC-DH
definitely outperforms standard DH (where

344

Verify (mJ)
15.97
338.02
196.23

computations are done in p) with the same

security level. In order to achieve roughly the same
costs as EC-DH, DH keys must be restricted to
512bits in length.
We have not mentioned asymmetric encryption so far. Since the same mathematical formula
is used for the RSA cryptosystem and the RSA
signature scheme, the corresponding numbers
can be looked up in the table. This shows that
in some UC situations with constrained devices,
asymmetric cryptosystems may quickly become impractical while symmetric encryption
is cheaper by magnitudes. Contemporary public
key algorithms are not very well suited for the
capabilities and needs of UC (Buchmann, May,&
Vollmer, 2006). A back-of-the-envelope calcula-

Security for Ubiquitous Computing

tion shows that a pocket PCs battery with a 1500

mAh capacity and a 5V voltage would have lost
20% of its charge after 5000 executions of a DH
protocol or 10000 RSA signatures.
These facts illustrate that it is worthwhile to
precisely analyze the security demands of a UC
system since choices on the cryptographic layer
directly affect energy consumption. A comprehensive examination of energy costs is provided
by Seys (2006). Lightweight UC cryptography
has also been the subject of recent research activities, for example, by the ECRYPT Network
of Excellence in Cryptology (http://www.ecrypt.
eu.org). This issue will surely gain even more
momentum with the deployment of RFID tags
for industrial applications as we already sketched
in Scenario4.
Lightweight cryptography has several aspects.
On the one hand, it comprises the design choice of
algorithms and efficient engineering to implement
them in hard- and software. On the other hand, it
requires a re-assessment of the threat model by
assuming a less powerful opponent and ruling
out some particular attacks which are considered
less likely. This may imply that weaker security
mechanisms, like weak authentication (discussed
next), or lower security parameters are acceptable
for low-value transactions (cf.the discussion of
forward secrecy).
Cryptography offers a wealth of ideas that
can be applied to UC: For instance, traditional
protocols can be enhanced to provide privacy
protection using mechanisms like the private
authentication scheme explained below. We also
point out that threshold cryptography (see for
example, Gemmell (1997) for an introduction) is
a valuable tool for UC that merits further study.
For instance, it can be used in sensor networks
that are prone to so-called Byzantine faults, that
is,there is a risk that devices fall into the attackers
hands and then act incorrectly. Provided that the
number of such Byzantine nodes is lower than
an adjustable bound, the attacker learns nothing
about a secret (e.g., a common key for signing

responses) shared among the nodes. Interestingly, this statement holds in an unconditional
sense, that is, for an adversary with unlimited
computational power.

Sample Solutions
In this section five example solutions for secure
UC systems are described. References to the
literature and UC applications as well as links
to our scenarios are given in the text.
At first, we present two strategies of privacyenhancing technologies: the technical concept of
anonymity and a policy-driven approach. While
anonymity is favorable from a privacy perspective,
it nevertheless leads to new threats as malicious
parties may benefit from it, too. One such threat
is the so-called Sybil attack which is discussed in
the section Fighting Sybil and DoS Attacks. We
explain the technique of proof-of-work as a countermeasure which can also be used as a defense
against DoS. A typical situation in UC is the lack
of a priori trust relations among the participants.
In order to cope with such a setting, a secure
communication channel has to be bootstrapped
somehow. The section Bootstrapping Secure
Communication treats this aspect from a theoretical point of view. Above all, the resurrecting
duckling security policy model is explained here.
The following section gives several real-world
examples for out-of-band channels used to set
up secure UC communication. Finally we touch
on the hot topic of RFID security and discuss
in particular privacy concerns and protection
mechanisms of electronic travel documents.
We deliberately omitted reputation systems
from the set of examples, as this topic is the
subject of the following chapter. Trusted Computing is also an interesting direction of current
UC research, but would go beyond the scope of
this chapter. We refer the interested reader to the
work of Hohland Zugenmaier (2005).

345

Security for Ubiquitous Computing

Privacy-Enhancing Technologies

Blurring Data

A large number of UC environments rely on sensing and tracking technologies of users and devices
in order to carry out their tasks. For example, in
order to provide location-based services, a users
position has to be determined beforehand. But even
without a given UC service in place, the fact that
most communication takes place over a wireless
link opens the door for attacks based on traffic
analysis. All communication patterns that can be
successfully linked to a human jeopardize a users
privacy. In this sense, the notion of confidentiality
has to be extended to message source confidentiality and/or message destination confidentiality.
This property of hiding the fact that particular
communication relationships exist at all is also
called sender or recipient anonymity, respectively. Anonymity is significantly more difficult
to achieve than message content confidentiality,
as it requires careful design of communication
protocols. Examples in the Internet setting include
mix networks(Chaum, 1981), onion routing (see
http://tor.eff.org for a popular implementation), or
anonymous re-mailers. Those protocols however
cannot be directly adapted to the UC world, since
they typically make assumptions about connectivity and resources. A first step in this direction is the
idea of Mist Routing(Al-Muhtadi, Campbell,
Kapadia, Mickunas,& Yi, 2002). Mix Zones
are a variant of the concept of mix networks to
locations (seeBeresford& Stajano, 2004).
As mentioned before, anonymity is of particular interest in UC when it comes to achieving
location privacy, that is, to prevent people bearing
devices emitting radio signals from being tracked
without noticing it. We discuss two different
strategies to preserve a users privacy. First, a
number of technical means are presented in order
to blur data that could later be used to identify
a person/subject. Second, the policy-based approach to achieve user privacy as proposed by
Langheinrich (2002).

Gruteser and Grunwald (2003) propose a middleware-based approach to blur location information from clients before passing the location
information onwards to a location based service
provider. They assume that clients communicate
their position as very precise location information to a location server. Position is determined
on the client itself, for example, via GPS or by
the wireless service provider through signal
triangulation. Location-based service providers
access location information through the location
server (see Figure 1).
A client specifies that he wants to be indistinguishable from at least k - 1 other clients within
a given area and time frame. In other words,
the clients want to stay k-anonymous. To reach
k-anonymity, their algorithm, an adaptive form
of a quadtree-based algorithm, adjusts the resolution of location information along spatial and
temporal dimensions in order to meet the specified
anonymity constraints, say k.
Providing anonymity requires careful system
design as identification may happen on each
network layer(Avoine& Oechslin, 2005). As a

346

Figure 1. Anonymizing middleware for locationbased services

Security for Ubiquitous Computing

practical consequence, user-controlled pseudonyms on the application layer are pretty useless
if the devices itself can be re-identified. This is
the case when static IP or Media Access Control
addresses are used. A proposal for temporary
addresses for anonymity on the data link layer is
made inOrava, Haverinen,and Honkanen (2002).
UC environments like scenario2 are particularly
suitable for addresses that are picked at random,
since the number of devices that are connected
to each other at the same time is by magnitudes
smaller than in an Internet setting. This guarantees, with a high probability, that device addresses
do not coincide.

Private Authentication
The following example illustrates how cryptographic protocols can be modified to provide
location privacy on the application layer. In typical
schemes for mutual authentication like the one
used in TLS/SSL(Dierks& Rescorla, 2006),
the parties reveal their identity during the handshake. As a consequence, Mallory may pretend
to be a legitimate party and start a handshake in
order to see who is in her vicinity. She may also
eavesdrop while Alice and Bob are authenticating
each other and learn about their communication
relation. Private authentication tackles this problem. The approach of Abadi (2002) allows Alice
to prove her identity to some self-chosen set S of
peers in order to establish a private and mutually
authenticated channel. Entities outside of S cannot detect Alices presence while entities inside
S do not learn more than their own membership.
Without loss of generalization, we restrict ourselves to the case where Bob is the only member
of S. If the set S contains more than one element,
Alice simply goes through parallel executions of
the protocol for each element. In order to start a
communication with Bob, Alice broadcasts the
plaintext hello accompanied with

(1)

Here KB is Bobs public key used for encryption, (KA ,K-1A) denotes her own key pair, K a
session key, and t is a timestamp. As Bob is the
only entity who knows the private key K-1B , he is
able to extract from c the senders identity (which
is unforgeable due to the digital signature). Assuming that Alice is on his whitelist of peers, he
answers with a message protected by K. Bobs
presence in turn cannot be detected provided that
the cryptosystem is which-key concealing (i.e.,KB
cannot be deduced from c) and Bob ensures ts
recency. Otherwise Mallory could mount a replay
attack by sending c and checking whether there
is a corresponding answer. Abadi (2002) also
describes a second protocol without timestamps
at the price of additional communication rounds.
This variant might be useful in cases where one
cannot assume synchronized clocks.

A Policy-Based Mechanism
Langheinrich (2002) proposes pawS, a system that
provides users with a privacy enabling technology. This approach is based on the Platform for
Privacy Preferences Project (P3P, seehttp://www.
w3.org/TR/P3P/), a framework which enables
the encoding of privacy policies into machinereadable XML. Making use of a trusted device,
the so-called privacy assistant (PA), the user
negotiates his privacy preferences with the UC
environment. For this purpose, the PA is able
to detect a privacy beacon upon entering a UC
environment, for example, a smart space. The
privacy beacon announces the available services,
for example, a printer, or a video camera, with
a reference to their data collection capabilities
and policies. The PA in turn contacts the users
personal privacy proxy located on the Internet,
which contacts the corresponding service proxies

347

Security for Ubiquitous Computing

at their advertised addresses (through the privacy

beacon) and asks for the privacy policies of the
services located in the smart space. These privacy
policies are compared to the users privacy preferences and a result may be to decline usage of a
tracking service, which will result in disabling a
video camera. Further, the work of Langheinrich
(2001) includes four design principles for UC
environments, the main ideas of which we sum
up as follows:

348

Notice: Since future UC environments will

be ideally suited for unnoticed operation,
monitoring and tracking, the author proposes
some kind of announcement mechanism
that allows users to notice the data collection capabilities in their environment.
This announcement mechanism should be
standardized and agreed on.
Choice and consent: To preserve privacy,
a future UC system should offer users the
choice of allowing or denying any kind of
data collection. A system should respect this
choice and operate only on user consent.
Since a user will not always explicitly express
her will, her choice and the derived consent
should be expressed in machine-readable
policies
Proximity and locality: Locality information for collected data should be used by the
system to enforce access restriction. This
restriction may be based on the location of a
user who wants to use the data, for example,
a user is able to record an audio stream at a
meeting while attending, but not from the
opposite side of the world (proximity). Also
the audio stream should not be disseminated
through the network (locality). This idea is
also present in location-limited channels and
location-based authentication (see below).
Access and recourse: A system should
give users easy access to collected personal
information (e.g., using a standardized interface). Also, they should be informed about

any usage of their data. Thus, abuse would

be noticed.
Following these principles, Langheinrichs
goal is to allow people and systems which want
to respect a users privacy to behave in such
a way. This should help in building a lasting
relationship based on mutual trust and respect.
However his goal is not to try to provide perfect
protection for personal information that is hardly
achievable anyway.

Fighting Sybil and DoS Attacks

As mentioned in the section First View: UC
Characteristics and Associated Risks, DoS
may be a serious attack in UC. In a world of
widely-deployed pervasive computing devices,
a phenomenon comparable to unsolicited email
as we experience it today in the Internet setting
would be disastrous. In this section we describe
a mechanism that can be use as a remedy for DoS
attacks and for so-called Sybil attacks as well
(see Levine, Shields,and Margolin (2006) for
an overview). The latter are an inherent threat to
recommender or reputation systems, which are
the subject of the following chapter. The term
Sybil stems from the name of a schizophrenic
literary character that has a total of 16 personalities. Analogously, an attacker takes part in a
recommender system under multiple identities in
order to manipulate the trustworthiness rating of
nodes. DoS and Sybil attacks have some similar
characteristics which is why they are jointly dealt
with in this section.
In a setting where UC nodes can be identified
in a reliable way, the risks for both attacks are
quite low. Clearly, a necessary condition for a
Sybil attack is a nodes ability to remain anonymous. In the DoS case, illegitimate messages
can be filtered out by so-called whitelists and
blacklists containing accepted and non-accepted
entities, respectively. Filtering in its turn can be
done on each network layer provided the sender

Security for Ubiquitous Computing

can be (re-)identified. The earlier the filtering is

done, the better, as this saves processing time:
For instance, if data can already be discarded
based on the senders IP address, this is easier
than first decrypting the payload and verifying
a signature or MAC.
However, there are situations where identifying information is not reliable or not available at
all. Remember that we even argued that sender
anonymity is in fact a desirable feature from a
privacy point of view. As a consequence, this
means that DoS or Sybil attacks cannot be prevented completely. In order to make them at least
more difficult, the system must ensure that sending
requests or creating a new identity respectively
requires some effort.

Proof-of-Work
So-called proof-of-work (PoW) techniques treat
the computational resources of each user of a resource or service as valuable. To prevent arbitrarily
high usage of a common resource by a single user,
each user has to prove that she has made some
effort, that is, spent computing resources, before
she is allowed to use the service. This helps to
prevent for example, users from sending millions of spam emails, or from mounting denial
of service attacks.
The idea of PoW is to require the sender of a
message or service request to provide the answer
to a computational challenge along with the actual
message. If the verification of the proof fails, the
recipient discards the whole request without further processing. Obviously, the costs of creating
such a proof must be some order of magnitude
higher than for system setup and proof verification. A challenge in a PoW scheme may either
be generated by the recipient or calculated based
on implicit parameters that cannot be controlled
by the sender (in order to prevent the re-usage of
a proof). The second variant is very well suited
for UC as no infrastructure or multi-round communication is required.

In the following example we sketch a

mechanism that is similar to the one applied by
Hashcash(Back, 1997). Here, the challenge is
time-dependent. The proof of work consists of
partially inverting a one-way hash function h by
brute-force search. The sender of a message has
to find a solution X to the following equation:

(2)

Here, id is the recipients identifier (e.g.,

a pseudonym), t is a time stamp and h is the
output of h shortened to the leading bits. The
tuple (id,t,X) is sent along with the payload. The
following checks are made at the receivers side:
At first, she checks whether id is her own ID and
whether the message is suitably recent, that is, t
differs less than a predefined clock skew bound
of s seconds from her local time. Furthermore the
above condition must hold. In order to prevent a
PoW from being reused, (t,X) can be checked for
freshness against a database that caches previous
values for s seconds.
The sender has to try out 2 possibilities for X
on average, until a solution is found. This implies
that the sender has to invoke the hash function 2
as many times as the recipient. The complexity for
the sender grows exponentially in . Depending on
the performance of the devices in use, should be
adjusted in a range of, say, 1020 in practice.
We will revert to the concept of PoW in the
following chapter in the context of micropayment
schemes.

Bootstrapping Secure
Communication
Consider the situation where two UC devices want
to establish a secure communication channel between each other in the presence of an adversary.
Here secure may have the meaning of confidentiality and/or authenticity. As confidentiality is

349

Security for Ubiquitous Computing

achieved by encryption, the parties have to secretly

agree on a shared key or exchange their public
encryption keys in an authentic waydepending
on whether they use a symmetric or asymmetric
cryptosystem. The requirements for authenticity
are quite similar: a MAC key has to remain secret
as well and the public key for the verification of
a digital signature has to be authentic.
The bootstrapping of a secure channel thus
boils down to either initially obtaining some
information in a confidential or authentic way.
Observe that, by using public key cryptography, an authentic communication relation can
be enhanced to an encrypted one. If there is no
(common) central authority like a certification
authority or a key distribution center the devices
have a trust relation with, they have to resort to
direct exchange of cryptographic information
via so-called out-of-band communication. This
is a well-known concept which is applied, for
instance, in the context of secret key distribution
by messengers or the verification of public key
fingerprints on the phone.
Out-of-band communication requires a
secondary link, typically with some technical
limitations, but with the huge advantage of being
safe from attacks. Such limitations, that make
this secondary channel unattractive for general
data transfer, include bandwidth, signal strength,
error rate, latency, and so forth. What is typically
considered a downside simultaneously reduces the
risk of attacks on this channel. Balfanz, Smetters,
Stewart, and Wong (2002) use the term of location-limited channels and emphasize that humans
can be in control of the communication links.
These authors assume an authentic out-of-band
communication channel while the resurrecting
duckling security policy model by Stajanoand
Anderson (1999) requires a confidential one.
Stajano and Anderson use a confidential channel in order to securely pair mobile devices for a
certain period of time; they use the term secure
transient association. We will discuss their model
in greater detail now.

350

Secure transient association addresses the

problem of device authentication in the absence
of a central and always available authority, for
example a public key infrastructure. Devices authenticate each other and agree upon a shared key
by physical contact. For a user, physical contact
has the advantage that it is simple to understand
and it is clear which devices are involved in the
paring. The two devices take different roles:

A slave (or duckling) obeys a master

A master (or mother duck) controls a slave

Initially the slave is in state imprintable. As

soon as a master has sent a key to the slavethis
process is called imprintingvia a physical contact, the slave follows commands only given by
the master who owns the key. The slave is now
in the state imprinted.
The association can be broken in three ways.
First, the master sends a special kill command to
the slave. The slave will reset itself into the initial
state imprintable. Second, after a predefined time
interval has gone by, the slave kills itself. And
third, after the completion of a certain transaction,
the slave will reset itself to imprintable.
The two states of a slave and its transition are
depicted in Figure 2. Note that the slave needs to
be constructed in a tamper resistant manner. This
guarantees that it is uneconomical for an attacker
to attack by artificially causing a slave to die, that
is, to force a slave into state imprintable. Otherwise, this would allow an attacker to imprint the
slave using its own key and gain control over the
slave. We quote Stajanos four formal principles
that make up the resurrecting duckling security
policy(Stajano, 2002):
1.

Two state principle: The entity that the

policy protects, called the duckling, can be in
one of two states: imprintable or imprinted.
In the imprintable state, anyone can take it
over. In the imprinted state, it only obeys
its mother duck.

Security for Ubiquitous Computing

Figure 2. Two states of a slave (duckling)

2.

3.

4.

Imprinting principle: The transition from

imprintable to imprinted, known as imprinting, happens when a principal, from then
on known as the mother duck, sends an
imprinting key to the duckling. This must
be done using a channel whose confidentiality and integrity are adequately protected.
As part of the transaction, the mother duck
must also create an appropriate backup of
the imprinting key.
Death principle: The transition from imprinted to imprintable is known as death.
It may occur under a very specific circumstance, defined by the particular variant of
the resurrecting duckling policy model that
one has chosen. Allowable circumstances,
each corresponding to a different variant of
the policy, include
o Death by order of the mother duck.
o Death by old age after a predefined
time interval.
o Death on completion of a specific
transaction.
Assassination principle: The duckling must
be constructed in such a way that it will be
uneconomical for an attacker to assassinate
it, that is, to cause the ducklings death
artificially in circumstances other than the
one prescribed by the Death principle of the
policy.

An appropriate key backup of the imprinting

key is necessary in order to keep control of the
slave in case the master is broken. Coming back to
the patient monitoring scenario from the section
Four UC Settings, the resurrecting duckling
security policy works for secure transient pairing
of a heartbeat monitor with a patients record.
Here, the record takes the role of the master and
the heartbeat monitor the role of the slave. Being at first in an imprintable state, as soon as the
heartbeat monitor is attached to the patient, it
is also imprinted by the patients record. Being
imprinted, only the master is able to read the
heartbeat and store it in the record. As soon as the
patient is fully recovered and leaves the hospital,
the patients record sends a kill command to the
heartbeat monitor. The pairing is broken and the
monitor is available for another patient.

Out-of-Band Channels
Many communication technologies are found in
the UC world that are suitable for cryptographic
out-of-band transmission. We now give a nonexhaustive list of examples that have been already
used in practice. Usually switching to a secondary
channel requires some human intervention, but
may also be triggered automatically by an appropriate protocol command on the primary channel. McCune, Perrig,and Reiter (2005) provide

351

Security for Ubiquitous Computing

a comparison of different out-of-band channels

with respect to their security, convenience, and
hardware requirements.

352

Infrared light is used in IrDA interfaces for

data transmission. It requires a direct line
of sight and distances of 1meter or less
between the communicating parties. The
salesman from the first scenario may use
IrDA to establish a secure connection to a
printer by exchanging a short-lived symmetric session key for the next print job or
alternatively go through a pre-authentication
protocol based on public key cryptography
(like in Balfanz, Smetters, Stewart, & Wong
(2002)). The idea of an IrDA-based magic
wand that can be used to point to a device
for establishing a secure connection is described by Spahic, Kreutzer, Khmer,and
Chandratilleke (2005).
Dynamically generated 2D barcodes can
encode about 50-100 bits of information.
They are suitable for modern mobile phones,
which have high-resolution displays and
built-in cameras. Claycomband Shin (2006)
have recently shown how to use colorized
barcodes to exchange a public key fingerprint
with an application to secure communications with ATMs. The authentication of an
802.11 access point using a sticker with a barcode on its casing is described inMcCune,
Perrig,and Reiter (2005).
Devices that have a speaker and a microphone could use audio to transfer a
fingerprint modulated on a piece of music.
While the diffusion of the signal is larger
than in the case of infrared, the presence of
a second audio sender can be detected by a
human. This technology is also suitable for
location-based authentication of devices.
An example is a caf rewarding customers
with media downloads by distributing access keys for the service via music played

inside the building(Coulouris, Dollimore,&

Kindberg, 2005).
Buhan, Doumen, Hartel,and Veldhuis
(2006) propose a user-friendly, two-party
key agreement protocol that takes as an input
biometric data of the device owners obtained
via a grip pattern sensor. A downside of this
approach, however, is the fact that biometric
templates are static information and become
useless once they fall into an opponents
hands.
Capkunand Cagalj (2006) apply ultrasonic
to measure the distance and spatial direction
of UC communication partners. Their concept of integrity regions forces the attacker
to be physically close to the targets. When
devices are operated by humans and the
maximum distance is appropriately short,
say 1 meter or less, the presence of a man-inthe-middle can be ruled out. Mayrhoferand
Gellersen (2007) show how to transmit short
messages over ultrasound in such a way that
only a recipient at an expected distance from
the sender can decode the signal.
UC devices that do not support any of the
above mechanisms, but have at least a user
interface, can let the user enter a sequence
of numbers by hand. Bluetooth applies an
algorithm called pairing that allows two
devices to agree on a common secret key
derived from a random value (transmitted
in the clear) and a static PIN (typically only
a few characters long) that has to be known
to each device.
Even a simple user interface consisting, for
instance, of an LED and a pushbutton is
enough to transmit 1 bit of information telling the device that it is now safe to exchange
public keys via the primary channel. This
may happen when the devices are connected
in an environment that is considered safe,
for exampleat the users home.

Security for Ubiquitous Computing

The last example is an application of weak

authentication as described by Arkkoand Nikander (2002). According to the temporal separation
principle, at a certain point in time the parties
consider safe they exchange a key via the primary channelwithout further protection, that
isunauthenticated and in plaintext. Provided that
no attacker is present in this phase, a long-term
secure communication channel can be set up.
The weak authentication mechanism completely
defeats passive attacks (by exchanging public
keys in an authentic fashion). It also raises the bar
for active attacks, since a man in the middle who
engaged during key agreement has to intercept
and re-encrypt each subsequent message in order
to remain undetected. A similar mechanism is
used in the Secure Shell (SSH) protocol.

RFID Security from EPC to

E-Passport
In this section the security of RFID systems is
discussed, mainly by means of two examples: tags
with a 96bit Electronic Product Code (EPC) are
used for the lifecycle management of commercial
goods as described in Scenario4. Such tags are
at the lower end of the scale since they offer very
limited functionality. At the other extreme, tags
embedded in modern identity cards or e-passports
(electronic passports) support state-of-the art
cryptographic algorithms.
The threat of abusive tracking of UC systems
in general and RFID tags in particular has already
been mentioned. Basically, tracking is what EPC
tags are built for: Unlike barcodes, they do not
simply indicate the manufacturer and product
type, but a serial number that is distinct for each
object. As access control and reader authentication is only an option in the EPC standards (see
http://www.epcglobalinc.org), tags are inherently
prone to skimming. That means that unauthorized
parties reasonably close may read the chip without
its owner noticing it. In the case of EPC tags that
operate at a frequency of around 900 MHz, the

specified reading range is 3 meters or more. Epassports use so-called proximity tags operating at
a frequency of 13.56MHz and a reading range of
at most 15 cm. Note that fraudulent RFID readers
do not necessarily adhere to standards and may
therefore exceed reading ranges significantly (see
e.g., Kirschenbaum& Wool, 2006).

Countermeasures Against Skimming

and Tracking
EPC tags illustrate a common limitation of most
RFID chips: Their identifier is defined during
the production process and cannot be changed
afterwards. This allows a tagged object to be
misused for tracking a person it is linked to. In
the discussion of anonymity in the section Privacy-Enhancing Technologies, we have already
seen that identification is a problem affecting all
network layers. RFID systems are no different in
that respect as tracking may in theory happen on
the application, the data link, or the physical layer.
Consider, for instance, tags used in access control
systems like the Hitag, Legic, or Mifare series.
In order to prevent cloning and replay attacks,
such tags may encrypt their communication on
the application level. While this hides personal
information of their owner, the tags themselves
can be easily traced using their static data link
layer ID.
In the EPC system, there is at least a basic, yet
very effective privacy protection mechanism: A
reader may send a tag the so-called kill command rendering it irreversibly inoperable. In
order to prevent DoS attacks, the kill command
requires authentication via a secret 32-bit number.
The METRO groups vision of an RFID-based
store (see http://www.future-store.org) already
encompasses a dedicated device behind the cash
desk to kill tags. However, this comes at the
price of excluding the tag ID from being used
by the consumer at her home or for the purpose
of recycling.

353

Security for Ubiquitous Computing

There a several non-destructive countermeasures against skimming and tracking, for a comparison see for example, Weis, Sarma, Rivest,and
Engels (2003) or Juels, Rivest,and Szydlo (2003).
For instance, RFID chips in identity cards can be
shielded in a Faraday cage made of metal foil which
is integrated in a wallet. Clipped tags(Karjoth&
Moskowitz, 2005) provide a user-friendly way
to deactivate a tag by manually separating chip
and antenna. Such a method provides immediate visual feedback about the tags state. Instead
of completely deactivating the tag, the antenna
may be transformed in a way that reduces reading range considerably. Blocker tags(Juelset al.,
2003) are devices that fool readers by simulating
a pile of tags in order to prevent them to single out
and communicate with a single one. As the EPC
addressing scheme is hierarchical, a blocker tag
may restrict the range of IDs selectively.

E-Passport Security Measures

Compared to EPC tags, the tags embedded in
e-passports contain information that is much
more sensitive. The specifications(ICAO, 2004a,
2004b) for machine readable travel documents
provided by the International Civil Aviation
Organization (ICAO), a United Nations body,
govern the storage of personal and especially
biometric data on the chip. Mandatory pieces of
information include name, date of birth, nationality, and a digitized portrait photograph. While
the standard defines iris scans and fingerprints as
optional features, the member states of the European Union have agreed on including fingerprints
in passports issued to their citizens(European
Union, 2006).
As a baseline security mechanism, all e-passports have to implement passive authentication.
This means that the data stored on the chip is
digitally signed by the issuing nation. The signature algorithm may be one of RSA, DSA, or
EC-DSA. While passive authentication prevents
modifications of the chip, it does not help against

354

chip cloning; only the optional feature active

authentication does. Here, the chip is given a
key pair the public key of which is included in
the signed data set. During a challenge-response
protocol, the chip proves possession of the private
key and therefore its authenticity to the reader.
While other chip data can be read and cloned,
the private key is not disclosed similarly to a
smartcard.
The ICAO specifications offer Basic Access
Control (BAC) as an optional protection against
eavesdropping, skimming, and tracking. When
implemented on the chip, a reader is required
to authenticate before being able to read nonsensitive data, that is, any information except
fingerprints (and iris scans if present) which
require additional authentication. EU e-passports
implement BAC while US e-passports do not. This
opens the door for American-sniffing bombs or
even personalized RFID-enabled bombs(Juels,
Molnar,& Wagner, 2005). The design of the BAC
mechanism is guided by the idea that a passport
has to be handed over to border control personnel, that is, physical access to the document is a
prerequisite to reading the RFID chip. In order
to start the communication, a reader must pass
a challenge-response protocol with the chip. In
this protocol, the reader has to implicitly prove
that it knows two cryptographic keys derived
from the passport number, the bearers date of
birth, and the expiration date. These parameters
are part of the so-called Machine Readable
Zone (MRZ) which is why an e-passport has
to be swiped through an optical reader first. The
keys derived from the MRZ are used for 3DES
encryption and integrity protection respectively.
Note that, although the 3DES encryption uses a
112-bit key, the entropy obtained from the MRZ
is much lower. The number of possible values
can be at most 256 in theory, but practical figures
may be significantly lower: The feasibility of a
brute-force attack on Dutch e-passports, which
provide encryption strength of around 35 bit due
to their predictable serial numbers, is described

Security for Ubiquitous Computing

by Smith (2006). Note that even strong encryption

does not prevent tracking by parties that have read
the MRZ once, as the derived keys are static.
Another challenge-response scheme is used to
protect the fingerprints on the chip. In this protocol, the chip authenticates the reader in order to
decide whether to disclose the fingerprint or not.
This mechanism involves national authorities that
issue certificates to border control organization of
foreign countries upon condition that fingerprints
are treated in accordance with data security and
protection legislation. These organizations in
turn issue certificates to individual readers. A
detailed description of the infrastructures used
for issuing certificates to passport manufacturers
and to readers respectively is given by Straub,
Hartl, and Ruppert (2006).

Research Outlook
This chapter gave an overview of the field of UC
security taking into account UC characteristics
and limitations. By means of representative settings, ranging from mobile work to RFID-based
real-time enterprises, common security challenges
were exhibited. At the heart of the chapter was
the presentation of a bunch of concrete solutions
to such challenges. This list is not claimed to be
exhaustive, but it is an initial step toward a taxonomy. The solutions stand pars pro toto for UC
security, as they are generic and scalable enough
to be adaptable to a large number of scenarios.
Social acceptance of UC technology strongly
depends on the security of UC systems. On the one
hand, the growing use of UC in everyday life must
not lead to security breaches as demonstrated,
for instance, in a recent attack on RFID-enabled
credit cards (Heydt-Benjamin et al., 2007). It is
also a safe bet that conflicts of interest will again
occur between privacy and law enforcement as
we have already seen in the aftermath of 9/11, for
example, concerning the traceability capabilities
of UC. We hope that this chapter motivates the

reader to delve deeper into the subject and refer

to the following textbooks:

IT security in general: Eckert (2006), Stallings (2006),

Focus on cryptographic aspects: Buchmann (2004), Stallings (2005); Menezes,
Oorschot,and Vanstone (1996),
UC security in particular: Stajano (2002),
RFID: Finkenzeller (2003).

Among the promising areas of UC security

research we could not discuss in detail here are
UC and its relation to Grid Computing (Storz,
Friday, & Davies, 2003) and relay attacks on RFID
systems (Kfir & Wool, 2005; Hancke 2006).

References
Abadi, M. (2002).Private authentication. In Proceedings of theWorkshop on Privacy-Enhancing
Technologies (PET).
Aitenbichler, E., Kangasharju, J.,& Mhlhuser,
M. (2004). Talking assistant: A smart digital
identity for ubiquitous computing. Advances in
pervasive Computing(pp.279284). Austrian
Computer Society (OCG).
Al-Muhtadi, J., Campbell, R., Kapadia, A., Mickunas, M.,& Yi, S.
(2002). Routing through the
mist: Privacy preserving communication in ubiquitous computing environments. In Proceedings
of IEEE International Conference of Distributed
Computing Systems (ICDCS)(pp.6574).
Arkko, J.,& Nikander, P. (2002). Weak authentication: How to authenticate unknown principals without trusted parties. In Proceedings of
the IWSP: International Workshop on Security
Protocols.
Avoine, G.,& Oechslin, P. (2005). RFID traceability: A multilayer problem. In A.S. Patrick&
M.Yung(Eds.), Financial cryptography and data

355

Security for Ubiquitous Computing

security, 9th International Conference, FC 2005,

Roseau, The Commonwealth of Dominica, February 28 - March 3, 2005, Revised Papers(Vol.3570,
pp.125-140).Springer.
Back, A. (1997). Hashcash. Retrieved October
17, 2007, http://www.cypherspace.org/~adam/
hashcash/.
Bahl, P., Chandra, R., Padhye, J., Ravindranath,
L., Singh, M., Wolman, A.,Zill, B.
(2006). Enhancing the security of corporate Wi-Fi networks
using DAIR. In Proceedings of the MobiSys
2006: Proceedings of the 4th international
conference on Mobile systems, applications and
services(pp.114).New York.
Balfanz, D., Smetters, D., Stewart, P.,& Wong,
H. (2002). Talking to strangers: Authentication in
adhoc wireless networks. In Proceedings of the
Symposium on Network and Distributed Systems
Security (NDSS 02), San Diego, California.
Beresford, A.R.,& Stajano, F. (2004, March). Mix
zones: User privacy in location-aware services.In
Proceedings of the IEEE International Workshop
on Pervasive Computing and Communication
Security (PerSec).
Bu ch m a n n , J. (20 0 4). In t r o d u c t i o n t o
cryptography.Springer.
Buchmann, J., May, A.,& Vollmer, U. (2006).
Perspectives for cryptographic long-term
security.Comm.ACM, 49(9), 5055.
Buhan, I.R., Doumen, J.M., Hartel, P.H.,& Veldhuis, R. N.J.(2006, June).Feeling is believing:
A location limited channel based on grip pattern
TRbiometrics and cryptanalysis(Tech. Rep.No.
CTIT-06-29). Retrieved October 17, 2007, from
http://eprints.eemcs.utwente.nl/5694/.
Capkun, S.,& Cagalj, M.(2006).
Integrity regions: Authentication through presence in wireless networks.In Proceedings of the WiSe 06:
Proceedings of the 5th ACM workshop on Wireless
security(pp.110), New York.

356

Chaum, D. (1981). Untraceable electronic mail,

return addresses, and digital pseudonyms.Commun. ACM, 24(2), 8488.
Claycomb, W.R.,& Shin, D.(2006).Secure real
world interaction using mobile devices.In Proceedings of thePERMID / Pervasive 2006.
Coulouris, G., Dollimore, J.,& Kindberg,
T.(2005).Distributed systems: Concepts and
design.Addison-Wesley Longman.
Dierks, T.,& Rescorla, E. (2006). The transport layer security (TLS) Protocol Version
1.1.RFC4346.
Dobbertin, H., Bosselaers, A.,& Preneel,
B.(1996).RIPEMD-160, A strengthened version
of RIPEMD. In Proceedings of theFast Software
Encryption(pp.7182).Springer LNCS 1039.
Eastlake, D., & Jones, P. (2001). US Secure hash
algorithm 1 (SHA1).RFC3174.
Eckert, C. (2005).

Security issues of mobile devices. In Proceedings of the Security in Pervasive

Computing, Second International Conference,
SPC 2005(Vol.3450).Boppard: Springer.
Eckert, C. (2006). IT-Sicherheit. Konzepte Verfahren: Protokolle. Oldenbourg.
E
uropean Union (2006).
EU passport specification, Working document (28/06/2006)(Tech.R
ep.).
Fleisch, E.,& Mattern, F. (2005). Das Internet
der Dinge: Ubiquitous Computing und RFID in
der Praxis.Springer.
Gemmell, P.S.(1997).
An introduction to threshold cryptography.RSA Laboratories CryptoBytes, 2(3), 712.
Gligor, V. (2005). Cryptolite: How lite can secure crypto get? Information Security Summer
School.
Grlach, A., Heinemann, A., Terpstra, W.W.,&
Mhlhuser, M.(2005). Location privacy. In

Security for Ubiquitous Computing

A.Boukerche(Ed.), Handbook of algorithms

for wireless networking and mobile computing(pp.393411).Chapman & Hall/CRC.
Greiner, L.(2006).The urge to converge.netWorker, 10(2), 2630.
Gruteser, M.,& Grunwald, D. (2003). Anonymous
usage of location-based services through spatial
and temporal cloaking. In Proceedings of the First
International Conference on Mobile Systems,
Applications, and Services (MobiSys)(pp.31
42).USENIX.
Han kerson, D., Menezes, A.J.,& Vanstone, S. (2004). Guide to elliptic curve
cryptography.Springer-Verlag.
Heydt-Benjamin, T.S., Bailey, D.V., Fu, K., Juels,
A., & OHare, T.(2007). Vulnerabilities in firstgeneration RFID-enabled credit cards. Financial
Cryptography.
Hohl, A.,& Zugenmaier, A. (2005). Safeguarding
personal data using trusted computing in pervasive
computing. Privacy, Security and Trust within the
Context of Pervasive Computing(pp.147155).
Springer.
Husemann, D.,& Nidd, M. (2005). Pervasive
patient monitoring Take two at bedtime.ERCIM
News, 7071.

Privacy for Emerging Areas in Communication

Networks: SecureComm,Athens, Greece.
Juels, A., Rivest, R.,& Szydlo, M.
(2003, October).
The blocker tag: Selective blocking of RFID tags
for consumer privacy. In V.Atluri(Ed.), Conference on computer and communications security:
ACM CCS(pp.103111).Washington, DC.
Karjoth, G.,& Moskowitz, P.A.(2005).Disabling
RFID tags with visible confirmation: Clipped
tags are silenced.In Proceedings of the WPES
05: Proceedings of the 2005 ACM Workshop on
Privacy in the Electronic Society(pp.2730).
New York.
Kirschenbaum, I.,& Wool, A. (2006). How to
build a low-cost, extended-range RFID skimmer. In Proceedings of theUSENIX Security
Symposium(pp.4357).
Knospe, H.,& Pohl, H. (2004, NovemberDecember). RFID security. Information Security
Technical Report, 9(4), 3950.
K rawcz yk, H., Bellare, M.,& Canet ti,
R.(1997).HMAC: Keyed-hashing for message
authentication.RFC2104.
Langheinrich, M. (2001). Privacy by design Principles of privacy-aware ubiquitous systems. In
G.D. Abowd, B.Brumitt,& S.A. Shafer(Eds.),
Ubicomp(Vol.2201, p.273-291).Springer.

ICAO (2004a). Development of a logical data

structure: LDS for optional capacity expansion
technologies, Revision 1.7(Tech.Rep.). Retrieved October 17, 2007, from http://www.icao.
int/mrtd/.

Langheinrich, M. (2002). A privacy awareness

system for ubiquitous computing environments.
In G.Borriello& L.E. Holmquist(Eds.), Ubicomp(Vol.2498, pp.237245).Springer.

ICAO. (2004b). PKI for Machine Readable Travel

Documents offering ICC Read-Only Access, Version 1.1(Tech.Rep.). Retrieved October 17, 2007,
from http://www.icao.int/mrtd/.

Levine, B.N., Shields, C.,& Margolin, N.B.(2006,

October).A survey of solutions to the sybil
attack(Tech. Rep.No.2006-052). Amherst, MA:
University of Massachusetts Amherst.

Juels, A., Molnar, D.,& Wagner, D. (2005, September). Security and privacy issues in e-passports.
In Proceedings of the Conference on Security and

Mayrhofer, R.,& Gellersen, H. (in press).

On the security of ultrasound as out-of-band
channel.In Proceedings of the IPDPS 2007: In-

357

Security for Ubiquitous Computing

ternational Parallel and Distributed Processing

Symposium.IEEE Computer Society Press.
McCune, J.M., Perrig, A.,& Reiter, M.K. (2005).
Seeing-is-believing: Using camera phones for
human-verifiable authentication. In Proceedings of the IEEE symposium on security and
privacy(pp.110124). IEEE Computer Society.
Menezes, A., Oorschot, P.C. van,& Vanstone, S.A.(1996).Handbook of applied
cryptography.CRC Press.
Orava, P., Haverinen, H.,& Honkanen, J.-P. (2002).

Temporary MAC addresses for anonymity.

Potlapally, N.R., Ravi, S., Raghunathan, A.,&
Jha, N.K.(2003).Analyzing the energy consumption of security protocols. In Proceedings
of theISPLED03(pp.3035).
Rivest, R.L., Shamir, A.,& Adleman, L.(1978).A
method for obtaining digital signatures and public-key cryptosystems.Commun. ACM, 21(2),
120126.
Robinson, P. (2005). An application-led approach for security research in ubicomp (position paper).In Proceedings of theUbiapp
Workshop,Pervasive 2005.
Seys, S. (2006). Cryptographic algorithms and
protocols for security and privacy in ad hoc
networks. Unpublished doctoral dissertation,
Katholieke Universiteit Leuven.
Smith, R.W.(2006).E-passport hack demonstrated on Dutch TV.Retrieved October 17, 2007,
from, http://www.heise.de/english/newsticker/
news/69197.
Spahic, A., Kreutzer, M., Khmer, M.,& Chandratilleke, S. (2005). Pre-authentication using
infrared. Privacy, Security and Trust within
the Context of Pervasive Computing(Vol.780,
pp.105112).
Stajano, F. (2002). Security for ubiquitous computing. John Wiley & Sons.

358

Stajano, F.,& Anderson, R.J. (1999). The resurrecting duckling: Security issues for ad-hoc
wireless networks. In Proceedings of the Security Protocols, 7th International Workshop,
Cambridge, UK, April 19-21, 1999, Proceedings(p.172-194).
Stallings, W.(2005).Cryptography and network
security.Prentice Hall.
St alli ngs, W.(20 06).Net work sec urit y
Prentice Hall.
essentials.
Straub, T., Hartl, M.,& Ruppert, M. (2006).

Digitale Reisepsse in Deutschland - Prozesse

und Sicherheitsinfrastruktur.
In Proc.
Sicherheit
2006(p.233-243).
Straub, T.,& Heinemann, A.(2004).An anonymous bonus point system for mobile commerce
based on word-of-mouth recommendation. In
L.M. Liebrock(Ed.), Applied computing 2004.
Proceedings of the 2004 ACM Symposium on
Applied Computing(pp.766773),New York:
ACM Press.
Vogt, H. (2005). Small worlds and the security
of ubiquitous computing. In Proceedings of the
WOWMOM 05: Proceedings of the First International IEEE WoWMoM Workshop on Trust,
Security and Privacy for Ubiquitous Computing(pp.593-597), Washington, DC.
Weis, S.A., Sarma, S.E., Rivest, R.L.,& Engels,
D.W. (2003). Security and privacy aspects of lowcost radio frequency identification systems.In
Proceedings of theSecurity in Pervasive Computing(p.201-212).

Additional Reading
Eckert, C. (2006). IT-Sicherheit. Konzepte: VerOldenbourg.
fahren: Protokolle.

Security for Ubiquitous Computing

Finkenzeller, K.(2003).RFID handbook: Fundamentals and applications in contactless smart

John Wiley & Sons.
cards and identification.
Hancke, G.P. (2006).

Practical attacks on proximity identification systems (Short Paper).In

Proceedings of the
2006 IEEE Symposium on
Security and Privacy, 328-333.
Kfir, Z.,& Wool, A.(2005).Picking virtual pockets using relay attacks on contactless smartcard
systems. In Proceedings of the1st Intl. Conf. on
Security and Privacy for Emerging Areas in Communication Networks(p.47-58). IEEE.

Stajano, F. (2002). Security for ubiquitous computing. John Wiley & Sons.
Stallings, W.(2006).Network security essentials.
Prentice Hall.
Storz, O., Friday, A.,& Davies, N.(2003).Towards
ubiquitous ubiquitous computing: An alliance
with the grid.In Proceedings of the System
Support for Ubiquitous Computing Workshop
(UbiComp Workshop 2003).

This work was previously published in the Handbook of Research on Ubiquitous Computing Technology for Real Time Enterprises, edited by M. Mhlhuser and I. Gurevych, pp. 337-362, copyright 2008 by Information Science Reference, formerly
known as Idea Group Reference (an imprint of IGI Global).

359

Section II

Development and Design

Methodologies
This section provides in-depth coverage of conceptual architectures, frameworks and methodologies
related to the design and implementation of interfaces and technologies that promote successful human-computer interaction. Throughout these contributions, research fundamentals in the discipline are
presented and discussed. From broad examinations to specific discussions on particular frameworks
and infrastructures, the research found within this section spans the discipline while also offering detailed, specific discussions. Basic designs, as well as abstract developments, are explained within these
chapters, and frameworks for designing and personalizing interfaces for medical equipment, televisions,
and automobiles are discussed.

361

Chapter 2.1

Measuring the Human Element

in Complex Technologies
Niamh McNamara
University College Cork, Ireland
Jurek Kirakowski
University College Cork, Ireland

Abstract
Measuring satisfaction can provide developers with valuable insight into the usability of a
product as perceived by the user. Although such
measures are typically included in usability
evaluations, it is clear that the concept itself is
under-developed. The literature reveals a lack of
cumulative, systematic research and consequently
the field is in disarray (Hornbk, 2006). Clearly,
the area needs a strong theoretical foundation
on which to base research. This paper reviews
the literature on user satisfaction and proposes a
conceptualisation and definition of the concept
that will aid researchers in the development of
valid measures.

USER SATISFACTION AS
A USABILITY PARAMETER
The ISO 9241-11 standard suggests assessing
usability in terms of performance by measur-

ing effectiveness and efficiency, and in terms

of internal state by measuring satisfaction. It
recommends that at least one measure of each
parameter should be included in usability evaluations. Despite this, Lindgaard and Dudek (2003,
p. 430) have observed that the:
Human-computer interaction literature has
traditionally concerned itself almost exclusively
with effectiveness and efficiency with satisfaction
being regarded (as) mainly a by-product of great
usability.
Although satisfaction is a central concept in
HCI it has been described as underdeveloped
(Tractinsky, 2004, p. 352). It is the criterion most
often included in usability evaluations (Kirakowski, 2002) but there is little understanding
on the part of evaluators of what exactly they are
measuring, and often the tools they employ are
poorly developed. This article will illustrate the
consequences of the lack of a cumulative, systematic approach to research that is evident in the

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Measuring the Human Element in Complex Technologies

literature on user satisfaction (Hornbk, 2006). It

will be shown that much of what is known about
satisfaction is based on relatively little empirical
work and represents opinion rather than fact. The
diversity of adjectives used in satisfaction measures is symptomatic of the lack of a cumulative,
systematic approach to research in the field. Hornbk (2006, pp. 90, 91) observes that comparisons
across studies become difficult because of this
diversity, as does the understanding of distinct
dimensions of subjective satisfaction.
Clearly, the field needs a theoretical foundation on which to base research beginning with a
sound conceptualisation of the construct and a
suitable definition.
There are very few definitions of satisfaction
to be found in the HCI literature. Possibly the
closest the field has to a standard definition is that
provided by ISO 9241-11, where satisfaction is
defined as freedom from discomfort and positive
attitudes towards the product. However, this is
quite a broad definition that gives little insight into
the nature of the satisfaction construct. Although
there is no widely-accepted definition, there seems
to be a number of common elements that most
researchers agree on. Satisfaction is:
1.

2.
3.

4.

A response (either cognitive or emotional)

on the part of the individual (Chin, Diehl, &
Norman, 1988; LaLomia & Sidowski, 1990;
Lindgaard & Dudek, 2003)
To an interactive system
Based on the degree to which the user
feels the device facilitates the completion
of a particular goal (LaLomia & Sidowski,
1990; Bevan & MacLeod, 1994; Kirakowski,
1999)
That is linked to usage (Lund, 2001; Kirakowski, 2002)

These characteristics are similar to those of the

attitude construct in social psychology. Indeed,
satisfaction is often described in the literature as
an attitude (Melone, 1990; Nielsen, 1993; Kei-

362

nonen, 1998; Shneiderman, 1998). Melone (1990)

suggests that conceptualising satisfaction in this
way provides the necessary theoretical foundation
for research and retains the essential elements
embodied in the user satisfaction construct.
This article starts by reviewing attempts to
measure and uncover the structure of user satisfaction from early HCI, research on Information Systems, and from recent HCI. Checklist
approaches to usability in general (e.g., Reiterer
& Oppermann, 1993; Ravden & Johnston, 1989)
will not be reviewed here as they are too broad
in scope; neither will the influential research of
Dzida, Herda, and Itzfeldt (1978) for the same
reason. Although each of these approaches creates
methods of assessment which involve questionnaires, they only look superficially like satisfaction
questionnaires; they are closer to checklists.
The measurement of satisfaction would greatly
benefit from the provision of a sound theoretical
base for research (Melone, 1990; Hornbk, 2006).
Melone (1990) suggests that this would be best
achieved by conceptualising satisfaction as an
attitude. Although HCI uses the term attitude
regularly in discussions on satisfaction, rarely are
the implications of this way of conceptualising
satisfaction addressed. This article will go on to
illustrate how this conceptualisation does indeed
retain satisfactions essential elements while also
clarifying some theoretical issues facing researchers, specifically (1) determining the nature of
the users response (i.e., whether it is cognitive
or emotional), (2) conceptualising satisfaction
as a characteristic of the interaction (quality in
use) rather than as a product attribute (quality of
features) and (3) understanding the relationship
between satisfaction and usage behaviour. Following conceptualisation of satisfaction as an attitude,
a definition of the construct in keeping with its
character as a usability parameter is offered.
Finally, four proposals are made for the future development of measuring instruments for
satisfaction.

Measuring the Human Element in Complex Technologies

UNCOVERING THE STRUCTURE OF

USER SATISFACTION
Early Satisfaction Research in HCI
Most of the research on satisfaction was carried
out in the context of developing questionnaires
to measure user satisfaction during the late 1970s
and 1980s. At this time, developers were becoming
more aware of the need to evaluate the usability
of their products. Various types of questionnaires
were developed. Some were single-use instruments constructed for specific studies that often
only provided a global measure of satisfaction,
while others were developed in industry, such
as Brookes (1996) system usability scale (which
was developed in the 1980s while the author was
employed by the Digital Equipment Corporation,
DEC) and Lewis (1995) IBM usability questionnaires. While instruments like SUS and the IBM
questionnaires have proven to be quite useful
when evaluating software, they have offered
little in the way of progress in understanding
the nature of user satisfaction. For this, one has
to turn to instruments developed in an academic
context. While few such instruments exist, they
are worth examining primarily for their efforts
to uncover the structure of user satisfaction, and
also as a potential starting point for research on
user satisfaction with new technology.
The questionnaire for user interface satisfaction (QUIS) (Chin, Diehl, & Norman, 1988) and
the computer user satisfaction inventory (CUSI)
(Kirakowski & Dillon, 1988) are two early satisfaction questionnaires developed at roughly the
same time, but that conceptualised the construct
in two different ways.
QUIS, which is still widely used today, was
originally constructed to measure the users subjective rating of the human-computer interface
(Chin et al., 1988, p. 213). The authors viewed
satisfaction as dependent on system features, and
respondents were asked to report their satisfaction
with the use of these features on a 9-point scale.

The questionnaire had five sections referring to

different aspects of the system, namely, Overall
Reactions, Screen, Terminology and System Information, Learning, and System Capabilities.
Chin et al (1988) carried out a factor analysis on
the responses to version 5 of the questionnaire
from which they extracted the following four
factors: (1) Learning, (2) Terminology and Information Flow, (3) System Output and (4) System
Characteristics. Carrying out factor analyses
provides an important insight into the nature of
a construct, however Chin et al. (1988) did not
provide a comprehensive explanation of what the
four factors measured. Although they attempted
to uncover the underlying structure of satisfaction, the authors failed to provide a meaningful
conceptualisation of the construct.
This is in contrast to the development of the
Computer User Satisfaction Inventory (CUSI,
Kirakowski & Dillon, 1988). This research was
informed by Banduras (1977) Self-efficacy
Theory. Self-efficacy refers to the individuals
belief that he or she can accomplish a particular
task. The authors hypothesised that satisfaction
could be measured on two dimensions: Affect,
which assesses the (individuals) feelings of fear
or pleasure and Competence, which assesses
the individuals feelings of mastery over the
computer system (Kirakowski & Corbett, 1988,
p.330). Kirakowski and Dillon (1988) not only
created a questionnaire that could be used as an
alternative to other more resource- and labourintensive methods, but also attempted to provide
a conceptual understanding of the satisfaction
construct.
CUSI includes an affective dimension that is
not present in QUIS. While CUSIs developers
did not provide a definition of satisfaction per
se, it is clear that they did believe it to contain
elements relating to user emotions (Affect) and
performance (Competence). This would appear
consistent with later definitions of satisfaction
such as ISO 9241-11, Bevan and MacLeod (1994)
and Bevan and Azuma (1997).

363

Measuring the Human Element in Complex Technologies

The conceptualisation of usability as quality

in use in the 1990s greatly influenced research
on user satisfaction. Satisfaction was no longer
widely regarded as a product property, but as a
property of the interaction between the user and
the product (Bevan, Kirakowski, & Maissel, 1991).
This proved to be a much more successful approach to measuring satisfaction than the previous
quality of features approach espoused by Chin et
al (1988). Their focus on specific attributes of the
product resulted in model instability and limited
applicability across applications for questionnaires that followed this route. Thus, attention
focused on constructing models of satisfaction
that (attempted) to measure the way the software
being rated fulfils some latent need within the user
community (Kirakowski, 1999, p.129).
Examining user reactions to an interactive
system applies a psychological approach to the
investigation of user satisfaction. The software
usability measurement inventory (SUMI) (Kirakowski & Corbett, 1993; Porteous, Kirakowski, &
Corbett, 1995; Kirakowski, 1996), based on this
approach to user satisfaction measurement, has
become the de facto instrument for evaluating
software applications (Kirakowski, 2002). It measures user-perceived software quality, which
it is hypothesised, consists of five dimensions:
Affect, Controllability, Efficiency, Helpfulness
and Learnability.
This model has also been successfully applied
to the evaluation of user satisfaction with Web
sites in the Website Analysis and MeasureMent
Inventory (WAMMI) (Kirakowski, Claridge, &
Whitehand, 1998), illustrating its applicability
across technological domains. The five factors
of the model were modified slightly for Web
sitesthe main difference being that the Affect dimension is replaced by an Attractiveness
dimension.
We now turn to research on satisfaction from
the discipline of management information systems
(MIS). While this literature is often included in
reviews of user satisfaction research, rarely is

364

it noted that the constructs of user satisfaction

and user information satisfaction are entirely
different. The following section will explain
these differences and advise caution in borrowing
concepts across disciplines.

Information Systems Research on

Satisfaction
HCI and management information systems (MIS)
are two disciplines that have a significant interest
in assessing satisfaction with technology, and
consequently, often cite each others research.
However, researchers from both disciplines do not
seem to recognise the vastly different perspectives each takes on what they perceive to be the
same construct, particularly concerning the role
of usability in operationalisations and definitions
of satisfaction.
The term user information satisfaction is itself
suggestive of the differences between the ways
satisfaction is conceptualised in the HCI and MIS
disciplines. MIS is primarily interested in the
benefits of technology for decision-making in an
organisation and its impact on productivity. The
goal of HCI, on the other hand, is to create technology that supports users (Preece, Rogers, Sharp,
& Benyon, 1994). The definitions of satisfaction
from MIS are broadly similar to ISO 9241-11s
definition of satisfaction. For example, Bailey &
Pearson (1983) define it as the sum of feelings or
attitudes, both positive and negative, toward the
spectrum of factors affecting a specific situation
(p. 531). Similarly, Doll & Torkzadeh (1988) define
satisfaction as the affective attitude towards a
specific computer application by someone who
interacts with the application directly (p.261).
However, the operationalisations of the construct
between MIS and HCI are quite different. Typically, the MIS literature treats usability as one
dimension of user satisfaction, while HCI treats
satisfaction as a usability parameter. The user
information satisfaction construct is influenced
by user role rather than conceptualisations of

Measuring the Human Element in Complex Technologies

usability, while the user satisfaction construct

is typically influenced by both. This may be the
reason why the usability or ease of use dimension was only included in (MIS) user information
satisfaction questionnaires with the advent of
end-user computing.
The computing environment and the role of
the end-user were undergoing rapid changes during the late 1980s, and with this came a change
in the way satisfaction was conceptualised in
the MIS literature. Early studies such as those
conducted by Bailey & Pearson (1983) and Ives,
Olson, and Baroudi (1983) dealt with a system
that included the computer, the end-user, and the
programmer/analyst. As the end-user became
more autonomous, attention was increasingly
focused solely on the application, and the output
and usability began to be considered. Organisational structures such as EDP personnel were not
included in evaluations as prominently as had
been the case previously, although some evaluations still included the adequacy of the technical
support provided.
A new definition of satisfaction in accordance
with these changes was proposed by Doll and
Torkzadeh (1988). They defined end-user computing satisfaction as the affective attitude towards
a specific computer application by someone who
interacts with the application directly (p. 261).
Doll and Torkzadeh (1988) argued that because
the end-user was now interacting with the system
directly instead of through programmers or analysts, there was a need to consider the usability
(ease of use) of the application. The proposed
model, which the authors claimed was more
suitable for an end-user computing environment
than previous efforts, consisted of five factors,
namely Content, Accuracy, Format, Ease of Use,
and Timeliness.
Doll and Torkzadeh (1988) did include items
referring to ease of use that might lead one to infer
that this conceptualisation of end-user computing
satisfaction is closer to HCIs user satisfaction
than user information satisfaction. However, this

is not the case. Doll & Torkzadeh (1988) still focus

on the information product, that is, satisfaction
with the information the system provides to the
user to assist in decision-making. Furthermore,
it seems to oversimplify the concept of usability
by simply including the following two items to
measure it: is the system user friendly? and
is the system easy to use? (Doll & Torkzadeh,
1988, p. 268). Clearly, these items do not reflect
the complexity of the usability construct.
Doll and Torkzadehs (1988) end-user computing satisfaction measure became the standard
instrument within the MIS community. Much
of the research that was conducted after the
publication of this questionnaire either involved
further validation studies of the instrument or
attempts to investigate the relationship between
satisfaction and other variables such as usage, user
involvement in design, and so on. The research
focus shifted from purely measurement issues to
theoretical issues, particularly the need to identify
the nature of the satisfaction construct and the importance of establishing a sound theoretical base
for research. Thus, while the satisfaction construct
differs between the two disciplines of MIS and
HCI, one thing HCI can and should learn from
MIS is the importance of providing a theoretical
foundation for research (Melone, 1990).

Recent Approaches to Satisfaction

in HCI
Towards the late 1990s, little further progress had
been made in HCI in revealing the true nature of
the satisfaction construct. At this time, researchers were becomingly increasingly interested in
investigating the wider user experience. There are
two emerging approaches to satisfaction research
in light of this development.
In the first approach, researchers believe that
the satisfaction construct can encompass the
wider elements of the user experience (e.g., Han
& Hong, 2003; Lindgaard & Dudek, 2003; Yun,
Han, Hong, & Kim, 2003), particularly with the

365

Measuring the Human Element in Complex Technologies

inclusion of new dimensions. The operationalisations proposed by these researchers are very
different to those already discussed. For example,
Lindgaard and Dudek (2003, p. 430) define satisfaction as an expression of affect and suggest it
consists of four dimensions: Aesthetics, Likeability, Expectation, & Emotion. Similarly, Yun et al.
(2003) define satisfaction as the users subjective
impression of the device which is informed by performance and emotional aspects. The dimensions
of satisfaction they propose include Luxuriousness, Simplicity, Attractiveness, Colourfulness,
Texture, Delicacy, Harmoniousness, Salience,
Ruggedness, and Overall Satisfaction. Finally,
Han and Hong (2003) define satisfaction as meaning that products should provide pleasure and /
or attractiveness (p.1441). The corresponding
dimensions they propose include Volume, Shape,
Elegance, Harmoniousness, Simplicity, Comfort,
Attractiveness, and Overall Satisfaction. Clearly,
these conceptualisations of satisfaction have little
to do with perceived usability; rather they refer
to aesthetic concerns.
The second approach taken by researchers
involves a critique of the relevance of satisfaction.
It is argued that the construct should be replaced
(Jordan, 1998, 2000; Dillon, 2001). Just like the
call to go beyond usability (Logan, 1994), Dillon (2001) called for the creation of a construct
that goes beyond satisfaction. This is because
he believed that the concept of satisfaction as it
had been understood up to then was inadequate
for application to new technology. He proposed
the Affect construct as a replacement which is
predominantly concerned with user emotions.
Finally, recent approaches to satisfaction are
moving from a view of the construct as a characteristic of the interaction to one that conceptualises
satisfaction as an intrinsic product attribute. Even
though satisfaction has been shown not to be a
product characteristic (Kirakowski, 1999), new
conceptualisations of satisfaction proposed by Yun
et al. (2003) and Han and Hong (2003) take an ergonomic approach and conceptualise satisfaction

366

in terms of product properties. Yun et al. (2003)

and Han & Hong (2003) argue that satisfaction is
dependent on design. Thus, each of their dimensions consists of particular product properties
known as human interface elements (HIEs) that
they believe contribute to satisfaction. Using the
method of Kansei Engineering, they maintain that
it is possible to predict a users level of satisfaction
with a product before it is released. In doing this,
Yun et al. (2003) and Han and Hong (2003) ignore
the impact of the context of use on the interaction
between the user and the device. Consequently,
we see the same difficulties inherent in their approach that were encountered by others who took
an ergonomic approach to defining and measuring
usability. Before any evaluation could take place
using this approach, evaluators would have to
ascertain which particular dimensions of satisfaction and which HIEs were relevant to the device
under evaluation, making it a time-consuming
method, and one that is virtually impossible to
apply across devices.
It seems that some researchers are keen to align
satisfaction with the concept of experience for no
other reason except that it is the only subjective
construct consistently referred to in the usability
literature (Dillon, 2001). However, there is a clear
distinction, which is often not made, between
the concepts of usability and experience. (This
issue is outside the scope of this article; see McNamara and Kirakowski (2005, 2006) for further
discussion.)
The satisfaction construct is becoming increasingly broad as is evident in the new definitions
and operationalisations of satisfaction that have
been proposed in the literature. Indeed, the only
common link remaining between recent conceptualisations of satisfaction and earlier ones seems
to be that it is still viewed as a multidimensional
construct. Unfortunately, there is little theoretical
or empirical justification offered as the basis for
these new models of satisfaction. Indeed many
new dimensions appear to be borrowed from
other disciplines such as Consumer Psychology.

Measuring the Human Element in Complex Technologies

The absence of clear content boundaries has led

to this indiscriminate borrowing of concepts and
theories across disciplines.

What is Satisfaction?
It is clear from this review of the user satisfaction
literature that there are three theoretical challenges
facing researchers.
First, little insight is given into the nature of
the construct. Some only offer broad definitions
of satisfaction and little else (e.g., ISO 9241-11).
Others offer both a definition and model of the
construct. But in many cases the model does not
seem to correspond to the definition. For example,
Yun et al. (2003) define satisfaction in terms of
both performance and affective elements, yet their
operationalisation of the construct does not include
any dimensions relating to user performance.
Some view the user response as exclusively cognitive (e.g., Chin et al., 1988), others as exclusively
emotional (e.g., Han & Hong, 2003; Lindgaard &
Dudek, 2003), and still others who take a combined
approach including both cognitive and affective
elements in their models of satisfaction (e.g. Porteous et al., 1995; Kirakowski et al., 1998).
Second, some researchers view satisfaction
as a characteristic of the interaction between the
user and the product (e.g. Porteous et al., 1995;
Kirakowski et al., 1998) which is in keeping with
the quality in use approach to usability proposed in
ISO 9241-11 and widely accepted by the majority
of usability professionals. However, others view
satisfaction as a product attribute (e.g., Chin et
al., 1988; Han & Hong, 2003; Yun et al., 2003)
indicating a return to the generally discredited
quality of features approach to measurement.
Finally, it appears to be taken for granted that
satisfaction is linked to usage. Many mention
this link (e.g., Lund, 2001; Kirakowski, 2002),
but there is little evidence of any investigations
(in the HCI literature at least) into the nature of
this relationship.

It is unlikely that these issues can be resolved

without better theoretical foundations for research.
It is argued that conceptualising satisfaction as
an attitude can provide such a base (Melone,
1990).

USER SATISFACTION AS AN
ATTITUDE
The study of attitudes, their structure, formation, and their relationships with other constructs
(most notably behaviour) has occupied a central
position within Social Psychology almost since
its inception (Palmerino, Langer, & McGillis,
1984; Ajzen, 2001).
Although there does not exist a standard definition of attitude (Fishbein & Ajzen, 1975; Dawes
& Smith, 1985; Olson & Zanna, 1992), it is clear
that evaluation is an important component. This is
illustrated in the following four definitions of attitude taken from classics in the attitude literature:
(1) (A term) used to refer to a relatively enduring
tendency to respond to someone or something in
a way that reflects a positive or negative evaluation of that person or thing (Manstead, 1996,
p. 3); (2) (a tendency) to evaluate an entity with
some degree of favour or disfavour, ordinarily
expressed in cognitive, affective, and behavioural
responses (Eagly & Chaiken, 1993, p.155); (3)
a persons feelings toward and evaluation of
some object, person, issue, or event (Fishbein
& Ajzen, 1975, p.12) and finally, (4) a summary
evaluation of a psychological object captured in
such attribute dimensions as good-bad, harmful-beneficial, pleasant-unpleasant, and likableunlikable (Ajzen, 2001, p. 28).
These definitions demonstrate that theorists
define attitudes in affective (e.g., Fishbein &
Ajzen, 1975), behavioural (e.g., Manstead, 1996),
and cognitive (e.g., Ajzen, 2001) terms. The
definition offered by Eagly and Chaiken (1993)
where attitudes are defined in terms of all three
components (affect, behaviour, and cognition)

367

Measuring the Human Element in Complex Technologies

represents the most common conceptualisation of

attitude structure. Affect refers to the individuals
emotional response to the target of the attitude. Behaviour is the action engaged in by the individual
towards the target, which is assumed to reflect the
attitude held by the individual. Cognition refers to
the knowledge and beliefs held by the individual
that link the target to particular attributes.
Definitions of attitude vary because theorists
disagree regarding the relative importance of the
various processes of affect, behaviour, and cognition involved in attitude formation.
Explaining attitude formation on the basis
of affective processes is based primarily on two
areas of research: the principles of conditioning,
both operant and classical; and, on Zajoncs (1980)
research on the Mere exposure effect. Classical
conditioning involves pairing an object with a
stimulus that elicits an affective response. Through
repeated association, the attitude object comes
to elicit the affective response and an attitude is
formed. Operant conditioning typically involves
reinforcement of positive or negative statements
by an individual. The attitude that subsequently
forms depends on whether positive or negative
statements were reinforced. The mere exposure
theory states that familiarity with objects, based
on the frequency with which they are encountered,
increases liking for those objects.
Bems (1972) Self-perception theory is often
used to describe attitude formation in terms of
behavioural processes. According to this theory,
individuals infer their attitudes from their behaviour, that is, engaging in a particular behaviour
towards an object (reflecting favour or disfavour)
is indicative of their underlying attitude. Bem
(1972) states that individuals take context into
account when interpreting their own behaviour.
Behaviour is seen to be reflective of an underlying
attitude only if there are no forces external to the
individual that might be inferred to be causing
his or her behaviour.
Finally, according to the cognitive perspective, individuals gain information about an object

368

through direct or indirect experience with it. This

information results in the formation of beliefs
about the object which links it to particular attributes. If the beliefs are mainly favourable, the
resulting attitude will be positive, and if the beliefs are mostly unfavourable, a negative attitude
towards the object will result.
It seems to make sense to consider satisfaction
as an evaluative state that can be formed through
any of the affective, behavioural, or cognitive
processes. Rather than broadening the construct
beyond recognition, it is proposed here that it is
best to define it (based on the definition of attitude proposed by Ajzen, 2001) as a summary
evaluation of an interactive system captured in
terms of how easy the user thinks it is to use. This
definition is in keeping with the conceptualisation of satisfaction both as an attitude and as a
usability parameter.
Now that a definition of satisfaction has been
offered, the ways in which the attitude
literature
can inform three important theoretical issues facing satisfaction researchers will be discussed: (1)
the nature of the user response (i.e. whether it is
cognitive or affective-emotional); (2) whether this
response is best described as a characteristic of
the interaction (quality in use) or as an intrinsic
product attribute (quality of features); and (3)
understanding the relationship between satisfaction and usage.

The Nature of the User Response

As noted in the review of the user satisfaction
literature, some view satisfaction as a cognitive
response and others as an emotional response. A
minority take a combined view. Consequently,
there is a debate as to which view is the correct
one. Consumer Psychology typically views satisfaction as an emotional response (Giese & Cote,
2000), and given that this literature has influenced
recent operationalisations of user satisfaction,
it is no surprise that definitions proposed by
Lindgaard and Dudek (2003) and Han and Hong

Measuring the Human Element in Complex Technologies

(2003) have promoted the view of satisfaction as

an exclusively emotional response to a product.
Some attitude literature in psychology suggests
that attitudes can be formed on the basis of emotional processes. Thus, these researchers could
point to this literature as providing support for
their conceptualisation of satisfaction. However,
attitude theorists also do make a clear distinction
between affect and emotion, which suggests that
satisfaction (or any attitude), cannot be described
as simply an emotional response tout court.
There are two ways in which affect, as it is
conceptualised by attitude researchers, can be
distinguished from general emotions and moods
(Judd & Johnson, 1984). First, affect relates to
a specific object (Judd & Johnson, 1984; Eiser,
1994). Second, the valence of affect is generally
conceptualised as being unidimensional, that
is, either positive or negative, while the more
general study of emotionsis full of attempts to
invoke more richly defined distinctions (Judd &
Johnson, 1984, p. 66).
This seems to clarify the distinction between
viewing satisfaction as an emotional response
and viewing it as an attitude that can be formed
on the basis of affective processes. However, it
does not resolve the issue of which set of processes
are primarily responsible for the attitude held by
an individual. Is it the affective processes (i.e.,
the feelings the object elicits) or the cognitive
processes (i.e., beliefs about that object)?
Evidence supporting both sides of the debate
can be found in the attitude literature. Proponents
of the affective perspective argue (in response to
the dominant cognitive perspective on attitudes)
that affect is more important in attitude formation than cognition because affective aspects are
more accessible in memory than cognitive aspects
(Ajzen, 2001). Their position is said to be further
strengthened by research which states that if there
is disagreement between beliefs and feelings,
(i.e., if they are of opposite valence), feelings take
precedence (Ajzen, 2001).

Proponents of the cognitive perspective, in

turn, argue that cognition (specifically awareness)
mediates affective processes. Thus, cognitive
processes are more influential in forming attitudes
than affective processes. Researchers argue that
operant conditioning effects are greater when experimental participants are aware of a contingency
between responses and reinforcement (Eagly &
Chaiken, 1993). Effects of this type are also seen
in studies on classical conditioning. Such effects
are also seen in research on the mere exposure
effect in that participants may be aware that they
should respond in a particular way to stimuli that
occur more frequently than others. However, there
is research involving subliminal presentation of
stimuli that produced both conditioning and mere
exposure effects. This suggests that conscious
stimulus recognition is not necessary for positive attitudes to form toward objects (Eagly &
Chaiken, 1993).
Eagly and Chaiken (1993, p. 421) note that
there is a lack of clear criteria to distinguish
between affective and cognitive explanations of
attitude formation. For example, some learning
theorists believe that it is not responses that are
learned in instances of classical conditioning, but
mental representations of relationships between
stimuli and responses. The authors maintain it is
somewhat misleading to refer to conditioning as
involving purely affective mechanisms; cognitive
processes, they say, can also be used to explain the
effects of classical conditioning. According to the
principles of classical conditioning, conditioned
responses occur automatically in the presence of
a conditioned stimulus, that is, they are not mediated by cognitive processes. However, Eagly and
Chaiken (1993) argue that associative network
models of memory can provide an alternative
explanation for why these responses occur. This
theory states that each object and each response
is represented by a concept node and an evaluative node, respectively, in memory, and that these
are linked by an associative bond. A conditioned
stimulus (also represented by a concept node)

369

Measuring the Human Element in Complex Technologies

that is not originally linked with the response

becomes linked with it through repeated association. A network is formed between all the concept
and evaluative nodes. Activating any one node
activates the entire network. It is in this manner
that attitudes are formed. Clearly, when the same
process can be explained using appeals to both
affective and cognitive mechanisms, it is difficult
to determine which explanation is correct.
Ajzen (2001, p. 29) notes, The debate over the
cognitive versus affective basis of attitudes has
yet to be resolved. So it seems most prudent to
follow a research strategy in which attitudes are
hypothesised to form as a result of both affective
and cognitive processes (Judd & Johnson, 1984;
Ajzen, 2001) rather than either one or the other.
The relationship between these two components
should be characterised as one of reciprocal
causality (Judd & Johnson, 1984, p. 79). For
example, objects that elicit more intense affect
are more easily remembered and judged more
quickly than those that elicit less intense affect.
In addition, affective responses influence how
the individual organises and stores information
(Judd & Johnson, 1984).
The implication of this discussion for research
on user satisfaction is that it is inappropriate to
conceptualise the satisfaction construct as an
expression of affect where affect denotes emotion only. However, conceptualising satisfaction
as an attitude does not preclude the possibility of
an affective component considering the reciprocal
relationship between affect and cognition.

Satisfaction: Quality in Use or

Quality of Features?
The second issue facing satisfaction researchers
that could benefit from application of theories
from the attitude literature is whether it is best
to view the construct as a characteristic of the
interaction between the user and the product or
as a product attribute. As mentioned earlier, it is
widely accepted that usability is best conceptu-

370

alised as a characteristic of the interaction. Given

that satisfaction is a usability parameter, one could
assume that it also would be conceptualised in a
similar manner. However, recent approaches to
investigating the construct have proposed a return to the widely discredited quality of features
approach. Quality of features implies that the
level of user satisfaction is dependent on product
features (e.g., Han & Hong, 2003; Yun et al., 2003).
This approach, referred to very insightfully as
design reductionism by Hassenzahl, Beu and
Burmester (2001), is unlikely to prove successful. Desmet and Hekkert (2002, p. 67) assert that
there is not a direct relationship between emotion
and product design, and that emotions are not
elicited by product characteristics as such, but by
construals based on these characteristics.
A similar issue arose in the attitude literature.
Research on attitudes often neglected the fact
that attitudes are social constructs, and instead,
they were generally conceptualised as individual
attributes (Manstead, 1996). However, newer conceptualisations of attitude argue

for the inclusion

of context, a variable not often considered in the
literature, but which has a significant bearing on
attitudes (Palmerino, Langer, & McGillis, 1984;
Ajzen, 2001). It is suggested that context may
condition attitude (Ajzen, 2001.)
Palmerino et al. (1984, p. 181) define an attitude as a relationship between two entities
where one entity is a person and the other is a
person or object and point out that the form of
this relationship (i.e., the attitude) can be radically
different given different contexts (p.179).
The conceptualisation of attitude proposed
by Palmerino et al. (1984) seems well suited to
user satisfaction research for two reasons. First,
because of the inclusion of context and second, because it views attitude as the relationship between
an individual and a psychological object, rather
than hypothesising that the attitude resides totally
in the person or (absurdly) totally in the object.
This bears a striking similarity to the notion of
quality in use where usability is conceptualised

Measuring the Human Element in Complex Technologies

as a property of the interaction between the individual and the device under evaluation.
This discussion suggests that viewing satisfaction as a product characteristic is inadequate
(which is what is implied by Yun et al., 2003, and
Han & Hong, 2003). Attitude research studies
provide the necessary theoretical basis for proposing that satisfaction is a characteristic of the
interaction, and support an approach such as that
followed by Porteous et al. (1995) which focuses
on user reactions to the product rather than referring to specific product features.

Investigating the Relationship

between Satisfaction and Behaviour
Satisfaction is considered to be an important
indicator of future user behaviour (Kirakowski,
2002). However, there is little empirical evidence
available to support this claim. Although Melone
(1990) cautioned that attitudes only may play a
moderating role in determining behaviour, researchers tend to oversimplify the relationship
between satisfaction and usage.
It was thought that knowledge of an individuals
attitude could be used to predict his or her behaviour (see the discussion in Eiser, 1994). However,
when studies by LaPiere (1934), DeFleur and
Westie (1958), and Wicker (1969) found there to
be little or no correlation between verbal expressions of attitude and overt behaviour, many called
for the abandonment of research into the attitude
construct altogether. Fishbein and Ajzen (1975)
maintain that poor correlations between measures
of attitude and overt behaviours were a consequence of the flawed attempt to predict specific
behaviours using general measures of attitude.
They developed their own attitude framework.
This was an attempt to standardise the terminology
used, and to clarify the relationships between the
various hypothesised components of the attitude
construct and the relationships of the attitude
construct itself with other constructs. Fishbein
and Ajzen (1975) maintain that the concepts of

beliefs, attitude, intention, and behaviour have

been used interchangeably in the literature, but
by making distinctions between them, many of
the inconsistencies found in research on attitudes
can be reduced.
In their Theory of Reasoned Action, Fishbein
and Ajzen (1975) contend that beliefs are the
fundamental building blocks that determine
attitudes. Intentions are a function of the beliefs
the individual has about the behaviour in question,
that is, the consequences of performing a specific
behaviour and the individuals evaluation of those
consequences. Whether the individual performs
the behaviour in question depends on his or her
behavioural intentions, which in turn are determined by both the individuals belief structure and
what Fishbein and Ajzen (1975) term the subjective norm: this refers to the extent to which you
believe that your acting in a given way will earn
the approval or disapproval of other people whose
opinions you value (Eiser, 1994, p. 21).
In the end, the Theory of Reasoned Action
provides a good justification for proposing that
satisfaction by itself may never be sufficient to
predict behaviour. However, through the Affective
route, attitudes must surely influence behaviour
and, therefore, we will expect to see a broad degree
of relationship between the two. The hypothesis
is, therefore, that discrepancies between satisfaction and usage must be attributable to biasing
contextual factors. It is not enough in any one case
to show that there is a poor relationship between
satisfaction and behaviour; all one needs for that
are measuring instruments of poor discriminability. If a poor relationship is found, then one must
hunt to find the biasing contextual factors.

CONCLUSION
The preceding discussion illustrates that Melone
(1990) was correct in asserting that a broad conceptualisation of satisfaction as an attitude retains
the constructs essential elements while enabling

371

Measuring the Human Element in Complex Technologies

access to the necessary theoretical foundation for

research in this area. The discussion has gone on
to attempt to uncover what these essential elements
are. It was argued that using theories and models
from the attitude literature can help with some of
the challenges faced by researchers, particularly
in constructing definitions and operationalisations of satisfaction which may lead to scales of
measurement.
The gist of this paper can be summarised in
four statements.
First, it is proposed here that the construct
should not be broadened as this changes the construct into unrecognisable forms. It will lead to
greatest consensus to define satisfaction (based on
Ajzens 2001 definition of attitude) as a summary
evaluation of an interactive system captured in
terms of how easy a user thinks a system is to use.
For HCI, satisfaction as so defined is an element of
the construct of Usability. Defining it in this manner ensures that satisfaction remains a measure of
usability as quality in use, rather than a broader
measure of the entire user experience.
Second, it is suggested that research into the
structure of the satisfaction concept is important.
This kind of research seeks to discover the main
directions along which users are thinking when
they interact with a particular class of products. It
is likely that such thinking will change over time
as technology is assimilated into the way we live.
Structural information can be used formatively
(in the development of technology), as well as
summatively (in the assessment of technology.)
Third, it is proposed that satisfaction is the
result of an evaluative state in the user that
can be formed through affective, cognitive, or
behavioural processes. These processes may be
related through emotion, but satisfaction is not
an exclusively emotional response on the part of
the user toward a product.
Finally, it is suggested that in accordance with
the conceptualisation of usability as quality of
use, the level of user satisfaction with a product is
dependent on the interaction between a user and a

372

product, and the context in which that interaction

takes place. Satisfaction should not be regarded
as being solely a product attribute.
We propose that these are four fundamental
principles in the development of scales of measurement of the concept of user satisfaction in HCI.

REFERENCES
Ajzen, I. (2001). Nature and operation of attitudes.
Annual Review of Psychology, 52, 27-58.
Bailey, J. E., & Pearson, S. W. (1983). Development
of a tool for measuring and analysing computer
user satisfaction. Management Science, 29(5),
530-545.
Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioural change. Psychological
Review, 84(2), 191-215.
Bem, D. (1972). Self-perception theory. Advances
in Experimental Social Psychology, 6, 1-62.
Bevan, N., & Azuma, M. (1997). Quality in use:
Incorporating human factors into the software
engineering life cycle. In Proceedings of the 3rd
International Conference on Software Engineering Standards. Retrieved from http://doi.ieeecomputersociety.org/10.1109/SESS.1997.595963
Bevan, N., Kirakowski, J., & Maissel, J. (1991).
What is usability? In Proceedings of the 4th International Conference on HCI, Stuttgart.
Bevan, N., & Macleod, M. (1994). Usability measurement in context. Behaviour & Information
Technology, 13, 132-145.
Brooke, J. (1996). SUS: A quick and dirty usability scale. In P. W. Jordan, B. Thomas, B. A.
Weerdmeester, & I. L. McClelland, (Eds.), Usability evaluation in industry. London: Taylor
& Francis.
Chin, J., Diehl, V., & Norman, K. (1988). Development of an instrument measuring user satisfaction

Measuring the Human Element in Complex Technologies

of the human computer interface. In Proceedings

of the SIGCHI Conference on Human Factors in
Computing Systems (pp. 213-218), Washington,
DC.

Giese, J. L., & Cote, J. A. (2000). Defining consumer satisfaction. Academy of Marketing Science
Review. Retrieved from http://www.amsreview.
org/articles/giese01-2000.pdf

Dawes, R.M., & Smith, T. L. (1985). Attitude

and opinion measurement. In G. Lindzey & E.
Aronson (Eds.), Handbook of social psychology:
Volume 1 Theory and method (pp. 509-566). New
York: Random House.

Han, S. H., & Hong, S. W. (2003). A systematic approach for coupling user satisfaction with product
design. Ergonomics, 46(13-14), 1441-1461.

DeFleur, M. L., & Westie, F. R. (1958). Verbal

attitudes and overt acts: An experiment on the
salience of attitudes. American Sociological
Review, 23(6), 667-673.
Desmet, P. M. A., & Hekkert, P. P. M. (2002).
The basis of product emotions. In W. S. Green
& P. W. Jordan (Eds.), Pleasure with products:
Beyond usability (pp.61-68). London: Taylor &
Francis.
Dillon, A. (2001). Beyond usability: Process, outcome and affect in human computer interaction.
Retrieved on June 26, 2006, from http://www.
ischool.utezas.edu/~adillon/Journals/BeyondUsability.pdf
Doll, W. J., & Torkzadeh, G. (1988). The measurement of end-user computing satisfaction. MIS
Quarterly, 12(2), 259-274.
Dzida, W., Herda, S., & Itzfeldt, W. D. (1978).
User perceived quality of interactive systems.
IEEE Transactions on Software Engineering,
4(4), 270-276.
Eagly, A. H., & Chaiken, S. (1993). The psychology of attitudes. New York: Wadsworth Group /
Thomson Learning.
Eiser, J. R. (1994). Attitudes, chaos and the connectionist mind. Oxford: Blackwell Publishers.
Fishbein, M., & Ajzen, I. (1975). Belief, attitude,
intention and behaviour: An introduction to theory
and research. Reading, MA: Addison-Wesley.

Hornbk, K. (2006). Current practice in measuring usability: Challenges to usability studies and
research. International Journal of Human-Computer Studies, 64, 79-102.
Hassenzahl, M., Beu, A., & Burmester, M. (2001).
Engineering joy. Retrieved on June 26, 2006,
from http://www.iib.bauing.tu-darmstadt.de/it@
iib/9/s1has_lo.pdf
ISO (1998). ISO 9241: Ergonomic requirements
for office work with visual display terminals, Part
11: Guidance on usability.
Ives, B. S., Olson, M., & Baroudi, J. (1983). The
measurement of user information satisfaction.
Communications of the ACM, 26, 530-545.
Jordan, P. W. (1998). Human factors for pleasure in
product use. Applied Ergonomics, 29(1), 25-33.
Jordan, P. W. (2000). Designing pleasurable products: An introduction to the new human factors.
London: Taylor & Francis.
Judd, C. M., & Johnson, J. T. (1984). The polarising effects of affective intensity. In J.R. Eiser
(Ed.), Attitudinal judgment. New York: SpringerVerlag.
Keinonen, T. (1998). One-dimensional usability:
Influence of usability on consumers product
preference. Netherlands: University of Art &
Design.
Kirakowski, J. (1996). The software usability
measurement inventory: Background and usage.
In P. W. Jordan, B. Thomas, B. A. Weerdmeester,
& I. L. McClelland (Eds.), Usability evaluation

373

Measuring the Human Element in Complex Technologies

in industry (pp. 169-178). London: Taylor &

Francis.
Kirakowski, J. (1999, Aug. 22-27). User satisfaction and user preference: What are we doing
when we measure? In Proceedings of the 8th
International Conference on Human-Computer
Interaction. Adjunct Conference Proceedings,
(pp.129-130). Munich.
Kirakowski, J. (2002). Is ergonomics empirical?
Ergonomics, 45(14), 995-997.
Kirakowski, J., Claridge, N., &Whitehand, R.
(1998). Human centred measures of success in
website design. In Proceedings of the 4th Conference on Human Factors & the Web, Basking
Ridge, NJ, Retrieved on June 26, 2006, from
http://research.microsoft.com/users/marycz/hfweb98/kirakowski/index.htm
Kirakowski, J., & Corbett, M. (1988). Measuring
user satisfaction. In Proceedings of the 4th Conference of the British Computer Society on People &
Computers IV, (pp. 329-338). Manchester.
Kirakowski, J., & Corbett, M. (1993). SUMI:
The Software Usability Measurement Inventory.
British Journal of Educational Technology, 24(3),
210-212.
Kirakowski, J., & Dillon, A. (1988). The computer
user satisfaction inventory (CUSI): Manual and
scoring key. Cork, Ireland: Human Factors Research Group.
LaLomia, M. J. & Sidowski, J. B. (1990). Measurements of computer satisfaction, literacy and
aptitudes: A review. International Journal of
Human-Computer Interaction, 2(3), 231-253.
LaPiere, R. T. (1934). Attitudes vs. actions. Social
Forces, 13(2), 230-237.
Lewis, J. (1995). IBM computer usability satisfaction questionnaires: Psychometric evaluation
and instructions for use. International Journal of
Human-Computer Interaction, 7, 57-78.

374

Lindgaard, G., & Dudek, C. (2003). What is this

evasive beast we call user satisfaction? Interacting
with Computing, 15(3), 429-452.
Logan, R. J. (1994). Behavioural and emotional
usability: Thomson Consumer Electronics. In
M. E. Wiklund (Ed.), Usability in practice. New
York: AP Professional.
Lund, A. (2001). Measuring usability with the USE
questionnaire. Usability Interface, 8(2). Retrieved
from http://www.stcsig.org/usability/newsletter/0110_measuring_with_use.html
Manstead, A. S. R. (1996). Attitudes and behaviour. In G. R. Semin & K. Fielder (Eds.), Applied
social psychology (pp. 3-29).
Melone, N. P. (1990). A theoretical assessment of
the user-satisfaction construct in information systems research. Management Science, 36, 76-91.
Nielsen, J. (1993). Usability engineering. Boston:
Academic Press.
Olson, J. M., & Zanna, M. P. (1992). Attitudes
and attitude change. Annual Review of Psychology, 44, 117-154.
Palmerino, M., Langer, E., & McGillis, D. (1984).
Attitudes and attitude change: Mindlessnessmindfulness perspective. New York: SpringerVerlag.
Porteous, M., Kirakowski, J., & Corbett, M. (1995).
SUMI user handbook. Cork: University College
Human Factors Research Group.
Preece, J., Rogers, Y., Sharp, H., & Benyon, D.
(1994). Human computer interaction. Reading,
MA: MIT Press.
Ravden, S. J., & Johnson, G. I. (1989). Evaluating
usability of human-computer interfaces: A practical method. Chicester, UK: Ellis Norwood Ltd.
Reiterer, H., & Oppermann, R. (1993). Evaluation
of user interfaces: EVADIS II. A comprehensive
evaluation approach. Behaviour & Information
Technology, 12(3), 137-148.

Measuring the Human Element in Complex Technologies

Shneiderman, B. (1998). Designing the user interface. New York: Addison-Wesley.

sponses to attitude object. Journal Social Issues,

25(4), 41-78.

Tractinsky, N. (2004). A few notes on the study

of beauty in HCI. Human-Computer Interaction,
19(4), 351-357.

Yun, M. H., Han, S. H., Hong, S.W., & Kim, J.

(2003). Incorporating user satisfaction into the
look-and-feel of mobile phone design. Ergonomics, 46(13-14), 1423-1440.

Wicker, A. W. (1969). Attitudes vs. actions: The

relationship of verbal and overt behavioural re-

Zajonc, R. B. (1980). Feeling and thinking: Preferences need no inferences. American Psychologist,
35, 151-175.

This work was previously published in International Journal of Technology and Human Interaction, Vol. 4, Issue 1, edited by B.C.
Stahl, pp. 1-15, copyright 2008 by IGI Publishing, formerly known as Idea Group Publishing (an imprint of IGI Global).

375

376

Chapter 2.2

Human-Centric Evolutionary
Systems in Design and
Decision-Making
I. C. Parmee
University of the West of England, UK
J. R. Abraham
University of the West of England, UK
A. Machwe
University of the West of England, UK

Abstract
The chapter introduces the concept of user-centric
evolutionary design and decision-support systems, and positions them in terms of interactive
evolutionary computing. Current research results
provide two examples that illustrate differing
degrees of user interaction in terms of subjective
criteria evaluation; the extraction, processing,
and presentation of high-quality information; and
the associated improvement of machine-based
problem representation. The first example relates
to the inclusion of subjective aesthetic criteria to
complement quantitative evaluation in the conceptual design of bridge structures. The second
relates to the succinct graphical presentation

of complex relationships between variable and

objective space, and the manner in which this
can support a better understanding of a problem domain. This improved understanding can
contribute to the iterative improvement of initial
machine-based representations. Both examples
complement and add to earlier research relating
to interactive evolutionary design systems.

Introduction
Uncertainty and poor problem definition are
inherent features during the early stages of design and decision-making processes. Immediate
requirements for relevant information to improve

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Human-Centric Evolutionary Systems in Design and Decision-Making

understanding can be confounded by complex

design representations comprising many interacting variable parameters. Design constraints and
multiple objectives that defy complete quantitative
representation and therefore require a degree of
subjective user evaluation further inhibit meaningful progression. Machine-based problem
representation may, initially, be based upon qualitative mental models arising from experiential
knowledge, group discussion, and sparse available
data. However, such representations, coupled with
user intuition, play a significant role in defining
initial direction for further investigation. Concepts based upon current understanding require
both quantitative and qualitative exploration to
generate relevant information that supports and
enables meaningful progress.
The chapter presents research and development
relating to powerful machine-based search and
exploration systems that, through appropriate user
interaction, allow both quantitative and qualitative
evaluation of solutions and the extraction of information from complex, poorly understood design
and decision-making domains. The integration
and capture of user experiential knowledge within
such systems in order to stimulate, support, and
increase understanding is of particular interest.
The objective is the realisation of user-centric
intelligent systems that overcome initial lack of
understanding and associated uncertainty, support an improving knowledge-base, allow the
integration of subjective judgement, and stimulate
innovation and creativity.

Interactive Evolutionary
Computation (IEC)
Interactive evolutionary computing (Takagi, 1996)
mainly relates to partial or complete human evaluation of the fitness of solutions generated from
evolutionary search. This has been introduced
where quantitative evaluation is difficult if not
impossible to achieve. Examples of application

include graphic arts and animation (Sims, 1991),

food engineering (Herdy, 1997), and hazard icon
design (Carnahan, 2004). Such applications rely
upon a human-centred, subjective evaluation of
the fitness of a particular design, image, taste, and
so forth, as opposed to an evaluation developed
from some analytic model.
Partial human interaction that complements
quantitative machine-based solution evaluation
is also evidentfor instance, the user addition
of new constraints in order to generate solutions
that are fully satisfactory within an evolutionary
nurse scheduling system (Inoue, Furuhashi, &
Fujii, 1999). Another example is the introduction
of new compounds as elite solutions into selected
evolving generations of a biomolecular design
process (Levine, Facello, & Hallstrom, 1997).
These examples utilise a major advantage of
stochastic population-based search techniques
that is, their capabilities as powerful search and
exploration algorithms that provide diverse, interesting, and potentially competitive solutions
to a wide range of problems. Such solutions can
provide information to the user which supports a
better understanding of the problem domain whilst
helping to identify best direction for future investigation (Parmee & Bonham, 1999), especially when
operating within poorly defined decision-making
environments. Extracted information supports
development of the problem representation in an
iterative, interactive evolutionary environment.
Interactive evolutionary design systems (IEDSs)
represent a human-centric approach (Parmee,
2002; Parmee, Watson, Cvetkovic, & Bonham,
2000) that generate and succinctly present information appertaining to complex relationships
between the variables, objectives, and constraints
that define a developing decision space.
In an attempt to categorise these various forms
of IEC, it is possible to view complete human
evaluation as explicit, whereas partial evaluation
and interaction are less explicit, more subtle forms
of human involvement. Completely implicit interaction occurs where users are unaware of their role

377

Human-Centric Evolutionary Systems in Design and Decision-Making

in the evolution of a system (e.g., the Web-based

tutorials of Semet, Lutton, Biojout, Jamont, &
Collet, 2003). A simple implicit/explicit spectrum
of interactive evolutionary approaches can thus
be developed (Parmee & Abraham, 2005).
The following sections concentrate on the
manner in which evolutionary search and exploration can generate high-quality information from complex design and decision-making
environments. Such information can be utilised
interactively to:

Integrating Aesthetics via

Interactive Evolutionary
Design Processes

This example brings together agent-based machine learning, evolutionary computing, and
subjective evaluation in search for aesthetically
pleasing, structurally feasible designs during the
conceptual design process. Although significant
theoretical work is evident with respect to the
inclusion of aesthetics in computer-based design,
application-based research has received less at
support the identification of high-perfortention (Moore, Miles, & Evans, 1996a, 1996b;
mance solutions where qualitative as well
Saunders, 2001).
as quantitative objectives play a major role;
Figure 1 illustrates the main components of
and
the system and the manner in which they interact.

modify and refine design problem represenThe user defines initial design requirements and
tation.
aesthetically evaluates the designs generated
by the Evolutionary Search, Exploration, and
The first example relates to aesthetic judgement
Optimisation System (ESEO) during the initial
of EC-generated designs and is closer to the more
generations. The agents have multiple tasks, which
traditional explicit interaction where user subjecinclude the creation of the initial population based
tive evaluation is evident. However, this subjective
on design requirements, the monitoring of designs
evaluation complements detailed, machine-based
for feasibility during the ESEO processes, and
quantitative evaluation. The second is current
evaluation of machine-based aesthetic criteria.
IEDS research relating to problem definition and
The ESEO identifies design solutions that can be
the iterative interactive improvement of machineconsidered high performance in terms of Strucbased design representations that sits further
tural Feasibility and Stability, Materials Cost, and
Human-Centric Evolutionary Systems in Design and Decision-Making
toward the implicit end of the spectrum.
Rule-Based Aesthetics.

Figure
1. The
user-centric
system
Figure
1. The
user-centric
system
USER

CONSTUCTION
AND
REPAIR AGENT
(C.A.R.A)

DESIGN USER INTERFACE

SETTINGS
USER EVALUATION
AND
PREFERENCES

INITIAL POPULATION
AND
MONITORING OF DESIGNS

human
378

AESTHETICS
FITNESS

EVOLUTIONARY
SEARCH,
EXPLORATION
AND
OPTIMIZATION
SYSTEM

evaluation as explicit, whereas partial

evaluation and interaction are less explicit, more
subtle forms of human involvement. Completely

AESTHETIC
PREFERENCES

DESIGNS
AND
USER
EVALUATION

AESTHETICS
AGENTS

MACHINE
LEARNING
SUB-SYSTEM

machine-based quantitative evaluation. The

second is current IEDS research relating to
problem definition and the iterative interactive

Human-Centric Evolutionary Systems in Design and Decision-Making

The project is initially considering three test

domains: bridges; liquid containers such as wine
glasses, chemical tanks, and so forth; and street
furniture in the form of bench type structures. The
following example concentrates on bridge design.
Research involves the ACDDM Lab at Bristol
UWE and the Institute of Machines and Structures
at Cardiff University. The early stages have concentrated on development of highly flexible and
robust representations of simple bridge structures
and the identification of optimal solutions via basic
evolutionary algorithms.

offer high utility with differing operators. GAs use

crossover as the main operator (Goldberg, 1989).
ESs are similar to real parameter GAs without
crossover, although crossover-like operators have
been introduced (Deb, 2001). EPs (Fogel, 1988) is
a purely mutation-based evolutionary algorithm
and therefore perhaps represents the simplest of the
three, as the selected representation does not need
to support crossover between differing variable
strings. Gas, however, require a representation that
is robust enough to handle repeated crossovers
whilst ensuring offspring feasibility. ESs could be
considered to lie somewhere between these two.
EPs have been selected in this work primarily
Representation
to overcome feasibility maintenance problems
relating to the component-based representation
Problem representation affects search efficiency
and crossover.
and evolutionary performance (Rosenman, 1997;
The initial goal was a flexible enough repGoldberg, 1989). Genetic algorithms (GAs),
resentation
to model all possible designs which
evolutionary strategies (ESs), and evolutionary
is robust enough to be manipulated by design
programming (EP) representations are genersearch and exploration processes (Machwe,
ally based on binary or real number parameter
Parmee, & Miles, 2005). A collection-based obstrings, but many alternatives are also available.
ject-oriented representation has been developed.
Component-based and hierarchical representaA single population member (chromosome) is a
tions support flexibility and robustness in terms
collection of primitive elements that represent a
of accommodating complex design entities with
design. Different elements with different design
many related sub-systems/components (e.g., Bentproperties can be included in the set of possible
ley, 2000; Cramer, 1985; Rosenman, 1996).
design primitives. The evaluation of fitness of the
It is also necessary to consider which stochastic
structure and checking the structural integrity
search process would best suit a chosen representautilises secondary properties of the particular
tion in terms of efficient negotiation of the design
primitive element type. Figure 2 further clarifies
space. As a high degree of solution search and exHuman-Centric Evolutionary Systems in Design and Decision-Making
the idea of using an object-based representation.
ploration is required, a population-based approach
A simple bridge design is basically divided into
would seem most appropriate. GAs, EPs, and ESs
Figure 2. Details of the object-based representation
Figure 2. Details of the object-based representation
P
O
P
U
L
A
T
I
O
N

Design 1
Design 2

Design P

Figure 3. Design before mutation

D
E
S
I
G
N

S
P
A
N

S
U
P
P
O
R
T

Element 1
Element 2
Element M

E
L
E
M
E
N
T

X pos: Integer
Y pos: Integer
Length: Integer
Height: Integer

Element 1
Element 2
Element N

379

O
P
U
L
A
T
I
O
N

Design 1

D
E
S
I
G
N

Design 2

Design P

S
P
A
N

Element 1
Element 2
Element M

E
L
E
M
E
N
T

X pos: Integer
Y pos: Integer
Length: Integer
Height: Integer

Element
1
Human-Centric
Evolutionary
Systems in Design and Decision-Making
U
P
P
O
R
T

Element 2

Element N

Figure 3. Design before mutation

Figure 3. Design before mutation

two separate collections. These are the span elespan

with
and without
support
require
mentbridges
collection
containing
elements
which
form
only
two basic
ofand
elementsthe
the span
of thetypes
bridge
the support angled
element
section
element
(to be elements
used as awhich
span element
collection
containing
form the
only)
andofa the
simple
rectangle
support
bridge.
In the element,
case of a which
simple,
can
be used as both
spanning
supporting
end-supported
span,a the
supportand
element
collecelement.
elements have an assumed contion will All
be empty.
stant.The
Therectangular
object-based
representation
takeof
elements
can eithercan
be part
advantage
of element
the principles
of object-oriented
a supporting
collection
or a span element
programming
such beam
as inheritancethat
if
collection. Simple
or angled beamis,span
we
want with
to add
new kind
of element,
bridges
anda without
support
require say
only
curved
span
section,
we could easily
do section
so by
two basic
types
of elementsthe
angled
extending
thebebasic
of element
element (to
used properties
as a span element
only)and
and
adding
the rectangle
extra properties
required
a simple
element,
whichfor
cana curved
be used
section.
as both a spanning and supporting element. All
Duringhave
initial
activity,
it may
be
elements
andesign
assumed
constant.
Thenot
objectdesirable
or possible to
strictly
parameterise
a
based representation
can
take advantage
of the
design.
Designs
make the transition
from abprinciples
of object-oriented
programming
such
stract
concepts to well-defined
specifications.
as inheritancethat
is, if we want
to add a new
The
object-based
representation
can
kindabove
of element,
say curved span
section, we could
represent
at all levels,
including
at theof
easily dodesigns
so by extending
the basic
properties
abstract
by using
objects/eleelementlevel
and adding
the high-level
extra properties
required
ments
and at the
well-defined level as a set of
for a curved
section.
specifications.
Functionality
can be
modified
During initial
design activity,
it may
not be
desirable or possible to strictly parameterise a
design. Designs make the transition from abstract
concepts to well-defined specifications. The above
object-based representation can represent designs
at all levels, including at the abstract level by using
high-level objects/elements and at the well-defined
level as a set of specifications. Functionality can
be modified based upon changing requirements.
Thus objects offer a straightforward way to cover
design activity at all levels. To produce the same

380

in a simple string-based chromosome would rebased

changing
requirements.
Thus ob-of
quire upon
additional
checks
to ensure consistency
jects
offer a straightforward
way atobitcover
the chromosome
is not violated once
or real
design
levels. or
Toitsproduce
the
numberactivity
is addedatorall
removed,
functionality
same
in a The
simple
string-based
chromosome
modified.
overall
system would
be overly
would
require
additional
checks to ensure concomplex
and difficult
to manipulate.
sistency of the chromosome is not violated once
a Mutation
bit or real number is added or removed, or its
functionality modified. The overall system would
beLet
overly
complex
difficult totomanipulate.
us assume
theand
chromosome
be mutated
represents the design in Figure 3.
Mutation
This is a basic beam bridge simply supported
at either end with a single intermediate support
Let
us and
assume
to B2).
be mutated
(B3)
two the
spanchromosome
elements (B1,
L is the
represents
the
design
in
Figure
3.
span and H is the maximum height of the bridge
This
is a basic
(see
Figure
4). beam bridge simply supported
at either
end
with aissingle
intermediate
support
The mutation
rule based,
and a rule
is se(B3)
and
two spanThere
elements
(B1, B2).
L is
lected
randomly.
are separate
rules
forthe
the
span
and
H
is
the
maximum
height
of
the
bridge
two arrays of elements. The supports
can onlyEvolutionary System
Human-Centric
(see
Figure
move
left or4).
right. Their height is based upon the
The mutation is rule based, and a rule is
selected randomly. There are separate rules
for the two arrays of elements. The supports
Figure
of of
a design
chromosome
Figure4.4.Structure
Structure
a design
chromosome
Introductio
can only move left or right. Their height is based

399

Design:
Span {B1, B2}
Support {B3}

Where:
B1 = {x1, y1, 1, h}
B2 = {x2, y2, 1, h}
B3 = {x3, y3, 1, h}

upon the thickness of the spanning element they

support. Hence there are only four rules for
supports: two rules for left and right movement,

Initial testing
freeform asse
bridge structu
based approac
proved a diffic
rule-based ag
ward and rapi
been develope
population of
ous repair
system perfo
optimisation a
tures.
The constr
have the task

its X value attribute decreased by two units and

simple structural criteria
the B2 object in the span array will have its
height value attribute decreased by two units.
The CARAs can currently create three kinds of
The height attribute of the support will be
bridges:
Human-Centricadjusted
Evolutionary
Systems
in Design and Decision-Making
automatically
to make
it continuous
and remove any overlap at its new position. The

simple beam bridges without supports

mutated design is shown in Figure 5.
(Type 1a);
Figure 5. Design after mutation
Figure 5. Design after mutation

400

thickness of the spanning element they support.

Hence there are only four rules for supports: two
rules for left and right movement, and two for
increasing and decreasing width. The depth of
each span element can vary, but they must have a
level upper surface and must be continuous with
no overlap or space between them. Thus for a
simple element in a span, there are just two rules,
namely to increase or decrease the span depth.
Now, for example, if the selected rule for support
(B3) is to move it left by a constant distance (say
two units) and for span to decrease thickness of
B2 support by constant units (say two units again),
then the B3 object in the support array will have
its X value attribute decreased by two units and
the B2 object in the span array will have its height
value attribute decreased by two units. The height
attribute of the support will be automatically
adjusted to make it continuous and remove any
overlap at its new position. The mutated design
is shown in Figure 5.

Introduction of Agency
Initial testing of the representation involved
freeform assembly and evolution of simple bridge
structures using GA, EP, and agent-based approaches. Evolving feasible structures proved a
difficult and lengthy process, whereas rule-based
agent assembly was straightforward and rapid.
Thus a combined approach has been developed
where agents create the initial population of
structures and provide a continuous repair

capability, and an evolutionary system performs

search, exploration, and optimisation across the
space of possible structures.
The construction and repair agents (CARAs)
have the task of assembling structures with
varying size and shape regarding span and supports. Agents are given specifications relating to
restrictions on placement of supports and types
of span section. Thus CARAs are informed of
the design environment and will create initial
population designs within it. The evolutionary
process then takes care of the SEO process, with
the CARAs keeping a check on potentially disruptive processes and repairing the structure, where
necessary, to ensure feasibility.

Simple Structural Criteria

The CARAs can currently create three kinds of
bridges:

simple beam bridges without supports (Type

1a);
simple beam bridges with supports (Type
1b); and
simple beam bridges with sloping span sections and supports (Type 2).

An initial population can consist of a mixture

of these designs which are then evaluated against
structural and aesthetic criteria. In terms of structure, solutions are assessed via simple length/depth
ratios and minimising material. Column design
is subjected to buckling criteria.
381

(1 + Fi )

system is run for 100 generations. A few members from the initial population are shown in
To ensure the overall integrity of the strucFigure 6.
ture, the above equations are used. It is evident
The initial population consists of three difthat the closer the dimensionsHuman-Centric
of the spanEvolutionary Systems in Design and Decision-Making
ferent kinds of designs: a simply supported
elements are to the ideal ratio (R), the lower the
Figure
6. Sample
of mixed
population
bridge shapes
Figure
6. Sample
of mixed
initialinitial
population
of bridgeofshapes

401

Fitness Evaluation for Type 1A

Type 1a is treated as a simple beam deflection problem under uniform distributed loading. A simple
heuristic relating to an ideal length-to-height ratio
for a span element of 20:1 is utilisedthat is, the
closer a span section is to the ideal ratio (R), the
better its fitness.
Li and Hi are the length and height of the ith
span element.
l
Fi = R ( i )
hi

(1)

1

Stability =
(1 + f i )

(2)

To ensure the overall integrity of the structure,

the above equations are used. It is evident that the
closer the dimensions of the span elements are to
the ideal ratio (R), the lower the value of Fi will
be. At the minimum, all Fis are equal to zero and
thus stability is equal to one.

Fitness Evaluation for Type 1b and

Type 2
In Type 1b the buckling in the columns due to
the weight of the loaded beam is also considered
using:
P' =

2 EI

H2

where: P = maximum possible load, E = modulus of elasticity, I = moment of inertia, and H =

column height. The load on column is determined
from the length of the beam between the end supports, calculating the loading and distributing this
loading across intermediate supports. A column
satisfying the buckling criteria can either increase
or decrease in thickness. Otherwise it can only
increase in thickness.

Example

A basic EP approach is used with a population size

of 100 solutions. Tournament selection is utilised
with a tournament size of 10, and the system is
run for 100 generations. A few members from the
initial population are shown in Figure 6.
The initial population consists of three different kinds of designs: a simply supported span, a
span with columns, and an angled span bridge.
After 100 generations the optimal designs shown
in Figure 7 are achieved. The angled span bridges
turn out to be most efficient in terms
of structuralEvolutionary System
Human-Centric
criteria (i.e., stability and material usage). The
Figure 7. Optimised bridges
Figure 7. Optimised bridges

(3)

design represe
detailed aesthe
is evaluated b
rule-based ae
Aesthetic _

382

span, a span with columns, and an angled span

bridge. After 100 generations the optimal designs shown in Figure 7 are achieved. The

where wi are w
rules (Ai = A

Human-Centric Evolutionary Systems in Design and Decision-Making

other two design types have been evolved out of

the population.

Aesthetics and User Evaluation

It is very difficult to integrate subjective aesthetic
criteria with machine-based design unless user
interaction plays a significant role. Moore et al.
(1996a, 1996b) have published significant work in
this area which has contributed to a New South
Wales RTA set of guidelines on bridge aesthetics
(Road and Traffic Authority, 2004). However, aesthetics evaluation can only be partially quantified
by generic guidelines and rules. While aesthetically pleasing shapes can be explicitly specified to
some extent, complete aesthetic evaluation must
also involve the designerthat is, valuation must
involve both rule-based and subjective factors. In
the present system the following aesthetics have
been hard coded:

symmetry of support placement (A1),

slenderness ratio (A2),
uniformity in thickness of supports (A3),
and
uniformity in thickness of span sections
(A4).

Many other such rules are currently being

evaluated and included in the work as the design
representation is developed to support detailed
aesthetic evaluation. Each aesthetic rule is evaluated by a separate aesthetic agent. The rule-based
aesthetic fitness is calculated as:
4

Aesthetic _ Fitness = wi Ai
i =1

(4)

where wi are weights for each of the aesthetic

rules (Ai = A1 to A4) which can also be modified online.
In addition, user-assigned aesthetic fitness
(Ufit) is the fitness given to a design directly by
the user on a scale of 0 to 10 (10 being the best).
The user can also mark solutions for preservation
into the next generation. Overall user evaluation
operates thus:
1.
2.

3.
4.
5.

User stipulates the frequency of user interaction (e.g., once every 10 generations).
User aesthetically evaluates a preset number of population members from the initial
population (usually the top 10 members, i.e.,
those with highest fitness regarding stability, material usage, and explicitly defined
aesthetic criteria).
The EP system runs.
Population members are aesthetically evaluated by the user every n generations.
Repeat steps 3 and 4 until user terminates
the evolutionary process.

The overall fitness function now includes

aesthetic fitness and user-assigned aesthetic
fitness. Furthermore, weights have been added
(w1 to w4) to each of the objectives which the
user can modify online to influence evolutionary
direction (See Box 1).
Figure 8 shows aesthetically pleasing crosssections after 30 generations with user evaluation
every 10 generations. The aesthetic objectives
(A1 to A4) are clearly reflected in them. The span
elements are of the same size. The supports are
of nearly uniform thickness, and their placement
is also symmetric.

Box 1.
Fitness = ( w1* Stability ) + (

w2
) + ( w3* Aesthetic _ Fitness ) + ( w4 * Ufit ) (5)
Material _ Usage

383

Human-Centric Evolutionary Systems in Design and Decision-Making

Human-Centric Evolutionary Systems in Design and Decision-Making

Figure 8. Aesthetically pleasing crossFigure

8. Aesthetically pleasing cross-sections
sections

Furthermore, due to user interaction, the optiFigure

8 shows
pleasing
crossmised
shapes
are notaesthetically
limited to angled
sections
but
sections
after
30
generations
with
user
evaluatake on a variety of different aesthetically pleasing
tion every
aestheticdefined
objecshapes
that10
notgenerations.
only satisfy The
the explicitly
tives
(A1
to
A4)
are
clearly
reflected
in
them.
aesthetic guidelines (A1 to A4) but also the implicit
The span of
elements
of the same size. The
aesthetics
the user are
(Ufit).
supports are of nearly uniform thickness, and
their placement isLearning
also symmetric.
Incorporating
Furthermore, due to user interaction, the
optimised shapes are not limited to angled secCurrent work involves the introduction of supertions but take on a variety of different aesthetivised learning taking user evaluation into account.
cally pleasing shapes that not only satisfy the
Since there is a natural classification in the deexplicitly defined aesthetic guidelines (A1 to A4)
signs (i.e., angled spans, supported beams, and
but also the implicit aesthetics of the user (Ufit).
unsupported beams), learning is attempted at two
levels. The first level determines user preference
Incorporating
for
one of the threeLearning
types of bridge design. This
is achieved by evaluating the relative difference
Current work
involvesfitness
the introduction
subetween
user-assigned
(or rank) forofeach
pervised
learning
taking
user
evaluation
into
type of design. The second level assesses what
account.
Since there
is finds
a natural
classification
kind
of features
the user
pleasing
in the difin
the
designs
(i.e.,
angled
spans,
supported
ferent designs. Figures 7 and 8 indicate that for
beams,
and
unsupported
beams),
learning
is
the
angled
spans,
there are two
features
that catch
attempted
at
two
levels.
The
first
level
deterthe eye immediately. These are the peak of the
mines (that
user ispreference
onerise
of point)
the three
bridge
the locationfor
of the
and
types
of
bridge
design.
This
is
achieved
by
the thickness of the span sections. Such features
evaluating
the into
relative
difference
between
userare
converted
fuzzy
variables
to create
an
assigned
fitness
(or
rank)
for
each
type
of
aesthetic model of the particular bridge type. For
design.
Thesection
secondthe
level
assesses
whatvariables
kind of
the
angled
following
fuzzy
features
the
user
finds
pleasing
in
the
different
are used to specify the aesthetic model:
designs. Figures 7 and 8 indicate that for the
angled
spans,
there
are two
features that catch
1.
Peak:
Left,
Central,
Right
the eye
immediately.
the peak ofLeft
the
2.
Difference
in These
Span are
Thickness:
bridge
(that
is
the
location
of
the
rise
point)
and
Thicker, Equal Thickness, Right Thicker
the thickness
of the span Low,
sections.
SuchHigh
fea3.
Average Thickness:
Medium,
tures
are
converted
into
fuzzy
variables
to
4. Column Thickness: Low, Medium, High
createUser-Assigned
an aesthetic Fitness
model of
the Low,
particular
5.
(Ufit):
Medium, High

384

bridge type. For the angled section the followcanare

be created
supported
ingSimilar
fuzzy models
variables
used tofor
specify
the
beam
spans
and
unsupported
beam
spans.
Based
on
aesthetic model:
this model a fuzzy rule generator has been implemented.
Initial
results
have been
encouraging. The
1. Peak:
Left,
Central,
Right
intention
is
that
as
search
progresses
there willLeft
be a
2. Difference in Span Thickness:
gradual
lessening
of the
degree of user
interaction
Thicker,
Equal
Thickness,
Right
Thicker
allied
with
an
increasing
degree
of
autonomous
3. Average Thickness: Low, Medium, High
machine-based
solution evaluation
involvingHigh
both
4. Column Thickness:
Low, Medium,
aesthetic
and
structural
criteria.
This
addresses
the
5. User-Assigned Fitness (Ufit): Low,
problem
of userHigh
fatigue.
Medium,
Similar models can be created for supported

Evolving
Problem
Space
beam spans andthe
unsupported
beam spans.
Based
through
Interactive
on this model a fuzzy rule generator has been
Evolutionary
Processes
implemented. Initial results
have been encourag-

ing. The intention is that as search progresses

The
of the utilisation
of usertheresecond
will beillustration
a gradual lessening
of the degree
of
centric
evolutionary
computing
relates
to
a
more
user interaction allied with an increasing degree
implicit
form of interaction.
Thissolution
is aimed evaluaprimarof autonomous
machine-based
ily
at
the
extraction
of
high-quality
information
tion involving both aesthetic and structural criteand
the succinct
presentation
of such
information
ria. This
addresses
the problem
of user
fatigue.
to the designer/decision maker in such a manner
that supports a better understanding of complex relationships
between
variables,
multiple objectives,
EVOLVING
tHE
PrObLEM
and
constraints
during conceptual
design. This
sPAcE
tHrOUGH
INtErActIVE
complements
the
initial
IED
concept
by further
EVOLUtIONArY PrOcEssEs
attempting to meld experiential knowledge and
intuition
with
powerfulofmachine-based
search,
The second
illustration
the utilisation of
userexploration,
and
information
processing.
centric evolutionary computing relates to a
Machine-based
representations
more
implicit formproblem
of interaction.
This issupport
aimed
exploration
through
the
evaluation
of
solutions
primarily at the extraction of high-quality inforagainst
relevant criteria.
Although
mationseemingly
and the succinct
presentation
ofinitial
such
representations
may
be
relatively
basic,
and
confiinformation to the designer/decision maker in
dence
model output
be low,asuch
representasuch ain manner
that will
supports
better
undertions
can
provide
essential
problem
insight
despite
standing of complex relationships between variapparent
shortfalls.
Identified
ables, multiple
objectives,
andhigh-performance
constraints dursolutions
based
upon
quantitative
criteria followed
ing conceptual design. This complements
the
by
qualitative
human
evaluation
provide an
initial IED concept by furthermay
attempting
to
indication
of
concept
viability
and
model
fidelmeld experiential knowledge and intuition with
ity.
An iterative
user/machine-based
process can
powerful
machine-based
search, exploration,
commence
where processing.
gradual improvements in underand information
standing
contribute
to the development
of better
Machine-based problem
representations
suprepresentations,
a
growing
knowledge
base,
and
the
port exploration through the evaluation of soluestablishment of computational models that support
403

Human-Centric Evolutionary Systems in Design and Decision-Making

rigorous analysisthat is, a process emerges that

supports the development of representation through
knowledge discovery.
An initial variable parameter set may be selected with later addition or removal of variables
as the sensitivity of the problem to various aspects
becomes apparent. Constraints may be treated in
the same way with the added option of softening
them to allow exploration of non-feasible regions.
Included objectives may change as significant
payback becomes apparent through a reordering
of objective preferences. Some non-conflicting
objectives may merge, whilst difficulties relating to others may require serious re-thinking
with regard to problem formulation. The initial
design space is therefore a moving feast rich in
information (Parmee, 2002).
The visualisation of variable and objective
space from cluster-oriented genetic algorithm
(COGA) output provides a variety of perspectives illustrating complex relationships (Parmee
& Abraham, 2004). This information is further
defined by data mining, processing, and visualisation techniques. The intention is to support
implicit learning and reduce complexity by supporting the designer development of a quantitative
and intuitional understanding of the problem.
This leads to the iterative model development
described above.

COGAs and the MiniCAPs Model

Cluster-oriented genetic algorithms provide a
means to identify high-performance (HP) regions
of complex conceptual design spaces and enable
the extraction of information from such regions
(Parmee, 1996). COGAs identify HP regions
through the online adaptive filtering of solutions
generated by a genetic algorithm. COGA can be
utilised to generate design information relating
to single and multi-objective domains (Parmee
& Bonham, 1999). The technique has been well
documented (see http://www.ad-comtech.co.uk/
Parmee-Publications. htm for relevant papers).

The research utilises the BAE Systems MiniCAPs model, a simplified version of a suite of
preliminary design models for the early stages
of military aircraft airframe design and initially
developed for research relating to the development of the IED concept. The model comprises
nine continuous input variables and 12 continuous output parameters relating to criteria such as
performance, wing geometry, propulsion, fuel
capacity, structural integrity, and so forth. Input
variables are:
1.
2.
3.
4.
5.
6.
7.
8.
9.

Climb Mach Number (CLMN)

Cruise Height (CH)
Cruise Mach Number (CRMN)
Gross Wing Plan Area (GWP)
Wing Aspect Ratio (WAR)
Wing Taper Ratio (WTR)
Wing Lead Edge Sweep (WLES)
Wing T/C Ratio (WTCR)
By Pass Ratio (BPR)

Identifying High-Performance
Regions Relating to Differing
Objectives
Figures 9(a), (b), and (c) show HP regions comprising COGA-generated solutions relating to three of
the 12 MiniCAPS objectives (Ferry RangeFR,
Attained Turn RateATR1, and Specific Excess
PowerSEP1) projected onto a variable hyperplane relating to two of the nine variables utilised
in the search process. This projection allows the
designer to visualise the
HP regions, identify their bounds, and subsequently reduce the variable ranges as described
in previously referenced papers. These papers
also introduce the projection of these differing
objective HP regions onto the same variable
hyperplane as shown in Figure 10 from which
the degree of objective conflict immediately becomes apparent to the designer. The emergence
of a mutually inclusive region of HP solutions
relating to the ATR1 and FR objectives indicates

385

Human-Centric Evolutionary Systems in Design and Decision-Making

Human-Centric Evolutionary Systems in Design and Decision-Making

Figure 9. COGA-generated high-performance regions relating to three differing objectives: (a) FRFerry
Range;9.(b)COGA-generated
ATR1Attained Turn
Rate; (c) SEP1Specific
Excess Power.
projectedobjectives:
onto the GWPA
Figure
high-performance
regions relating
to threeAlldiffering
(a)
(Gross
Wing
Plan
Area)/WAR
(Wing
Aspect
Ratio)
variable
hyperplane
(N.B.
Colour
versions
of
figures
FRFerry Range; (b) ATR1Attained Turn Rate; (c) SEP1Specific Excess Power. All
can be found
projected
ontoat:
thehttp://www.ad-comtech.co.uk/cogaplots.htm)
GWPA (Gross Wing Plan Area)/WAR (Wing Aspect Ratio) variable hyperplane
80

Gross wing plan area

65

50

35

20
1.5

2.625

3.75

4.875

Wing aspect ratio

(a)

(b)

(c)
a low
degree
of conflict,
whereas
HP region
conflict. However, searching through all possible
N.B.
Colour
versions
of figures
can the
be found
at: http://www.ad-comtech.co.uk/cogaplots.htm
relating to SEP1 is remote (in variable space) to
both the ATR1 and FR regions, indicating a higher
of the
conflict.
todegree
three of
12 MiniCAPS objectives (Ferry
There
is
much
information
contained inand
the
RangeFR, Attained
Turn RateATR1,
HP regions
relating
to appropriateprojected
variable ranges
Specific
Excess
PowerSEP1)
onto
for
single
objectives,
degree
of
conflict
between
a variable hyperplane relating to two of the nine
multiple objectives,
emergence
and definivariables
utilised inand
thethe
search
process.
This
tion
of
mutually
inclusive
(common)
HP
regions.
projection allows the designer to visualise the
This
representation
providesand
an excelHPgraphical
regions, identify
their bounds,
sublent
spatial
indication
of
the
degree
of
objective
sequently reduce the variable ranges as described in previously referenced papers. These

two-dimensional variable hyperplanes to visualise such information is not a feasible approach.

Recentalso
research
has resulted
in single of
graphical
papers
introduce
the projection
these
representations
that
can
present
all
variable
and
differing objective HP regions onto the same
objective
data whilst
providing
links 10
tofrom
other
variable
hyperplane
as shown
in Figure
visual
perspectives.
The
parallel
coordinate
box
which the degree of objective conflict immediplot (PCBP)
representation
in FigureThe
11 is
ately
becomes
apparent to shown
the designer.
one
such
graphic
that
provides
a
central
repository
emergence of a mutually inclusive region of HP
containing
much to
single
and multiple-objective
solutions
relating
the ATR1
and FR objecsolution
information.
tives indicates a low degree of conflict, whereas
the HP region relating to SEP1 is remote (in

386
405

Human-Centric Evolutionary Systems in Design and Decision-Making

Parallel Coordinate Box Plots

Parallel coordinate plots (Inselberg, 1985) appeared to offer potential in terms of providing a
single graphic illustrating complex relationships
between variable and objective space. Parallel
coordinate representation displays each variable
dimension vertically parallel to each other. Points
corresponding to a solutions value of that variable can then be plotted on each vertical variable
axis. It is thus possible to show the distribution
of solutions in all variable dimensions and the
correlation between different dimensions. The
disadvantage of the technique when attempting
to include multiple objectives is that the density
of the information presented hinders perception
(Parmee & Abraham, 2004, 2005). To overcome
the data density problem, three modifications
to the standard parallel coordinate representation
have been included:
1.
2.

3.

additional vertical axes for each variable so

that each objective can be represented,
an indication of the degree of HP region
solution cover across each variable range,
and
the introduction of box plots to indicate
skewness of solutions across each variable
range.

This PCBP provides a much clearer graphic

(see Figure 11). The vertical axis of each variable
is scaled between the minimum and maximum
value of the variable in the HP region solutions of
each objective. The length of the axis represents
the normalised ranges of variable values present
in an HP region. Where an HP solution set does
not fully extend across the variable range, the
axis is terminated by a whisker at the maximum
or minimum value of the variable. The colourcoded box plots relate to each objective (i.e., SEP1,
ATR1, and FR). The median is marked within the

box, and the box extends between the lower and

upper quartile values within the variable set. The
PCBP clearly visualises the skewness of solution
distribution relating to each objective in each variable dimension which provides an indication of
the degree of conflict between objectives.
For instance, it is apparent that all three objective boxes overlap in the case of variables 1, 2, 3,
6, and 9. However, significant differences in the
distribution of the boxes are evident in terms of at
least one objective where variables 4, 5, 7, and 8
are concerned. Variables 4 and 5 are Gross Wing
Plan Area and Wing Aspect Ratio. The conflict
between SEP1 and FR/ATR1 evident in Figure 10
is strongly reflected in the HP solution distribution
indicated by the whisker truncation of variable 4
in Figure 11 and in the box plots of that variable.
In terms of variable 5, the whisker terminations
relating to ATR1 and FR in Figure 11 reflect the
extent of the solution distribution across their HP
regions in Figure 10. The box plots also reflect
the relative distribution of HP solutions of all
objectives along that variable plane as illustrated
in Figure 10. Figure 12 shows a projection of the
ATR1 HP region onto the Cruise Height (variable
1) and Climb Mach No (variable 2) hyperplane.
The relatively uniform distribution of HP solutions across the hyperplane is reflected in the
appropriate variable plots of Figure 11.
The PCBP represents a single graphic from
which the designer can perceive which variables
are causing high degrees of objective conflict.
To get an alternative, very clear perspective of
these conflicts, any two of these variables can
be selected to also view the relevant graphics
similar to Figure 10. Further reinforcement can
be obtained from the perspectives explored in
the following section relating to projections
upon objective space. Improved understanding
can lead to developments of the computational
representation and to appropriate setting of objective preferences.

387

1 and FR
of conflict.
ined in the
e variable
of conflict
the emersive (comepresentadication of
However,
mensional
h informant research
sentations
ective data
l perspecbox plot
gure 11 is
ral reposiple-objec-

ts

1985) aproviding a
ationships

distribution of solutions in all variable dimensions and the correlation between different
dimensions. The disadvantage of the technique
when attempting to include multiple objectives
Human-Centric
Evolutionary
Systems Evolutionary
in Design andSystems
Decision-Making
Human-Centric
in Design and Decision-Maki
Human-Centric
is that the density of the information
presented Evolutionary Systems in Design and Decision-Making
hinders perception (Parmee & Abraham, 2004,
2005). To overcome the data density problem, three modifications to the standard parallel
between
and objective
Parallel
Figure10.
10.representation
All HP
HP regions
regions
projected
onto the
the
Figure
12.variable
Projection
of results
onto variable
Figure
All
projected
onto
Figure
12.
Projection
of space.
results
onto
Wing Aspect R
coordinate
have
been included:
coordinate
each
variGWPA/WARvariable
variable
hyperplane
1/variable
2 representation
hyperplane
for displays
Attained
Rate
GWPA/WAR
hyperplane
variable
1/variable
2 hyperplane
forTurn
Attained
and FR/ATR1
able
dimension
vertically
parallel
to
each
other.
(ATR1)
objective
Turn
Rate
(ATR1)
objective
reflected in th
1. additional vertical axes for each variable
Points corresponding to a solutions value of
cated by the w
so that each objective can be represented,
that
variable
can
then
be
plotted
on
each
vertiFigure 11 and
2. an indication of the degree of HP region
cal
variable
axis.
It
is
thus
possible
to
show
the
In terms of var
solution cover across each variable range,
distribution
of
solutions
in
all
variable
dimenrelating to AT
and
sions and the correlation between different
extent of the
3. the introduction of box plots to indicate
dimensions.
The
disadvantage
of
the
technique
HP regions in
skewness of solutions across each variwhen attempting to include multiple objectives
reflect the rela
able range.
is that the density of the information presented
of all objectiv
hinders
perception
(Parmee
&
Abraham,
2004,
illustrated in
This PCBP provides a much clearer graphic
2005).
To
overcome
the
data
density
probprojection
of
Human-Centric Evolutionary
System
(see Figure 11). The vertical axis of each
lem, three modifications to the standard parallel
Cruise Height
variable is scaled between the minimum and
coordinate representation have been included:
(variable 2) hy
distribution o
Figure 11. Parallel box plot of solution distri1. additional vertical axes for each variable
Figure 11. Parallel box plot of solution
plane
refle
Figure13.13.
Distribution
of FR
and
ATR1
Figureis 14.
bution
of each
objective
across
each variable
Figure
Distribution
of FR can
and
ATR1
solutions
variable
space)
to both
the ATR1
FR
so that
each objective
be
represented,
distribution
of each
objective
acrossandeach
plots of Figur
solutions
in
objective space
solutions aga
dimension
in2.
objective
space
regions, indicating
an indication
of the degree of HP region
variable
dimensiona higher degree of conflict.
The PCBP
There is much information contained in the
solution
cover
across
eachinvariable
range,
which the de
maximum
value
of the
variable
the HP region
v1 regions
v8
v2
v4
v5 to v6appropriate
v7
v9
v3 relating
HP
variable
and
ables are cau
solutions of each objective. The length of the
ranges for single objectives, degree of conflict
3. represents
the introduction
of box plots
to indicate
conflict. To g
axis
the normalised
ranges
of varibetween multiple objectives, and the emerskewness
of
solutions
across
each
varispective of th
able values present in an HP region. Where an
gence and definition of mutually inclusive (comable range.
variables can
HP solution
set does not fully extend across the
mon) HP regions. This graphical representarelevant graph
variable range, the axis is terminated by a
FR
tion provides an excellent spatial indicationSEP1
of
ATR1
This PCBP
provides a or
much
clearervalue
graphic
reinforcemen
whisker
at the maximum
minimum
of
the degree of objective conflict. However,
(see
Figure
11).
The
vertical
axis
of
each
spectives exp
the variable. The colour-coded box plots relate
searching through all possible two-dimensional
variable
is scaled(i.e.,
between
minimum
and
lating to proje
to
each objective
SEP1,the
ATR1,
and FR).
variable hyperplanes to visualise such informaproved unders
The median is marked within the box, and the
tion is not a feasible approach. Recent research
of the comput
box extends between the lower and upper
has resulted in single graphical representations
Figure 11. Parallel box plot of solution
propriate sett
quartile values within the variable set. The
that can present all variable and objective data
distribution of each objective across each
PCBP clearly visualises the skewness of soluwhilst providing
links toOutput
other visual
variable dimension
Projection
of COGA
on perspecto
Projection
tion distribution relating to each objective in
tives. The Space
parallel coordinate box plot
For comparative purposes, Figure 14 illustrates
Objective
Objective s
v1
v6 which
v8
each
variable
dimension
provides
an
v2
v4
v5
v7
v9
v3
(PCBP) representation shown in Figure 11 is
the
distribution
of COGA
output
and SPEA-II
2001),
which
tend
to
use
a
non-dominance
sUMMArY
indication of the degree of conflict between
one such graphic that provides a central reposi(Zitzler
et al., 2002) Pareto-front output in objecapproach.
The HP region solutions for ATR1 and FR can
The
HP region
objectives.
tory
containing
much
single and
multiple-objective For
space.
Using a standard
multi-objective
GA
be
projected
onto
objective
space
as shown in
comparative
purposes,
Figure
14
illusTheprojected
aesthetico
be
For instance, it is apparent that all three
tive solution information.
(MOGA),
it distribution
is possible toof
obtain
solutions
lying
trates
the
COGA
output
and
Figure 13. A relationship between the HP region
tential
in
Figure
13.term
A
objective boxes overlap in the case of variables
along
the Pareto-front,
but2002)
difficult
to exploreoutthe
SPEA-II
(Zitzler
et
al.,
Pareto-front
solutions and a Pareto-frontier emerges along
FR
that include
gion
solutionc
1, 2, 3, 6, and 9. However, significant differATR1
Parallel
coordinate
box Plots
SEP1
relationship
between
variable
and
the objective
the
outer edge
of the plot (Parmee
& Abraham,
put
in
objective
space.
Using
a
standard
multititative
throug
along
the
out
ences in the distribution of the boxes are evispace.
However,
it is likely that
the
designer
is
also
objective
GA
(MOGA),
it
is
possible
to
obtain
2004) despite the fact that the working principle
the system200
wi
Abraham,
dent in terms of at least one objective where
Parallel
plots to
(Inselberg,
1985) apinterested
in
solutions
thatPareto-front,
lie around particular
solutions
lying
along
the
but
diffiof
COGAcoordinate
is very different
that of evolutionary
tive
indication
ing
principle
o
variables 4, 5, 7, and 8 are concerned. Varipeared to offer potential
in (Deb,
terms of
providing
sections
of the Pareto-front.
cult
to
explore
the
relationship
between
varimulti-objective
algorithms
2001),
whicha
design
and lik
of
evolutionar
ables 4 and 5 are Gross Wing Plan Area and
single
illustrating complex
relationships
able and the objective space. However, it is
tend
to graphic
use a non-dominance
approach.
tural feasibilit

406
388

likely that the designer is also interested in

solutions that lie around particular sections of
the Pareto-front.
The COGA approach therefore provides a
good visual indication of the degree of conflict

The introd
cess also pos

How ma
should b

Human-Centric Evolutionary
Systems
in Design and
Decision-Making
Human-Centric
Evolutionary
Systems
in Design and Decision-Making

FR and ATR1

non-dominance

Figure 14 illusGA output and

areto-front outstandard multissible to obtain
front, but diffibetween variHowever, it is
o interested in
ular sections of

fore provides a
gree of conflict
nity to explore
and view their
nd the ability to
o-front relating
ation plus soluhis is in addition
objective space
ced papers. All
ents the original
mation extracon.

Figure
Distribution
of ATR1
FR solutions
Figure14.14.
Distribution
ofand
ATR1
and FR
against
SPEA-II
Pareto
front Pareto front
solutions
against
SPEA-II

The COGA approach therefore provides a

sUMMArY
AND ofcONcLUsION
good
visual indication
the degree of conflict
between objectives, an opportunity to explore
The aesthetics
revealsand
a significant
povarying
objectivework
preferences
view their eftential
in HP
terms
of the
development
systems
fect
upon
region
bounds,
and theof
ability
to
that include
criteria ranging
from purely
quangenerate
an approximate
Pareto-front
relating
to
titative
through
to purely
subjective.
the
objectives
under
investigation
plusUltimately
solutions
the system
will be required
a comparaaround
the Pareto-front.
Thistoisgive
in addition
to
tiveutility
indication
in terms
aesthetically pleasing
the
of COGA
inofsingle-objective
space
and likely
cost whilst
indicating
asdesign
described
in previous
referenced
papers.strucAll
oftural
thisfeasibility.
utility directly complements the original
The
introduction
of information
such an interactive
proIEDS
concept
regarding
extraction,
cess also poses
many questions such as:
processing,
and presentation.

How many designs from each population

should
presented
to the user?
SUMMARY be
AND
CONCLUSION

How should these be selected?

How many
evaluations
can apotential
user be
The aesthetics
work reveals
a significant
expected
to performofbefore
becoming
fain terms
of the development
systems
that include
tigued?
criteria ranging from purely quantitative through
to purely subjective. Ultimately the system will
These questions
been repeatedly
posed,
be required
to give ahave
comparative
indication
in
but seldom
successfully
addressed
within
the
terms
of aesthetically
pleasing
design and
likely
interactive
evolutionary
computing
(IEC) comcost
whilst indicating
structural
feasibility.
munity.
Our continuing
is addressing
The introduction
of suchresearch
an interactive
process
these
issues
and,
in
particular,
the
user-fatigue
also poses many questions such as:
aspect.

How many designs from each population

should be presented to the user?
How should these be selected?
How many evaluations can a user be expected
to perform before becoming fatigued?

These questions have been repeatedly posed,

but seldom successfully addressed within the
interactive evolutionary computing (IEC) community. Our continuing research is addressing
these issues and, in particular, the user-fatigue
aspect.
The integration of user preference and uservaried objective weights supports the transfer of
subjective evaluation from the user to a design/
decision-making system. In order to address
the third question above, a machine learning
system is required which learns the preferences
of the user during the interactive process. User
fatigue is perhaps the major stumbling block in
the development of successful systems. Much
work is required to develop an appropriate, fully
functioning, machine-learning sub-system. The
fuzzy rule-based learning system is not ideal,
and current work is investigating a case-based
approach.
The developing system should be seen as a
generic framework for the integration of userevaluation with any preliminary design/decisionmaking domain. The CARA-EP representation
concept should be portable across many problem
domains. Any system must significantly decrease
the load on the user as early as possible in the
evolutionary process. A multi-agent-based learning environment is therefore under investigation
that gradually replaces the subjective criteria
evaluation.
It is apparent from previous research and the
research presented here that COGA-generated
data can provide visual representations in variable
space of the degree of conflict between objectives
and excellent spatial indications of the distribution
of high-performance solution regions relating to

389

Human-Centric Evolutionary Systems in Design and Decision-Making

a number of objectives. It is also apparent that

the COGA HP solution sets, when projected onto
objective space, provide the designer with an opportunity to explore a wealth of HP solutions that
offer varying degrees of objective compromise
and a variety of design characteristics. The nondominance sorting of these solutions also provides
an approximate Pareto-frontier illustrating succinct available trade-offs. The direct mapping of
solutions between objective and variable space
facilitates an understanding of the relative utility
of solutions in terms of preferred variable ranges
and particular design characteristics.
The PCBP of Figure 11 offers a first point of call
for the designer to get an overview of the varied
information available from COGA output. The
intention is that the COGA graphical perspectives
will be available through simple menu/clicking
operations from the central PCBP image. These
differing perspectives are seen as essential aids
to understanding overall complexities relating
to the two dependant design spaces (variable vs.
objective space).
There is a wealth of information available
from COGA output relating to single objective solutions that is also inherent within the
multi-objective output. Hence the utility of the
approach should be assessed across both areas.
The information available from single-objective
HP regions has been fully discussed in previous
referenced papers.
User-centric techniques described in the chapter and variations of them are also currently being
applied in the conceptual design of submersible
vehicles (Parmee & Abraham, 2005), pharmaceutical drug design and discovery (Sharma &
Parmee, 2005), and conceptual software design
(Simons & Parmee, 2004). Details of this associated work can be found on the ACDDM Web site at
www.ad-comtech.co.uk/ACDDM_Group.htm.

390

ACKNOWLEDGMENT
The authors wish to acknowledge the contribution
to the bridge aesthetics work of Professor John
Miles of the Institute of Machines and Structures
at the University of Cardiff.

REFERENCES
Bentley, P. J. (2000, April 26-28). Exploring
component-based representationsThe secret
of creativity by evolution? In I. C. Parmee (Ed.),
Proceedings of the 4th International Conference
on Adaptive Computing in Design and Manufacture (ACDM 2000) (pp. 161-172). University of
Plymouth, UK.
Deb, K. (2001). Multi-objective optimisation
using evolutionary algorithms. New York: John
Wiley & Sons.
Carnahan, B., & Dorris, N. (2004). Identifying
relevant symbol design criteria using interactive
evolutionary computation. Proceedings of the
Genetic and Evolutionary Computing Conference
(GECCO), Seattle, USA (pp. 265-273).
Cramer, N. L. (1985). A representation for the
adaptive generation of simple sequential programs. Proceedings of the International Conference on Genetic Algorithms and Their Application, Pittsburgh (pp.183-187).
Fogel, D. B. (1988). An evolutionary approach
to the travelling salesman problem. Biological
Cybernetics, 60(2), 139-144.
Goldberg, D. E. (1989). Genetic algorithms in
search, optimization, and machine learning.
Reading, MA: Addison-Wesley.
Herdy, M. (1997). Evolutionary optimisation based
on subjective selectionevolving blends of coffee. Proceedings of the 5th European Congress
on Intelligent Techniques and Soft Computing

Human-Centric Evolutionary Systems in Design and Decision-Making

(EUFIT97), Aachen, Germany (pp. 640-644).

Inoue, T., Furuhashi, T., & Fujii, M. (1999). Development of nurse scheduling support system using
interactive evolutionary algorithms. Proceedings
of the IEEE International Conference on Systems,
Man, and Cybernetics (SMC99) (pp. 533-537).
Inselberg, A. (1985). The plane with parallel coordinates. The Visual Computer, 1, 69-91.
Levine, D., Facello, M., & Hallstrom, P. (1997).
Stalk: An interactive system for virtual molecular
docking. IEEE Computer Science Engineering
Magazine, 4(2), 55-65.
Machwe, A. Parmee, I. C., & Miles, J. C. (2005).
Overcoming representation issues when including
aesthetic criteria in evolutionary design. Proceedings of the ASCE International Conference in
Civil Engineering, Mexico [CD-ROM].
Moore, C. J., Miles, J. C., & Evans, S. J. N.
(1996a). Establishing a knowledge base for bridge
aesthetics. Structural Engineering Review, 8(2/3),
247-258.
Moore, C. J., Miles, J. C., & Evans, S. J. (1996b).
Innovative computational support in bridge aesthetics. Transport Research Record, 1549, 1-11.
Parmee, I. C. (1996). The maintenance of search
diversity for effective design space decomposition using cluster-oriented genetic algorithms
(COGAs) and multi-agent strategies (GAANT).
Proceedings of the 2nd International Conference
on Adaptive Computing in Engineering Design
and Control, Plymouth, UK (pp. 128-138).
Parmee, I. C., & Bonham, C. R. (1999). Towards the
support of innovative conceptual design through
interactive designer/evolutionary computing
strategies. Artificial Intelligence for Engineering
Design, Analysis and Manufacturing Journal,
14, 3-16.
Parmee, I., Watson, A., Cvetkovic, D., & Bonham,
C. R. (2000). Multi-objective satisfaction within

an interactive evolutionary design environment.

Journal of Evolutionary Computation, 8(2),
197-222.
Parmee, I. C. (2002). Improving problem definition
through interactive evolutionary computation.
Journal of Artificial Intelligence in Engineering
Design, Analysis and Manufacture (Special Issue:
Human-Computer Interaction in Engineering
Contexts), 16(3), 185-202.
Parmee, I. C., & Abraham, J. R. (2005). Interactive
evolutionary design. In Y. Jin (Ed.), Knowledge
incorporation in evolutionary computation (pp.
435-458). Berlin: Springer-Verlag.
Parmee, I. C., & Abraham, J. R. (2004). Supporting implicit learning via the visualisation of
COGA multi-objective data. Proceedings of the
IEEE Congress on Evolutionary Computation,
Portland (pp. 395-402).
Rosenman, M. A. (1997). An exploration into evolutionary models for non-routine design. Artificial
Intelligence in Engineering, 11(3), 287-293.
Rosenman, M. A. (1996). A growth model for
form generation using a hierarchical evolutionary
approach. Microcomputers in Civil Engineering
(Special Issue on Evolutionary Systems in Design),
11, 161-172.
Saunders, R. (2001). Curious design agents and
artificial creativity. PhD Thesis (Electronic
edition), Faculty of Architecture, University of
Sydney, Australia.
Road and Traffic Authority. (2004). New South
Wales (NSW) report: Bridge aesthetics: Design
guidelines to improve the appearance of bridges in
New South Wales. Report No. RTA/Pub 04.003.
Semet, Y., Lutton, E., Biojout, R., Jamont, Y.,
& Collet, P. (2003). Artificial ants colonies &
e-learning: An optimisation of pedagogical
paths. Retrieved from http://www.ad-comtech.
co.uk/Workshops.htm/

391

Human-Centric Evolutionary Systems in Design and Decision-Making

Simons, C. L., & Parmee, I. C. (2004). 35 years

on: To what extent has software engineering design achieved its goals? IEEE Software, 150(6),
337-350.
Sims, K. (1991). Artificial evolution for computer
graphics. Computer Graphics, ACM SIGGRAPH
Proceedings (Vol. 25, pp. 319-328).
Sharma, B., & Parmee, I. C. (2005). Drug discovery: Exploring the utility of cluster-oriented

genetic algorithms in virtual library design. Proceedings of the IEEE Congress on Evolutionary
Computation, Edinburgh, UK (pp. 668-675).
Takagi, H. (1998). Interactive evolutionary
computation: Cooperation of computational intelligence and human KANSEI. Proceedings of
5th International Conference on Soft Computing
and Information/Intelligent Systems, Japan (pp.
41-50).

This work was previously published in Handbook of Research on Nature-Inspired Computing for Economics and Management,
edited by J. Rennard, pp. 395-411, copyright 2007 by Information Science Reference, formerly known as Idea Group Reference
(an imprint of IGI Global).

392

393

Chapter 2.3

Adaptation and Personalization

of User Interface and Content
Christos K. Georgiadis
University of Macedonia, Thessaloniki, Greece

Abstract

INTRODUCTION

Adaptive services based on context-awareness are

considered to be a precious benefit of mobile applications. Effective adaptations however, have to
be based on critical context criteria. For example,
presence and availability mechanisms enable the
system to decide when the user is in a certain
locale and whether the user is available to engage
in certain actions. What is even more challenging
is a personalization of the user interface to the
interests and preferences of the individual user
and the characteristics of the used end device.
Multimedia personalization is concerned with the
building of an adaptive multimedia system that
can customize the representation of multimedia
content to the needs of a user. Mobile multimedia
personalization especially, is related with the
particular features of mobile devices usage. In
order to fully support customization processes, a
personalization perspective is essential to classify
the multimedia interface elements and to analyze
their influence on the effectiveness of mobile
applications.

Limited resources of mobile computing infrastructure (cellular networks and end user devices)
set strict requirements to the transmission and
presentation of multimedia. These constraints
elevate the importance of additional mechanisms,
capable of handling economically and efficiently
the multimedia content. Flexible techniques are
needed to model multimedia data adaptively for
multiple heterogeneous networks and devices
with varying capabilities. Context conditions
(the implicit information about the environment,
situation and surrounding of a particular communication) are of great importance.
Adaptive services based on context-awareness
are indeed a precious benefit of mobile applications: in order to improve their provided service,
mobile applications can actually take advantage
of the context to adjust their behaviors. An effective adaptation has to be based on certain context
criteria: presence and availability mechanisms
enable the system to decide when the user is in
a certain locale and whether the user is available

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Adaptation and Personalization of User Interface and Content

to engage in certain actions. Hence, mobile applications aim to adapt the multimedia content to
the different end user devices.
However, typically each and every person
receives the same information under the same
context conditions. What is even more challenging is a personalization of the user interface (UI)
to the interests and preferences of the individual
user and the characteristics of the user end device.
The goal of mobile applications is to increasingly
make their service offerings more personalized
toward their users. Personalization has the ability to adapt (customize) resources (products,
information, or services) to better fit the needs of
each user. Personalization in mobile applications
enables advanced customized services such as
alerts, targeted advertising, games, and improved,
push-based mobile messaging. In particular,
multimedia personalization is concerned with the
building of an adaptive multimedia system that
can customize the representation of multimedia
content to the needs of a user.
Multimedia personalization enlarges the
applications complexity since every individuals
options have to be considered and implemented. It
results in a massive amount of variant possibilities:
target groups, output formats, mobile end devices,
languages, locations, etc. Thus, manual selection
and composition of multimedia content is not
practical. A personalization engine is needed
to dynamically create the context-dependent
personalized multimedia content. General solution approaches concerning the personalization
engine, include personalization by transformation
(using XML-based transformations to produce
personalized multimedia documents), adaptive
multimedia documents (using SMIL-like presentation defined alternatives), personalization by
constraints (optimization problemconstraint
solving), personalization by algebraic operators
(algebra to select media elements and merge
them into a coherent multimedia presentation), or
broader software engineering approaches.
Mobile multimedia (M3) personalization es-

394

pecially, is related with the particular features of

mobile devices usage. Because of their mobility
and omnipresence, mobile devices have two characteristics worth noticing. First, users have limited
attention as they operate their mobile devices (this
is because they usually are concerned at the same
time in other tasks, (e.g., car driving)). Second,
users tend to treat their mobile devices in a quite
personal way, seeking for personal services and
personalized content. The preferences of users
are therefore noticeably affected. In many cases,
they favor content and services which do not
require transmitting large quantities of information. Thus, low-intensity content (e.g., ring tones,
weather reports, and screen icons) proved to be
very popular. This is not only because of the low
availability of mobile devices resources which
complicates the processing of large volumes of
information. Users demand further individually
customized content on the mobile Internet because
its personalization level is higher than that of the
fixed Internet.
Detailed issues concerning M3 personalization can be described, analyzing UI design issues.
Existing mobile applications offer a reasonably
easy, browser-based interface to help user access
available information or services. In order to support adaptation and personalization mechanisms
they should be also as far as possible concentrated
on the individual prerequisites of the human in
contact with it. In this chapter, after the presentation of background topics we discuss critical
issues of the mobile setting (characteristics of
mobile applications and mobility dimensions
in user interactions) that influence adaptation
and personalization technologies. Then, as an
application case, we focus on m-commerce applications and customer interfaces. All current
research studies tend to acknowledge that the
design rules of wired Internet applications are
only partially useful. They should not be directly
adopted in mobile computing area, because of
the considerably different user requirements and
device constraints. On the other hand, experi-

Adaptation and Personalization of User Interface and Content

ence gained from the fixed Internet formulated

as the well-accepted 7C framework, is always
welcomed. Hence, we classify the multimedia
interface elements and we analyze their influence on m-commerce sites effectiveness from a
personalization perspective.

BACKGROUND
Adaptation Objectives
The diversity of end device and network capabilities in mobile applications along with the
known multimedia challenges (namely, the efficient management of size, time, and semantics
parameters of multimedia), demand media content
and service to be flexible modeled for providing
easy-to-use and fast multimedia information.
Multimedia adaptation is being researched to
merge the creation of the services so that only
one service is needed to cover the heterogeneous
environments (Forstadius, Ala-Kurikka, Koivisto,
& Sauvola, 2001). Even though adaptation effects
could be realized in a variety of ways, the major
multimedia adaptation technologies are adaptive content selection, and adaptive presentation.
Examples of adaptation include down-scaling
the multimedia objects and changing the style
of multimedia presentation according to users
context conditions.
In general, adaptive hypermedia and adaptive
Web systems belong to the class of user-adaptive
systems. A user modelthe explicit representation of all relevant aspects of a users preferences,
intentions, etc.forms the foundation of all
adaptive systems (Bauer, 2004). The user model
is used to provide an adaptation effect, which is
tailoring interaction to different users. The first
two generations (pre-Web and Web) of adaptive systems explored mainly adaptive content
selection and adaptive recommendation based
on modeling user interests. Nowadays, the third
(mobile) generation extends the basis of the adaptation by adding models of context (location,

time, bandwidth, computing platform, etc.) to

the classic user models and explores the use of
known adaptation technologies to adapt to both
an individual user and the context of the users
work (Brusilovsky & Maybury, 2002).

Personalization Objectives and

Mechanisms
Personalization is a special kind of adaptation of
the UI which focuses on making a Web application
more receptive to the unique and individual needs
of each user (Cingil, Dogac, & Azgin, 2000). Personalization mechanisms presuppose two phases.
First, the accumulation of user information, in
order to build up a profile that illustrates a set
of descriptors essential to administrators (e.g.,
visitors interest, navigation paths, entitlements
and roles in an organization, purchases, etc.). The
second phase is the analysis of user information
to recommend actions specific to the user.
To develop the best recommendation, rulebased practices (allowing administrators to
specify principles for their applications to thrust
personalization) are usually combined with
filtering algorithms which analyze user profiles
(Pierrakos, Paliouras, Papatheodorou, & Spyropoulos, 2003). Simple filtering techniques are
based on predefined groups of users, classifying
their accounts by age groups, asset value etc.
Content-based filtering can be seen as locating
objects comparable to those a user was fond of
in the past. Finally, collaborative filtering builds
up recommendations by discovering users with
similar preferences.

CONTENT ADAPTATION AND

PERSONALIZED USER INTERFACE
Analyzing Mobile Setting
The characteristics of the mobile Internet applications can be appreciated from three different

395

Adaptation
and Personalization of User Interface and Content
Adaptation and Personalization of User Interface
and Content

Figure 1.
1. Analyzing
Figure
Analyzingmobile
mobilesetting
setting
adaptation &
personalization

adapti ve systems (mobile generation)

context mode l
user model

usage of mobile de vices

limited attention

temporal
mobility

personal character

spatial mob ility

context-sensitivity

contextual
mobility

instant
connectivity

constraints
of devices &
infrastructure

environmental
perspective

system
perspective

mobility dimensions
in user interactions

characteristics of
mobile applications

viewpoints: system, environment, and user (Chae

develop
best
rule& To
Kim,
2003). the
From
therecommendation,
systems viewpoint,
mobased
practices
(allowing
administrators
to
bile applications present disadvantages, because
specify
principles
for
their
applications
to
thrust
they provide a lower level of available system
personalization)
are usually
filresources. Mobile
devices,combined
especiallywith
cellular
tering
algorithms
which
analyze
user
profiles
phones, have lower multimedia processing ca(Pierrakos,
Paliouras, input/output
Papatheodorou,
&
pabilities, inconvenient
facilities
Spyropoulos,
2003).
Simple
filtering
techniques
(smaller screens/keyboards), and lower network
are
based on predefined
groups
of users,
clasconnection
speeds than
desktop
computers.
sifying
their
accounts
by
age
groups,
asset
However, from the environmental viewpoint
value
filtering
be seen
there etc.
is anContent-based
uncontested benefit:
theycan
enable
users
astolocating
objects
comparable
to
those
a user
access mobile Internet content anywhere
and
was
fond
of
in
the
past.
Finally,
collaborative
anytime. The term instant connectivity is used
filtering
builds
up recommendations
by discovfor mobile
browsing
to describe actually
the fact
ering users with similar preferences.

396

that it is possible to do it at the moment of need.

cONtENt
ADAPtAtION
ANDmust be
Users perspective
characteristics
PErsONALIZED
UsEr
regarded rather differently,
because they are to a
INtErFAcE
certain degree consequences of the system and

of the environment. In addition, the multi-dimen-

Analyzing
Mobile
setting
sioned concept
of mobility
influences on them

in many ways. Mobile users perform their tasks in

The
characteristics
the
mobileDifferent
Internetterms
apterms
of place, timeof
and
context.
plications
can
be
appreciated
from
three
differare used by the research community to describe
ent
viewpoints:
system,
andwithin
user
users
mobile setting,
andenvironment,
their interactions
(Chae
Kim,
2003). at
From
the systems
it, but&
these
converge
the ones
describedviewbelow
point,
mobile
applications
present
(Kakihara & Sorensen, 2001; Lee &disadvanBenbasat,
tages,
2004):because they provide a lower level of
available system resources. Mobile devices,
especially cellular phones, have lower multime-

269

Adaptation and Personalization of User Interface and Content

Spatial mobility denotes mainly the most immediate dimension of mobility, the extensive
geographical movement of users. As users
carry their mobile devices anywhere they
go, spatiality includes the mobility of both
the user and the device
Temporal mobility refers to the ability of
users for mobile browsing while engaged
in a peripheral task
Contextual mobility signifies the character
of the dynamic conditions in which users
employ mobile devices. Users actions are
intrinsically situated in a particular context
that frames and it is framed by the performance of their actions recursively

Because of their mobility (and in correspondence with its dimensions), we distinguish three
attributes regarding mobile device usage:
1.

2.

3.

Users have a tendency to treat their mobile

device in a quite personal and emotional
way (Chae & Kim, 2003). They prefer to
access more personalized services when
are involved in mobile browsing. Spatial
mobility must be considered as the major
reason behind this behaviour, which is quite
normal considering users perspective: the
mobile phone is a portable, ubiquitous and
exposed on everybodys view gadget, able
to signify users aesthetic preferences and
personality
Users have limited attention as they manage
their mobile devices (Lee & Benbasat, 2004).
This is because they usually are involved
at the same time in other tasks (e.g., walking). Temporal mobility is the reason of this
phenomenon
Users manage their mobile devices in broadly
mixed environments that are relatively unsteady from one moment to the next. Contextual mobility requires context-sensitivity
on mobile device operations. So, mobile
device is able to detect the users setting

(such as location and resources nearby) and

subsequently to propose this information to
the mobile application. In this way, mobile
device practically may offer task-relevant
services and information.

Application Case: User Interfaces in

M-Commerce Applications
Mobile Commerce Applications
The mobile sector is creating exciting new opportunities for content and applications developers. The use of wireless technologies extends
the nature and scope of traditional e-commerce
by providing the additional aspects of mobility
(of participation) and portability (of technology)
(Elliot & Phillips, 2004). One of the most rapidly
spreading applications within the m-commerce
world is the mobile Internet: the wireless access to
the contents of the Internet using portable devices,
such as mobile phones. Undoubtedly, delivering personalized information is a critical factor
concerning the effectiveness of an m-commerce
application: the organization knows how to treat
each visitor on an individual basis and emulate
a traditional face-to-face transaction. Thus, has
the ability to treat visitors based on their personal qualities and on prior history with its site.
M-commerce applications support mechanisms
to learn more about visitor (customer) desires,
to recognize future trends or expectations and
hopefully to amplify customer loyalty to the
provided services.

Personalized Multimedia in Interfaces

of M-Commerce Applications
The goal of adaptive personalization is to increase
the usage and acceptance of mobile access through
content that is easily accessible and personally
relevant (Billsus, Brunk, Evans, Gladish, & Pazzani, 2002). The importance of interface design
has been commonly acknowledged, especially

397

(Kristoffersen & Ljungberg, 1999). There

icons, ring melodies, etc.). These multiis also an additional point to take into
media UI elements are certainly used by
consideration regarding M3 presentation
mobile users in order to make their phones
adaptation and
personalization:
to
more personal
Adaptation and Personalization
of User
Interface and how
Content

Figure 2.2.Mobile
Mobile
multimedia
adaptation/personalization
categories
multimedia
adaptation/personalization
categories
M3 content
perception of
quality
usage
patterns
presence &
availability
personal
profile

M3 presentation

quality of
multimedia

selection of
multimedia
items

personal
profile

aesthetic
elements

limited
attention
limitat ions of
screen space
presence &
availability

operational
elements

M3 communic ation
personal
profile

between users

perception
of quality
usage
patterns
presence &
availability

regarding mobile devices adoption: interfaces

characteristics had been identified as one of the two
broad factors (along with network capabilities),
affecting the implementation and acceptance of
mobile phones emerged (Sarker & Wells, 2003).
Devices adoption is a critical aspect for the future
of m-commerce, because without widespread
proliferation of mobile devices, m-commerce can
not fulfill its potential.
Lee and Benbasat (2004) describe in detail
the influence of mobile Internet environment to
the 7C framework for customer interfaces. This
framework studies interface and content issues
based on the following design elements: customization (sites ability to be personalized), content
(what a site delivers), context/presentation (how
is presented), connection (the degree of formal
linkage from one site to others), communication
(the type of dialogues between sites and their us-

398

between
site & users

ers), community (the interaction between users),

273
and commerce (interface elements that support
the various business transactions) (Rayport &
Jaworski, 2001). A generic personalized perspective is presented in (Pierrakos et al., 2003) with
a comprehensive classification scheme for Web
personalization systems. Based on all these works,
we focus on multimedia design issues concerning
personalized UIs for m-commerce applications.
We present a reconsideration of the 7C framework
from an M3 customization aspect, in which we
distinguish the following mobile multimedia
adaptation/personalization categories:
M3 content is the main category. It contains
the parts of the 7Cs content and commerce
design elements, which deal with the choice of
media. Multimedia mix is the term that is
used in 7C framework regarding exclusively the
content element. However, in our approach

Adaptation and Personalization of User Interface and Content

multimedia elements regarding shopping carts,

delivery options etc. are also belong here because
they share a lot of commons concerning adaptation and personalization. It is commonly accepted
that large and high visual fidelity images, audio
effects, and motion on interfaces are multimedia
effects which might lead to a higher probability
of affecting users decisions in e-commerce environments (Lee & Benbasat, 2003). However, in
m-commerce setting things are different because
we can not assume that the underlying communication system is capable of delivering an optimum
quality of service (QoS). The bandwidth on offer
and the capabilities of devices are setting limitations. Therefore, a central issue to the acceptance
of multimedia in m-commerce interfaces is the
one of quality. The longer the response delay,
the less inclined will the user be to visit that specific m-commerce site, resulting in lost revenue.
Obviously, an end-to-end QoS over a variety of
heterogeneous network domains and devices is
not easily assured, but this is where adaptation
principle steps in.
Dynamic content adaptation of the media
quality to the level admitted by the network is
a promising approach (Kosch, 2004). Content
adaptation can be accomplished by modifying
the quality of a media object (resolution and its
play rate); so, it can be delivered over the network
with the available bandwidth and then it can be
presented at the end device (satisfying its access
and user constraints).
An essential issue for effective content adaptation is the perceptual quality of multimedia.
Quality of perception (QoP) is a measure which
includes not only a users satisfaction with multimedia clips, but also his ability to perceive, analyze, and synthesize their informational content.
When a personalization engine is called out to
adapt multimedia content, the perceptual impact
of QoS can be extremely valuable, and it can be
summarized by the following points (Ghinea &
Angelides, 2004):

Missing a small number of media units will

not be negatively perceived, given that too
many such units are not missed consecutively
and that this incident is infrequent
Media streams could flow in and out of
synchronization without substantial human
displeasure
Video rate variations are tolerated much
better than rate variations in audio
Audio loss of human speech is tolerated
quite well
Reducing the frame rate does not proportionally reduce the users understanding
(user has more time to view a frame before
changes)
Users have difficulty absorbing audio, textual and visual information concurrently, as
they tend to focus on one of these media at
any one moment (although they may switch
between the different media)
Highly dynamic scenes have a negative impact on user understanding and information
assimilation

Another important issue regarding M3 content

adaptation (both for quality and for the selection of
media items), is the usage patterns for the mobile
Internet. Users purchase more low-risk products
(e.g., books) than high-risk ones, because they can
not pay full attention to their interactions with
mobile devices. Also, users tend to subscribe to
content with low information intensity more than
to content with high information intensity (e.g.,
education), because mobile devices have inferior
visual displays.
Devices constraints and personalization
requirements emphasize the need for additional
effective content adaptation methods. Personalization mechanisms allow customers to feel
sufficiently informed about products and services they are interested in, despite the limited
multimedia information delivered by a restricted
display device. They can be considered as filters
which reject the delivery of multimedia content

399

Adaptation and Personalization of User Interface and Content

that users dont appreciate. More and more, mobile

applications exploit positioning information like
GPS to guide the user on certain circumstances
providing orientation and navigation multimedia
information, such as location-sensitive maps. To
facilitate personalized adaptation, multimedia
content is desirable to include personalization and
user profile management information (in the form
of media descriptors) (Kosch, 2004). In this way,
adaptive systems can utilize information from
the context (or user model) in use. Especially
personalized UIs are able to exercise all kinds
of personalization mechanisms (rule-based practices and simple, content-based or collaborative
filtering), to locate or predict a particular users
opinion on multimedia items.
M3 presentation is also an important adaptation/personalization category. It contains
all parts from 7Cs context, commerce, and
connection design elements related to multimedia presentation. M3 presentation refers to the
following aspects:

400

The aesthetic nature of multimedia in

interfaces (i.e., the visual and audio characteristics such as color schemes, screen
icons, ring melodies, etc.). These multimedia
UI elements are certainly used by mobile
users in order to make their phones more
personal
The operational nature of multimedia in
interfaces, including internal/external link
issues and navigation tools (in what ways
the moving throughout the application is
supported). An important issue here deals
with the limited attention of users when
interacting with their mobile devices. So,
minimal attention interface elements, able
to minimize the amount of user attention
required to operate a device are welcomed.
For example, utilizing audio feedback in
order to supplement users limited visual
attention is considered in general a desirable
approach in mobile setting (Kristoffersen &

Ljungberg, 1999). There is also an additional

point to take into consideration regarding M3
presentation adaptation and personalization:
how to overcome the limitations due to the
lack of screen space. Certainly, visual representations of objects, mostly through graphic
icons, are easier to manipulate and retain
than textual representations. But, small
screens need not set aside a large portion of
their space for infrequently used widgets.
In this context, potential adaptations can be
made by substituting visual elements with
non-speech audio cues (Walker & Brewster,
1999), or by using semi-transparent screenbuttons, that overlap with the main body of
content in order to make the most of a small
screen (Kamba, Elson, Harpold, Stamper,
& Sukaviriya, 1996).
All users are not having the same context
conditions and preferences. Personalization
mechanisms are used for both the aesthetic and
the operational nature of multimedia in interfaces.
Obviously, multimedia personalization engine
must be able to provide context-sensitive personalized multimedia presentation. Hence, when
a distracting user setting is acknowledged, the
adapted multimedia presentations on the interface
should call for only minimal attention in order to
complete successfully critical transaction steps.
Moreover, context-awareness of mobile devices
may influence M3 presentation adaptation/personalization regarding connection issues. Indeed,
the recommendation of a particular external link
among a set of similar ones may depend not only
from its content, but also from its availability
and efficiency under the specific conditions of
users setting.
M3 communication contains all parts from
7Cs communication and community design
elements related to multimedia. In our approach,
they belong to the same adaptation/personalization category because they deal with multimedia
enriched communication and interaction services.

Adaptation and Personalization of User Interface and Content

Mobile devices are inherently communication

devices. Location and positioning mechanisms
provide precise location information, enabling
them to better interact with applications to deliver greatly targeted multimedia communication
services. The perceptual quality of multimedia
and relative previously discussed issues, are also
important factors for effective multimedia communication adaptation.
With M3 communication personalization,
m-commerce administrators are able to make
use of information about users mobile setting
to catch the right type of multimedia communication for the right moment (taken into account
also the preferences of each user about the most
wanted type of communication between him or
her and the site). In addition, supporting adaptive (interactive or non-interactive) multimedia
communication between users enables opinion
exchange about current transactions and network
accesses. Undoubtedly, such functionality may
provide useful information for collaborative filtering techniques, resulting in more successful
personalized sites.

and personalization processes, it appears as an

interesting research opportunity.
In a multi-layered wireless Web site, more
sophisticated adaptation and personalization
mechanisms are introduced as we get closer to the
database layer. From that point of view, multimedia database management system (MMDBMS)
emerging technology may support significantly the
(mobile) multimedia content adaptation process.
Existing multimedia data models in MMDBMSs
are able to partially satisfy the requirements of
multimedia content adaptation because contain
only the basic information about the delivery of
data (e.g., frame rate, compression method, etc.).
More sophisticated characteristics such as the
quality adaptation capabilities of the streams are
not included. This information would be of interest
to the end user. Consequently, a lot of research
deals with extending the functionalities of current
MMDBMSs by constructing a common framework for both the quality adaptation capabilities
of multimedia and for the modeling/querying of
multimedia in a multimedia database (Dunkley,
2003; Kosch, 2004).

FUTURE TRENDS

CONCLUSION

Providing adaptation and personalization affects

system performance, and this is an open research
issue. A basic approach to improve performance
is to cache embedded multimedia files. However,
when personalized multimedia elements are used
extensively, multimedia caching can not maximize
performance. The trend is therefore to provide
personalization capabilities when server-usage
is light and disallow such capabilities at periods
of high request. Alternatively, users can have a
personalized experience, even at times of high
system load, if they pay for the privilege (Ghinea
& Angelides, 2004). In any case, the design of
a flexible context (or user) model, capable of
understanding the characteristics of mobile setting in order to facilitate multimedia adaptation

The advances in network technology, together

with novel communication protocols and the
considerably enhanced throughput bandwidths
of networks, attracted more and more consumers
to load or stream multimedia data to their mobile
devices. In addition, given the limited display
space, the use of multimedia is recommended so
that display space can be conserved. However,
mobile settings limitations regarding multimedia
are serious. In fact, enhancing the mobile browsing user experience with multimedia is feasible
only if perceptual and contextual considerations
are employed.
The major conclusion of previously presented
issues is that efficient delivery, presentation and
transmission of multimedia has to rely on context-

401

Adaptation and Personalization of User Interface and Content

sensitive mechanisms, in order to be able to adapt

multimedia to the limitations and needs of the
environment at hand, and even more to personalize
multimedia to individual users preferences.

REFERENCES
Bauer, M. (2004). Transparent user modeling for
a mobile personal assistant. Working Notes of the
Annual Workshop of the SIG on Adaptivity and
User Modeling in Interactive Software Systems
of the GI (pp. 3-8).
Billsus, D., Brunk, C. A., Evans, C., Gladish, B.,
& Pazzani, M. (2002). Adaptive interfaces for
ubiquitous Web access. Communications of the
ACM, 45(5), 34-38.
Brusilovsky, P., & Maybury, M. T. (2002). From
adaptive hypermedia to adaptive Web. Communications of the ACM, 45(5), 31-33.
Chae, M., & Kim, J. (2003). Whats so different
about the mobile Internet? Communications of
the ACM, 46(12), 240-247.
Cingil, I., Dogac, A., & Azgin, A. (2000). A
broader approach to personalization. Communications of the ACM, 43(8), 136-141.
Dunkley, L. (2003). Multimedia databases. Harlow, UK: Addison-WesleyPearson Education.
Elliot, G., & Phillips, N. (2004). Mobile commerce
and wireless computing systems. Harlow, UK:
Addison WesleyPearson Education.
Forstadius, J., Ala-Kurikka, J., Koivisto, A., &
Sauvola, J. (2001). Model for adaptive multimedia
services. Proceedings SPIE, Multimedia Systems,
and Applications IV (Vol. 4518).
Ghinea, G., & Angelides, M. C. (2004). A user
perspective of quality of service in m-commerce.
Multimedia Tools and Applications, 22(2), 187206. Kluwer Academic.

402

Kakihara, M., & Sorensen, C. (2001). Expanding

the mobility concept. ACM SIGGROUP Bulletin, 22(3), 33-37.
Kamba, T., Elson, S., Harpold, T., Stamper, T., &
Sukaviriya, P. (1996). Using small screen space
more efficiently. In R. Bilger, S. Guest, & M. J.
Tauber (Eds.), Proceedings of CHI 1996 ACM
SIGCHI Annual Conference on Human Factors
in Computing Systems (pp. 383-390). Vancouver,
WA: ACM Press.
Kosch, H. (2004). Distributed multimedia database technologies. Boca Raton, FL: CRC Press.
Kristoffersen, S., & Ljungberg, F. (1999). Designing interaction styles for a mobile use context. In
H. W. Gellersen (Ed.), Proceedings of the 1st International Symposium on Handheld and Ubiquitous
Computing (HUC 99) (pp. 281-288).
Lee, W., & Benbasat, I. (2003). Designing an
electronic commerce interface: Attention and
product memory as elicited by Web design. Electronic Commerce Research and Applications,
2(3), 240-253. Elsevier Science.
Lee, Y. E., & Benbasat, I. (2004). A framework for
the study of customer interface design for mobile
commerce. International Journal of Electronic
Commerce (1086-4415/2004), 8(3), 79-102.
Pierrakos, D., Paliouras, G., Papatheodorou, C.,
& Spyropoulos, C. (2003). Web usage mining as
a tool for personalization: A survey. User Modeling and User-Adapted Interaction, 13(4), 311-372.
Kluwer Academic.
Rayport, J., & Jaworski, B. (2001). Introduction
to e-commerce. New York: McGraw-Hill.
Sarker, S., & Wells, J. D. (2003). Understanding
mobile handheld device use and adoption. Communications of the ACM, 46(12), 35-40.
Walker, A., & Brewster, S. (1999). Extending the
auditory display space in handheld computing devices. Proceedings of the 2nd Workshop on Human
Computer Interaction with Mobile Devices.

Adaptation and Personalization of User Interface and Content

KEY TERMS
Content Adaptation: The alteration of the
multimedia content to an alternative form to meet
current usage and resource constraints.
MMDBMS: Multimedia database management system is a DBMS able to handle diverse
kinds of multimedia and to provide sophisticated
mechanisms for querying, processing, retrieving,
inserting, deleting, and updating multimedia.
Multimedia database storage and content-based
search is supported in a standardized way.
Personalization: The automatic adjustment of
information content, structure and presentation
tailored to an individual user.

QoS: Quality of service notes the idea that

transmission quality and service availability can
be measured, improved, and, to some extent, guaranteed in advance. QoS is of particular concern
for the continuous transmission of multimedia
information and declares the ability of a network
to deliver traffic with minimum delay and maximum availability.
Streaming: Breaking multimedia data into
packets with sizes suitable for transmission between the servers and clients, in order to allow
user to start enjoying the multimedia without
waiting to the end of transmission.
UI: (Graphical) user interface is the part of
the computer system which is exposed to users.
They interact with it using menus, icons, mouse
clicks, keystrokes and similar capabilities.

This work was previously published in Handbook of Research on Mobile Multimedia, edited by I. Ibrahim, pp. 266-277 , copyright 2006 by Information Science Reference, formerly known as Idea Group Reference (an imprint of IGI Global).

403

404

Chapter 2.4

User Interaction and Interface

Design with UML
Jesus M. Almendros-Jimenez
Universidad de Almeria, Spain
Luis Iribarne
Universidad de Almeria, Spain

Abstract

INTRODUCTION

This chapter will show you how to use and specialise UML diagrams for describing the user
interfaces of a software system. In order to accomplish the description of user interfaces, the
proposed technique considers three specialised
UML diagrams called user-interaction, user-interface, and GUI-class diagrams, which will be built
following a model-driven development (MDD)
perspective. These diagrams can be seen as the
UML-based UI models of the system. In addition,
this chapter is concerned with code-generation
to implement the user interfaces of the system
by using GUI-class diagrams and user-interaction diagrams. A case study of an Internet book
shopping system is introduced in this chapter to
proof and illustrate the proposed user interaction
and interface design technique.

The emergence of the unified modelling language

(UML) (OMG, 2005) as an industry standard for
modelling systems has encouraged the use of
automated software tools that facilitate the development process from analysis through coding.
The user interface (UI), as a significant part of
most applications, should also be modelled using
UML. UML diagrams could be used to model
user interfaces, and automatic CASE tools could
help to generate code for user interfaces from
UML designs. In general terms, visual modelling allows the developers to visualize source
code in a graphical form: graphical abstractions,
such as flow charts to depict algorithmic control flows and structure charts or simple block
diagrams with boxes representing functions and
subprograms, and so on. UML provides system

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

User Interaction and Interface Design with UML

architects with a visual language for specifying,

constructing, and documenting the artefacts of
software systems. In particular, user interfaces
should be visually modelled in order to describe
the behaviour of the window system in response
to user interactions.
This chapter is firstly devoted to show how
to use and specialise UML diagrams in order to
describe the user interface and user interactions of
a software system, following a particular modeldriven development (MDD) perspective. Modeldriven development involves creating models
through a methodological process that begins
with requirements and looks into a high-level
architectural design. Model-driven development
facilitates and improves the software analysis and
design and code generation facilities from models
prevent the loss of substantial information during
the transition of a model to its implementation.
In our MDD perspective, we consider the
following steps for user interface design and
modelling:
1.
2.

3.

4.

Firstly, we use a UML use case diagram for

extracting the main user interfaces.
Secondly, we describe each use case by
means of a special kind of UML activity
diagrams, called user-interaction diagrams,
whose states represent data output actions
and transitions represent data input events.
This perspective allows the designer to
model the user interaction (i.e., input-output
interaction) in each main user interface.
Thirdly, each input and output interaction
of the user-interaction diagrams allows the
designer to extract GUI components used
in each user interface. Therefore, we can
obtain a new and specialized version of the
use case diagram representing the user interface design, and a class diagram for GUI
components: user-interface and GUI-class
diagrams, respectively.
The user-interaction, user-interface, and
GUI-class diagrams can be seen as the

UML-based user interface models of the

system.
This chapter will also deal with code generation techniques. In our MDD perspective, the
UML-based user interface models can be used
for generating executable code with the following advantages:
1.

2.

3.
4.

Rapid prototyping of the developed software: Software modellers would find it useful to quickly generate user interfaces from
high-level descriptions of the system.
Model validation and refinement: Prototyping can detect fails in design and refinement and validation of model by testing user
interfaces and user requirements.
Model-based code generation: Generated
code would fit with developed models.
Starting point for implementers: Prototypes
can be refined until final implementation.

BACKGROUND
In the literature there are some works dealing with
the problem of user interfaces in UML.

Use Cases and UI Design

Some of these works (CommentEdition, 2000;
Constantine & Lockwood, 2001; Nunes & Falcao,
2001; Paterno, 2001) are focused on the utilisation
of UML use case diagrams as a starting point of
the user interface design, or even as a high-level
description of the structure of the user interface.
However, there are some considerations about
the use case diagram style. Following the UML
philosophy, a use case diagram could not be suitable for extracting the user interfaces. Use case
diagrams may include some use cases referred to
parts of the system not related to user interfaces
such as classes, human tasks, components of other
systems interacting with us, and so on. Or even

405

User Interaction and Interface Design with UML

decomposing use cases by means of include and

extend relationships, one could specify specific
parts of the system, which are not related with
the user interface. Therefore, in our opinion, a
specialized version of the use case model could
be required, or even some other UML diagrams
could be used to complete the use case view.

Adapting UML to UI Design

Other works (Campos & Nunes, 2005; Conallen,
1999; Heumann, 2003; Lieberman, 2004; Nunes,
2003; Nunes et al., 2001; Pinheiro da Silva & Paton, 2000) propose the introduction of new UML
elements in order to model user interfaces. In this
case, the need of new stereotypes or classifiers
in UML in order to distinguish GUI components
from other system components is well known.
In this case, there is a common point of view:
classes should be stereotyped with <<screen>>,
<<applet>>, and <<input form>>, to be
distinguished from data classes.
In addition, there are some works (Anderson,
2000; Elkoutbi & Keller, 2000) (Elkoutbi, Khriss,
& Keller, 1999) interested in the description of the
logic of GUI components, using activity and state
diagrams for that. In this case, user interaction can
be specified by means of states, where the system
shows output data and requests input data. It should
be specified how the user can navigate through the
user interface, opening, closing windows, picking
up in links, and so on. The specified logic should
describe the set of states (input, output, windows)
in which the user can be found.

Code Generation and UML

With respect to code generation, unfortunately,
the capabilities of code generator CASE tools to
transform design to an implementation are often
restricted to produce class definitions consisting
of attributes and operation signatures captured in
class diagrams, but not methods to implement the
procedural flow within the operations. Existing

406

approaches in this last sense turn statecharts into

executable code. Statecharts are used as object
controllers for specifying when an object is willing
to accept request. CASE tools supporting code
generation from statecharts are Statemate (Ilogix,
2006a), Omate (Harel & Gery, 1997), Rhapsody
(Ilogix, 2006b), and Fujaba (Schfer, 2002).

MAIN THRUST OF THE CHAPTER

With regard to previous works on UML (use
cases) and UI design, our contribution can be
summarized as follows:

Firstly, we consider use cases as a starting

point of the user interface design. Here we
define a new kind of diagram, user-interface
diagram: a specialized version of the use
case diagram as a high-level description of
the user interface.
In addition, following the MDD perspective,
we integrate this system view with a set of
specialized activity diagrams for user interaction design (user-interaction diagrams).
One of the benefits of this integration is
that each use case in the specialized use
case diagram is described by means of a
specialized activity diagram, and therefore
interfaces can be analyzed and built from
these two models.
In addition, a class diagram is generated
from the specialized activity diagrams. GUI
components can also be built from both
modelling techniques.

With regard to previous works on code generation, the UI models can get prototypes of the user
interface of our application. Through mapping
between UML and Java, we are able to generate
low-level Java code directly from the user interaction diagram. This code generation is adapted to
the special case of user interfaces, which is user
event-based and handles input and output data by
means of special kinds of UI components.

User Interaction and Interface Design with UML

User Interaction and Interface Design with UML

ORGANISATION OF THE CHAPTER

ORGANISATION OF THE CHAPTER
The next section describes our model-driven
The next section
describes
ourinterfaces.
model-driven
development
technique
for user
The
development
technique
for
user
interfaces.
chapter continues with the technique for The
the
chapter
continues
with the
technique models.
for the
code
generation
by using
user-interface
code chapter
generation
by using
models.
This
finishes
with user-interface
some conclusions
and
This
chapter
finishes
with
some
conclusions
and
future work.
future work.

MODEL-DRIVEN DEVELOPMENT
MODEL-DRIVEN
DEVELOPMENT
FOR
USER INTERFACES
FOR USER INTERFACES
Use Case Diagrams
Use Case Diagrams
Use case diagrams are used as starting point for
Use case diagrams
areUse
used
as starting
user-interface
design.
cases
are alsopoint
a wayfor
of
user-interface
design.
Use cases
are alsoand
a way
of
specifying
required
usages
of a system,
they
specifying
required
usages
of
a
system,
and
they
are typically used for capturing the requirements
areatypically
used is,
forwhat
capturing
the requirements
of
system (that
a system
is supposed
of
a
system
(that
is,
what
a
system
is supposed
to do). The key concepts associated with
the useto do).model
The key
associated
with
theusers
usecase
are concepts
actors and
use cases.
The
case
model
are
actors
and
use
cases.
The
users
and systems that may interact with the system
and represented
systems thatbymay
interact
with
the system
are
actors.
Actors
always
model
are
represented
by
actors.
Actors
always
entities that are outside the system. Use model
cases
entities that
thethe
system.
represent
the are
set ofoutside
tasks that
actorsUse
carrycases
out.
represent
the
set
of
tasks
that
the
actors
carry
In addition, the use cases can be decomposedout.
by
In addition,
the use
cases can be
decomposed
by
means
of include
relationships,
and
they can also
means
of
include
relationships,
and
they
can
also
be related by means of generalisation/specialisabe related
by means
of compare
generalisation/specialisation
relationships
that
more general and
tion
relationships
that
compare
more general and
particular tasks.
particular
In ordertasks.
to design a prototype of the user inIn
order
to design
a prototype
of the
user the
interface, the use
case diagram
should
include
terface,actors
the use
case
should
include
the
system
and
thediagram
set of (main)
tasks
for each
system
actors
and
the
set
of
(main)
tasks
for
each
one in which he or she takes part. From a point
oneview
in which
or she takes
part. From
a point
of
of userheinterface
modelling,
the use
case
of
view
of
user
interface
modelling,
the
use
case
diagram can be seen as a high-level description
diagram
canwindows
be seen as
a high-level
of
the main
of the
system. description
of the
main
windows
of
the
system.
To illustrate the functionality
of the MDDTo
illustrate
the
functionality
of theInternet
MDDbased technique we will explain a simple
based
technique
we
will
explain
a
simple
Internet
book shopping (IBS) model.
book
(IBS) model.
In shopping
the IBS example
(Figure 1), three actors
In
the
IBS
example
1),manager,
three actors
appear: the customer, the(Figure
ordering
and
appear:
the
customer,
the
ordering
manager,
and
the administrator. A customer directly makes
the administrator. A customer directly makes

Figure 1. A preliminary use case description of

Figure
A preliminary use case description of
the
IBS 1.
example
the IBS example

the purchases on the Internet, querying certain

issues of the product in a catalogue of books
the purchases on the Internet, querying certain
before making the purchase. The manager deals
issues of the product in a catalogue of books
with customers orders (total or partially). And
before making the purchase. The manager deals
finally, the systems administrator can manage
with customers orders (total or partially). And
the catalogue of books by adding and eliminating
finally, the systems administrator can manage
books in the catalogue or modifying those already
the catalogue of books by adding and eliminating
existing. The administrator can also update or
books in the catalogue or modifying those already
cancel certain component characteristics of an
existing. The administrator can also update or
order or those orders fulfilling certain searching
cancel certain component characteristics of an
criteria.
order or those orders fulfilling certain searching
This information is described with a use case
criteria.
diagram containing the identified actors and main
This information is described with a use case
tasks of the system. In our case study (see Figure
diagram containing the identified actors and main
1), the actors are the customer, the manager,
tasks of the system. In our case study (see Figure
and the administrator, and the main tasks are
1), the actors are the customer, the manager,
purchase, manage orders, manage parand the administrator, and the main tasks are
tial orders, manage catalogue, update
purchase, manage orders, manage parorders, and update partial orders.
tial orders, manage catalogue, update
From this use case diagram, one can identify
orders, and update partial orders.
the future windows of the system that will comply
From this use case diagram, one can identify
with the needs of the presentation logic (graphical
the future windows of the system that will comply
user interfaces).
with the needs of the presentation logic (graphical
user interfaces).

User Interaction Diagrams

User
Interaction Diagrams
The second modelling technique in our framework

is the activity diagram. However, we need to

The second modelling technique in our framework
specialise the activity diagram for user interface
is the activity diagram. However, we need to
design in the following sense.
specialise the activity diagram for user interface
Our activity diagrams include states and transidesign in the following sense.
tions. The states represent data output actions, that
407
331

User Interaction and Interface Design with UML

Our activity diagrams include states and transitions. The states represent data output actions, that
is, how the system responds to user interactions
showing data (or requesting them). Then the user
can introduce data and the corresponding event
is handled and specified by means of transitions. Transitions can be conditioned, that is, the
handled event is controlled by means of condition,
which can be referred to data/business logic or
a previous user interaction. In other words, it is
possible more than one transition from a state,
and to know which of them will run depends on
data/business logic or the previous user choices.
We call user-interaction diagrams to this kind
of activity diagrams used for user interaction
description.
Now, it is supposed that each use case in the
use case diagram is described by means a userinteraction diagram. However, from a practical
point of view, it is convenient to use more than
one user-interaction diagram for describing a
use case. This is so because the logic of a use
case is usually too complex. For this reason, a
user-interaction diagram can be deployed in
several user-interaction diagrams, where a part
of the main logic is separately described. For this
reason, user interaction diagrams can include
states that do not correspond to data output,
rather than representing subdiagrams. Now, it
is desirable in some cases to be able to combine
the logic of the subdiagrams and the main logic.
For this reason, we will use in the main diagram
transition conditions that can be referred to the
logic of the subdiagrams.
Activity diagrams describe input and output
user interactions. Given that we have decided to
implement our user interface by means of Java
swing package, we will consider the JFrame
class as a container class that opens new frame
windows (if needed). In addition, graphical components can be classified as input (a text field, a
button, etc) and output components (a label, list,
etc). Input/output components are associated with
terminal states and transitions by using the ap-

408

propriate stereotype. For instance, the stereotypes

JTextField, JList, JLabel are associated
with states and the stereotype JButton with
transitions. Since the graphical behaviour concerns with states and transitions, next we will
describe them separately.
States can be stereotyped or not. Stereotyped
states represent terminal states, which can be
labelled by <<JTextField>>, <<JList>>,
and <<JLabel>> stereotypes. For instance, let
us focus our attention in the purchase use case.
Figure 2 shows the whole user-interaction diagram
modelled for the purchasing process. Initially, the
behavioural description starts with an original
user-interaction description (Figure 2a). The
behaviour shows how the customer begins the
purchasing process of querying, adding, or removing articles of the shopping cart. After a usual
purchasing process, the shopping system requests
the customers card number and an address to carry
out the shipment whenever the shopping cart is
not empty. This diagram shows the graphical and
behavioural content of the applet window where
the purchases can be carried out.
The main user-interaction diagram (Figure
2a) is composed of three states (i.e., query catalogue, confirm proceed, and shopping cart).
Two of them are terminal states (i.e., confirm
proceed and query catalogue). A terminal
state is described in a new diagram whose states
correspond to graphical elements stereotyped (for
instance <<JTextField>> or <<JList>>)
and labelled by a text related to the graphical
element. The name of a separate user-interaction diagrams should be the same as that of the
state. A non-terminal state is also described in a
separate user-interaction diagram containing one
or more non-terminal states.
Transitions can be labelled by means of
stereotypes, conditions, or both. For instance, a
button is connected with a transition by using a
<<JButton>> stereotype, and the name of the
label is the name of the button. For example, a
show cart transition stereotyped as <<JBut-

User Interaction and Interface Design with UML

User Interaction and Interface Design with UML

Figure 2. The whole user-interaction diagram of the purchase use case

Figure 2. The whole user-interaction diagram of the purchase use case

ton>> will correspond with a button component

called Show cart.
Conditions can represent user choices or
business/data logic. The first one is a condition of
the users interaction with a graphical component
Conditions
can or
represent
userand
choices
or
(related
to button
list states),
the second
business/data
logic.
The
first
one
is
a
condition
of
one is an internal checking condition (not related
thetousers
interaction
with
component
the states,
but to
thea graphical
internal process).
For
(related
to
button
or
list
states),
and
the
second
example, in our case study the selections from a
one
an modelled
internal checking
condition
related
listisare
by conditions.
Note(not
in the
query
to catalogue
the states,
but
to
the
internal
process).
user-interaction diagram how theFor
reexample,
studyby
theaselections
fromstate
a
<<JList>>
sults in
listour
is case
modelled
listand
areamodelled
by
conditions.
Note
in
the
query
[Selected article] condition.
catalogue
user-interaction
diagram how
re- ,
[close]
Figure
2 shows some transitions
(i.e.,the
<<JList>>
state
sults
list
is
modelled
by
a
[exit], or [proceed]) that correspond with
and
a [Selected
conditions
of the article]
user choicecondition.
type. The [Exit]
[close],
Figure
2
shows
some
transitions
(i.e.,
output transition of the state query
catalogue
[exit]
, or
) that
(Figure
2a)[proceed]
means that the
user correspond
has pressed awith
button
[Exit]
conditions
of
the
user
choice
type.
The
called exit, which has been defined in a separate
output
of the
state query catalogue
querytransition
catalogue
user-interaction
diagram. Nev(Figure
2a)
means
that
the
user
has
pressed
empty]
ertheless, conditions like the [cart noa button
exit,are
which
has been defined
a separatein
called
condition
business/data
logic in
conditions,

which the human factor does not take part.

Furthermore, stereotyped transitions (buttons
in our example) and conditions connect (non) terminal states with (non) terminal states. A condition
would be an output of a non-terminal state if the
query
Nevuser catalogue
interacteduser-interaction
with a button or diagram.
a list component
[cart
no
empty]
ertheless,
conditions
like
the
inside the respective non-terminal state. The usual
condition
are business/data transition
logic conditions,
in
way of condition/event
can connect
which
the
human
factor
does
not
take
part.
(no)terminal states with (non) terminal states. A
Furthermore, stereotyped
transitions
(buttons
condition/event
transition between
states
means
inwhich
our example)
and
conditions
connect
(non)
ter-the
condition should be present to trigger
minal
states
withcase
(non)study,
terminal
A condition
event.
In our
an states.
event can
only be a
would
be
an
output
of
a
non-terminal
state
if the
button. For instance, to remove an article
from
user
or a list component
theinteracted
shopping with
cart, ait button
must previously
be selected
inside
non-terminal
fromthe
therespective
cart list (Figure
2c). state. The usual
way of condition/event transition can connect
(no)terminal
states with
(non) terminal states. A
User Interface
Diagrams
condition/event transition between states means
which
condition
should
be present
toobtained
trigger the
Therefore,
in our
technique
we have
a use
event.
In
our
case
study,
an
event
can
only
be
a
case diagram together with a set of user-interacbutton.
For
instance,
to
remove
an
article
from
tion diagrams, from which some correspond to
theuse
shopping
cart,
it must
previously
selected
cases and
others
to states
of usebe
cases.
Howfrom the cart list (Figure 2c).
409


of separate user interaction diagrams. In addition,

the logic of the use cases is not modified in the
Include Relationships
purchase user interaction diagram. It integrates
the logicUser
query
catalogue diagram by checkLet us consider the purchasing process described
Interaction and Interface Design with UML
ing which buttons (i.e., exit) the user pressed
in previous user-interaction diagrams.
when
he or she exits from shopping cart.
purchase use case is a frame that includes
The developer can also identify an inclusion
(uses) three other frames (use cases): query
relationship between manage catalogue and
(use case) is related in the user interface diagram
Figure
3. A user-interface
diagram
forpurthe purFigure
3. A user-interface
diagram
for the
withdraw article, Modify article and
by the inclusion relationship with the main use
chasing
process
chasing
process
add article use cases (Figure 4). In these
case. However, this is achieved whenever the
cases, four windows can be optionally opened
logic of the included use case is not modified by
(depending on a menu) from the manage catathe main use case. The logic of an included use
logue window. In addition, the administrator
case is modified in the main use case whenever
identification window is mandatory opened
the user interaction diagram of the main use case
from the Manage catalogue window in order
includes transitions that are referred to internal
to achieve the systems administrator tasks.
states or conditions of the included use case.
This typically happens when the main use case
Generalisation Relationships
defines its own logic using the included use case
as a piece of behaviour, but access to the included
In order to illustrate the generalisation/specialisause case. Otherwise, that is, whenever the logic
tion relationship, we will pay attention to three
of the included use case is modified, we relate the
use cases: purchase, query catalogue, and
main use case and the state (use case) by means
query catalogue by administrator. In
of the generalisation relationship.
previous sections, we have identified two cases
In addition, we have to identify a special case
of generalisation/specialisation.
of generalisation relationship, in which we have
a main use case including a state and the state
ever, it could be useful to have a new version of
itself is a specialisation version of another use
the
use
case
diagram,
to
know
what
are
the
main
case. Thisside
happens when the user interaction diaFigure 4. A piece of the user interface diagram of the administrator
user-interaction diagrams, that is, which are the
gram of the state specialises the user interaction
user-interaction diagrams corresponding to use
diagram of another use case. The specialisation
cases, and which are the secondary user-interconsists of the replacement of states and transiaction diagrams, that is, which are states of use
tions of the user interaction diagram by means
cases. For this reason, we will build a new verof more particular states and transitions with
sion of the use case diagram called user-interface
similar semantics. Here we are interested in the
diagram as follows. The user-interface diagram
identification of similarity between interactions
contains the same actors and use cases of the use
following the same pattern; however, the differcase diagram. In addition, we will add states (as
ence in our case consists of several kinds of GUI
use cases) of each use case corresponding to user
components and input/output data.
interaction diagrams. In order to connect use cases,
In other words, the specialisation of use cases
we will use the include or generalisation use
allows us to build new use cases with a more
case relations.
complex logic containing the specialized use case,
and adding transitions and states or modifying the
existent ones. On the contrary, the inclusion allows
User Interface Relationships
us to build new use cases with a more complex

logic without adding or modifying the states
and
Include and generalisation relationships have the
transitions
of
the
included
use
case.
following meaning in this user-interface diaGiven that the user interface diagram is a kind
gram, similarly to the same relations in the use
of
high-level
description of the user interface,
case diagram. If a use case contains in its user
the developer still could decide not to include
interaction diagram a state described by means of
in it all the states representing user interaction
another user interaction diagram, the state is also
diagrams. In other words, the user interface diaconsidered as a use case, and the included state

410

User Interaction and Interface Design with UML

User Interaction and Interface Design with UML

gram represents the set of windows of the system

by means of use cases. However, there could be
more
user interaction
diagrams than
windows.
has
to identify
use case relationships
in the
new
Usually,
some
user interaction
user
interface
diagram
as follows.diagrams can be
built for deploying some states but they will not
correspond
to system windows.
Include
Relationships
Once user-interaction diagrams have been
described,
thethe
designer
proceeds
to build
the user
Let
us consider
purchasing
process
described
interface
diagram.
It
contains
new
use
cases
that
in previous user-interaction diagrams.
are
some of the
states
the user
purchase
use non-terminal
case is a frame
thatof
includes
interaction
diagrams.
In
addition,
the
developer
(uses) three other frames (use cases): query
has to identify use case relationships in the new
user interface
diagram asdiagram
follows. for the purFigure
3. A user-interface
chasing process

Include Relationships

Let us consider the purchasing process described

in previous user-interaction diagrams.
purchase use case is a frame that includes
(uses) three other frames (use cases): query
catalogue, shopping cart, and confirm
proceed. These use cases are described by means
of separate user interaction diagrams. In addition,
the logic of the use cases is not modified in the
purchase user interaction diagram. It integrates
the logic query catalogue diagram by checking which buttons (i.e., exit) the user pressed
when he or she exits from shopping cart.

The developer can also identify an inclusion

relationship between manage catalogue and
withdraw article,
article
and
catalogue,
shopping Modify
cart, and
confirm
cases
(Figure 4).
these
add article
proceed
. These useuse
cases
are described
by In
means
cases,
four
windows
can
be
optionally
opened
of separate user interaction diagrams. In addition,
manage incata(depending
onuse
a menu)
themodified
the
logic of the
casesfrom
is not
the
logue
window.
In
addition,
the
administrator
purchase user interaction diagram. It integrates
window isdiagram
mandatory
opened
theidentification
logic query catalogue
by checkManage
catalogue
window
in
order
from
the
ing which buttons (i.e., exit) the user pressed
to achieve
theexits
systems
when
he or she
fromadministrator
shopping cart.tasks.
The developer can also identify an inclusion
Generalisation
relationship
betweenRelationships
manage catalogue and
withdraw article, Modify article and
In order
to illustrate
generalisation/specialisaarticle
use the
cases
(Figure 4). In these
add
tion relationship,
wecan
willbepay
attentionopened
to three
cases,
four windows
optionally
purchase,
query
catalogue,
and
use cases: on
cata(depending
a menu) from
the manage
query
catalogue
by the
administrator
. In
logue
window.
In addition,
administrator
previous sections,
we have
identified two
cases
identification
window
is mandatory
opened
of generalisation/specialisation.
from
the Manage catalogue window in order
query catalogue
The first
is theadministrator
to achieve
the case
systems
tasks. and
purchase case. In this case, the purchase
user-interaction Relationships
diagram contains a state (use
Generalisation
case) that specialises query catalogue in the
query catalogue user
sense. The
Infollowing
order to illustrate
the generalisation/specialisainteraction
diagram
describes
how to to
query
tion
relationship,
we will
pay attention
threethe
catalogue
of
the
IBS
by
introducing
the
searching
use cases: purchase, query catalogue, and
criteriacatalogue
and showingbytheadministrator
results. However,
the
query
. In
purchase
user
interaction
diagram
can
interrupt
previous sections, we have identified two cases
of generalisation/specialisation.

Figure
piece
user
interface
diagram
administrator
side
Figure
4.4.
AA
piece
ofof
thethe
user
interface
diagram
ofof
thethe
administrator
side

411

User Interaction
Interaction and Interface
Interface Design with
with UML
User
User Interaction and
and Interface Design
Design with UML
UML

Figure 5.
5. A
A piece
piece of
of the
the user-interface
user-interface diagram
diagram
Figure

first case
is the
catalogue
and
the The
querying
process
by query
adding the
searched items
purchase
case
. InIt this
case, the
to the shopping
cart.
is specified
bypurchase
adding the
user-interaction
diagram
contains
a state
Add to cart button as transition from
(and(use
to)
query
catalogue
in
the
case)
that
specialises
query catalogue. Therefore we can identify
catalogue
user
following
sense.relationship
The querybetween
purchase
a specialisation
interaction
describes
howsupposed
to query that
the
catalogue
. It is also
and query diagram
catalogue
of
the
IBS
by
introducing
the
searching
there will be a window for query catalogue
criteria
and purchase
showing the
results. However, the
inherits.
from which
purchase
user
interaction
can interrupt
The second case is the diagram
relationship
between
the
querying
process
by
adding
the
searched
items
query catalogue and query catalogue
by
to
the
shopping
cart.
It
is
specified
by
adding
the
administrator. In this case, the administrator
Add
to cart
button
as transition
(and to)
is supposed
to have
higher
privilegesfrom
for querying
query
catalogue
.
Therefore
we
can
identify
the catalogue and therefore the user interaction
purchase
adiagram
specialisation
relationship
betweenby
catalogue
adminof the query
and
query
catalogue
.
It
is
also
supposed
istrator (see Figures 5 and 6) specialises inthat
the
query
catalogue
there
will
be
a
window
for
query catalogue user interaction diagram
from which purchase inherits.
Figure 6. The user-interaction diagram for the
Figure
6. The user-interaction
diagram for the
query catalogue
by administrator
query catalogue by administrator

The second case is the relationship between

Thefollowing
second case
is the
in the
sense.
Therelationship
states of thebetween
query
query catalogue and query catalogue by
catalogue by administrator corresponding
administrator. In this case, the administrator

with the searching criteria and results are modiis

supposed
to have
privileges
for querying
catalogue. It is
fied
with respect
tohigher
the query
the
catalogue
therefore the
supposed
that and
the searching
anduser
resultinteraction
fields are
catalogue by admindiagram
of
the
query
different but the logic is similar. In other words, the
istrator (see Figures 5 and 6) specialises in the
query catalogue can be replaced by query
query catalogue user interaction diagram
catalogue by administrator given that the
query
in
the can
following
sense. In
The
states
be replaced.
thisstates
case,ofwethe
can
idencatalogue by administrator corresponding
tify a generalisation relationship between them.
with
the searching
criteria and results are modiWithdraw article and Modify
Analogously,
fied
with
respect
to
the query catalogue. It is
article combine both kinds of specialisation,
supposed
thatspecialized
the searching
and result
fields are
catalogue
once theyve
the query
different
but
the
logic
is
similar.
In
other
words,
the
by administrator in the same sense that
query catalogue can be replaced by query
purchase specializes query catalogue, and
catalogue by administrator given that the
they specialize indirectly query catalogue
states
can be6).replaced. In this case, we can iden(see Figure
tifyThe
a generalisation
relationship
between of
them.
complete user
interface diagram
our
Analogously,
case study canWithdraw
be seen in article
Figure 7. and Modify
article combine both kinds of specialisation,
once theyve specialized the query catalogue
GUI Class Diagram
by administrator in the same sense that
purchase specializes query catalogue, and
The next step of our model-driven technique
they specialize indirectly query catalogue
consists of the building of a class diagram for
(see Figure 6).
GUI components. The user-interface diagrams
The complete user interface diagram of our
obtained in the previous state give us the main
case study can be seen in Figure 7.
windows. Each use case connected to an actor
can be converted into a window, and if an actor

412



User Interaction and Interface Design with UML

User Interaction and Interface Design with UML

Figure 7. The Internet shopping user-interface diagram

Figure 7. The Internet shopping user-interface diagram

classtodiagram
isGuI
connected
more than one use case, it can be
considered a window by the actor that invokes (or
The nexteach
stepwindow
of ourofmodel-driven
embeds)
each use case.technique
consists
of
the
building
of a class
for
Therefore, the actor window
candiagram
be a menu
GUI components.
The in
user-interface
diagrams
window.
In addition,
the user-interaction
obtained
in
the
previous
state
give
us
the have
main
diagrams obtained from use cases, we
windows.
Eachinput
use and
caseoutput
connected
to an actor
also
described
components
for
can
be
converted
into
a
window,
and
if
an
actor
data output and request and user events. It gives
is connected
to more thanfor
oneeach
use case,
it can
us
the GUI components
window.
If be
a
considered
a
window
by
the
actor
that
invokes
(or
user-interaction diagram has a state described
embeds)
windowuser-interaction
of each use case.
by
meanseach
of another
diagram,

Therefore,
canthebeuse
a menu
we can
supposethe
thatactor
the window of
case
window.
In
addition,
in
the
user-interaction
could also contain a separate window for this
diagramstask.
obtained
fromnow,
use we
cases,
have
separate
However,
havewe
to take
also
described
input
and
output
components
for
into account the user-interface diagram and the
datacase
output
and requestInand
It gives
use
relationships.
theuser
caseevents.
of inclusion,
us
the
GUI
components
for
each
window.
the logic is also separate, and it is possibleIftoa
user-interaction
diagram
has a state
described
consider
a new window.
However,
in the
case of
by
means
of
another
user-interaction
diagram,
generalisation/specialisation, the window corwe can suppose
the windowand
of the
use case
responds
with a that
specialisation,
therefore
it
could
also
contain
a
separate
window
for
this
is better to consider new window by using the
separate task.
However, now, we have to take
inheritance
relation.

413


built and a set of user-interaction diagrams has

been obtained, now we can generate a GUI-class
diagram.
The GUI-class diagram is built from Java
swing classes. In the method, each use case
corresponds with a frame class. Use cases are

GUI-class diagram. The stereotype name of a

transition or state is translated into the appropriate
Java swing class. For example, the <<JButton>>
stereotype
of the Proceed
transition
that appears
User Interaction
and Interface
Design
with UML
in the purchase user-interaction diagram (see
Figure 2a) is translated into a JButton class.

Figure 8. Purchase class obtained from user-interface diagrams and user-interaction diagrams
Figure 8. Purchase class obtained from user-interface diagrams and user-interaction diagrams

User Interaction and Interface Design with UML

Figure 9. Query catalogue class obtained from user-interface diagrams and user-interaction diagrams

Figure 9. Query catalogue class obtained from user-interface diagrams and user-interaction diagrams

Figure 10. Shopping cart class obtained from user-interface diagrams and user-interaction diagrams

414

User Interaction and Interface Design with UML

Figure 10. Shopping cart class obtained from user-interface diagrams and user-interaction diagrams
Figure 10. Shopping cart class obtained from user-interface diagrams and user-interaction diagrams
Figure 10. Shopping cart class obtained from user-interface diagrams and user-interaction diagrams

Figure11.
11.Confirm
Confirmproceed
proceedclass
classobtained
obtainedfrom
fromuser-interface
user-interfacediagrams
diagramsand
anduser-interaction
user-interactiondiadiaFigure
Figure 11. Confirm proceed class obtained from user-interface diagrams and user-interaction diagrams
grams
grams



The windows obtained from use cases and the

GUI components obtained from each user-interaction diagrams allow us to build a class diagram,
which we call GUI-class diagram, where there
are windows classes connected by associations
or by inheritance. In addition, there are GUI
component classes connected by associations to
window classes.
Once the user-interface diagram has been
built and a set of user-interaction diagrams has
been obtained, now we can generate a GUI-class
diagram.

The GUI-class diagram is built from Java

swing classes. In the method, each use case
corresponds with a frame class. Use cases are
translated into classes with the same name as these
use cases. The translated classes specialise a Java
Frame class. The components of the frame (use
case) are described in user-interaction diagrams.
A terminal state is translated into that Java swing
class represented by the stereotype of the state.
The Java swing class is connected from the container class. For example, those terminal states
stereotyped as <<JTextField>> are translated

415



be obtained
e 13 shows
e window.
ery similar
except that
more than
en windows
agram as a
use cases:

purchase

mer will alpened from

n a query
mer inherits
relation of

purchase

nd that the
case of the

re 13c) apis pressed

b). Note in
n Figure 7,
he window
etween use
nformation

User Interaction and Interface Design with UML

User Interaction and Interface Design with UML

Figure 12. Confirm remove article class obtained from user-interface diagrams and user-interaction
Figure 12. Confirm remove article class obtained from user-interface diagrams and user-interaction
diagrams
diagrams
windows (Figure 13d) are also associated with
two buttons: the remove article button in the
shopping cart window and the proceed button
in the purchase window. Note again how these
windows are also described as inclusion relations
between use cases.
Moreover, observe the user-interaction diagrams shown in Figure 2 to better track the behaviour of the example. To develop the example,
we have used the Rational Rose for Java tool.
For space reasons, we have only included a part
of the GUI project developed for the case study.
Figure
The applet windows of the customers
GuI13.
Prototypes
Figure
13.
The applet windows of the customers
side
side
Finally, rapid GUI prototypes could be obtained
from the GUI-class diagram. Figure 13 shows
a first visual result of the purchase window.
Note how the purchase window is very similar
to the query catalogue window, except that
the second one includes three buttons more than
the first window. This similarity between windows
was revealed in the user interface diagram as a
generalisation relationship between use cases:
between the query catalogue and purchase
use cases. In the IBS design, the customer will always work on a purchase window opened from
the Customer window, and never on a query
catalogue window, though the former inherits
the behaviour of the latter (i.e., by the relation of
generalisation). Let us remark that purchase
intoinherits
a JTextField
class in the
GUI-class diagram.
from query
catalogue, and that the
Something
similar
happens
with
theuse
restcase
of stefive windows
(GUI)
are the
five
of the
reotyped
states
and
transitions.
Figures
8
to
12
client side (see Figure 3).
show the
classes
of window
GUI-class
diagram
Themain
shopping
cart
(Figure
13c)ofapthepears
customers
side.
when the show cart button is pressed
As the
it can
be seenwindow
in these (Figure
figures, the
on
purchase
13b).stereoNote in
typed
states
and
transitions
in
the
user-interaction
the user interface diagram, shown in Figure 7,
diagrams
translated
into Javawith
classes
the
how theare
button
is associated
the in
window
GUI-class
diagram.
The
stereotype
name
of
by means of an inclusion relation between ause
transition
or state
translated
appropriate
cases. On
the isother
hand,into
thethe
two
information

416
0

Java
swing class.
For example,
windows
(Figure
13d) are the
also<<JButton>>
associated with
stereotype
of thethe
Proceed
that
appears
removetransition
article
button
in the
two buttons:
in shopping
the purchase
user-interaction
diagram
(see
cart window and the proceed button
JButton
class.these
Figure
2a)purchase
is translated
into aNote
in the
window.
again how
windows are also described as inclusion relations
GUI
Prototypes
between
use cases.
Moreover, observe the user-interaction diaFinally,
GUIinprototypes
obtained
gramsrapid
shown
Figure 2 tocould
betterbetrack
the befrom
the
GUI-class
diagram.
Figure
13
shows
haviour of the example. To develop the example,
a first
visualused
result
the purchase
we have
theofRational
Rose forwindow.
Java tool.
purchase
window
is
very
similar
Note
how
the
For space reasons, we have only included a part
query
catalogue
window,
except
to the
of the
GUI project
developed
for the
casethat
study.
the second one includes three buttons more than
13. The
applet
windows
of thewindows
customers
theFigure
first window.
This
similarity
between
wasside
revealed in the user interface diagram as a
generalisation relationship between use cases:
between the query catalogue and purchase
use cases. In the IBS design, the customer will always work on a purchase window opened from
the Customer window, and never on a query
catalogue window, though the former inherits
the behaviour of the latter (i.e., by the relation of
generalisation). Let us remark that purchase
inherits from query catalogue, and that the
five windows (GUI) are the five use case of the
client side (see Figure 3).

User Interaction and Interface Design with UML

The shopping cart window (Figure 13c) appears when the show cart button is pressed
on the purchase window (Figure 13b). Note in
the user interface diagram, shown in Figure 7,
how the button is associated with the window
by means of an inclusion relation between use
cases. On the other hand, the two information
windows (Figure 13d) are also associated with
two buttons: the remove article button in the
shopping cart window and the proceed button
in the purchase window. Note again how these
windows are also described as inclusion relations
between use cases.
Moreover, observe the user-interaction diagrams shown in Figure 2 to better track the behaviour of the example. To develop the example,
we have used the Rational Rose for Java tool.
For space reasons, we have only included a part
of the GUI project developed for the case study.
A complete version of the project is available at
http://indalog.ual.es/mdd/purchase.

CODE GENERATION FOR USER

INTERFACES
Finally, our model driven technique allows us
to obtain code generation from the developed
models. The user-interaction diagrams describe
each user interaction of each use case by means
of states representing data output and transitions
representing data input and events.
Code generation consists of coding an eventbased user interface. We have decided to implement our user interface by means of the Java
swing classes, and therefore we will use Applet
and Frame classes for window components, and
button, label, list, etc., for other UI components.
According to the Java swing package, events are
handled by means of listeners and event handlers
in the corresponding window classes.
In order to explain the transition of the user
interaction diagrams to code, we should first
track a real interaction of the users with windows

from the example: the purchase behaviour from

the client side.
Here we can see the complete sequence of
user/windows interaction. The first window (purchase, slide 1) shows ten graphical components:
two labels (searching criteria and results), one
text field (textfield), one field with a list of results
(results), and six buttons. As we can see, the
search, close and exit buttons are enabled
whereas buttons add to cart, show cart,
and proceed are not active. The user can only
choose one searching criteria in the corresponding
field and then click search. This interaction with
the window after clicking the button makes the
user access the logic data to obtain the results and
then he or she gets to slide 2 window.
In that window (slide 2), we can observe that
the searching button has been deactivated and the
user can choose either one element from the list or
exit the application or clear the searching criteria,
in which case the search button is again activated.
Following the usual interaction sequence, when
the user is in the slide 2 window, he or she can
choose one item from the list and then he or she
will go to slide 3.
Moreover in slide 3, one button to add the
cart is activated. The user adds to the cart the
selected element by pressing the button and then
he or she gets to the slide 4 window. In this state,
there are some buttons that can show, on the one
hand, the content of the cart (show cart) and
on the other hand, they can make the purchase
(proceed). The user can choose different elements from the list and add them to the cart. He
or she can also enter new elements to the cart
from a new searching, clearing the list (clear),
and entering a new searching criteria (slide 5).
In this state, we can observe that the searching
button is activated and the adding to cart button is
deactivated. However, show cart and proceed
buttons are not activated because the cart is not
empty and the user at any time can see the cart
or make the purchase. If the user wishes to see
the content of the cart, he or she would get to the
slide 6 window (shopping cart).
417

code GenerAtIon for uSer

InterfAceS

representing data input and events.

Code generation consists of coding an eventbased user interface. We have decided to implement our user interface by means of the Java
swing
andand
therefore
weDesign
will use
User classes,
Interaction
Interface
withApplet
UML
and Frame classes for window components, and
button, label, list, etc., for other UI components.
According to the Java swing package, events are

Finally, our model driven technique allows us

to obtain code generation from the developed
models. The user-interaction diagrams describe
Figure 14. All complete sequence of the purchase interaction
Figure 14. All complete sequence of the purchase interaction

This new window (slide 6) has two labels, one

list and two buttons; the first one is deactivated to
avoid the removal of an element from a list until
the user hasnt selected one of them and the second
one allows us to close the window.
If the user selects one element from the list
(slide 7), the button to eliminate it from the cart
is activated. If so, he or she would get to the third
level of the window for requesting the users
confirmation to eliminate the element from the
cart (slide 8).

418

If the user confirms the elimination (in window slide 8) he or she would get to the last level
of the window shown in slide 9; otherwise, he or
she would get to the slide 7 again.


General Structure of Frame Classes

The interaction behaviour previously described is
defined by the designer through user-interaction
and user-interfaces diagrams. The relationship
of dependences between windows is described

2.

The user-interaction perspective, where

the designer models the behaviour of the
graphic components of the windows and
their interaction with the user; and
User Interaction and Interface Design with UML
3. The GUI perspective, where the user works
with a rapid prototype of the windows through

tive, including the type of relationship between

windows (in order to clarify the dependences
between them).
In this example, these five use cases lead to
five Java files of frame type. Their names are obtained from the use case name (which will be the

Figure 15. A summary for the user-interface and user-interaction diagram of the purchase function
Figure 15. A summary for the user-interface and user-interaction diagram of the purchase function

by means of user-interfaces diagrams (see again

Figure 3).
The behaviour of the window when the user
activates and deactivates buttons and changes the
state of the graphical components is described by
means of user-interaction diagrams (see again
Figure 2). In order to obtain the behaviour shown
in Figure 14, modelling it with user-interface
and user-interaction diagrams (Figures 2 and
3), we create code prototypes in Java from these
diagrams, making them correspond with specific

of
lines of Java code. The most general structure
our method is the Frame classes for the windows
of the system.
To explain how to make this relationship between diagrams and code, we are focussing on
the purchase user-interaction diagram. Figure 15
shows three perspectives of the purchase:

1.

The user-interface perspective, where the

designer studies the dependences between
windows through use cases;

419

User Interaction and Interface Design with UML

Table 1. Correspondence between diagram names

and file names
User-interface & userinteraction name

File name

purchase

purchase.java

query catalogue

querycatalogue.java

shopping cart

shoppingcart.java

confirm proceed

confirmproceed

confirm remove article

confirmremovearticle.java

Table 2. General structure of querycatalogue

(frame)
User-Interface

Code generation

Diagram

import java.awt.*;
import javax.swing.*;

query
catalogue

2.

3.

public class querycatalogue

extends JFrame {
/** @GUIcomponents */
...
public querycatalogue() {
try {
jbInit();
}
catch(Exception e) {
e.printStackTrace();
}
}
private void jbInit() throws
Exception {
...
}
// Methods implementation
...
}

The user-interaction perspective, where

the designer models the behaviour of the
graphic components of the windows and
their interaction with the user; and
The GUI perspective, where the user works
with a rapid prototype of the windows through
a rapid generation of the code from both
aforesaid perspectives (as previously seen).

In a symbolic way, we have represented in

the picture the GUI perspective (of the windows)

420

with its equivalent in the user-interface perspective, including the type of relationship between
windows (in order to clarify the dependences
between them).
In this example, these five use cases lead to
five Java files of frame type. Their names are obtained from the use case name (which will be the
same as the one for the user-interaction diagram,
one per use case). If compound nouns exist then
a simple name is created with the initials of each
word in capital letters (Table 1).
Classes extend JFrame class and present a
general structure of implementation based on
frames (see Table 2). The basic structure of frame
prototype created in the diagrams-to-code transition, is composed of four sections: one heading,
one class constructor, one initiator of the frame
( jbInit) and one basic prototype of the implementation of the detected methods (related with the
logic of presentation). In the heading (labelled
with /** @GUIcomponents */) the attributes
are considered to be graphical components that
have been detected in the corresponding userinteraction diagram. The class constructor is
the same for all code files created (except for the
name of the constructor). In the initiator of the
frame ( jbInit) it is included the code lines that
model the graphical content of the window. The
designer should establish the position and size of
the graphical components of the frame later on,
as the code generated is just a prototype of the
window. In the fourth section of the Frame, it is
created the basic methods related to the interaction of the user with the graphic elements that
he or she can interact with: buttons, lists, and
text fields.

GUI Components
In order to analyze how translating the diagrams
components to code inside the frame, well again
use the purchase example in Figure 15.
The behaviour of the purchase window begins
with the initial state query catalogue (represented

User Interaction and Interface Design with UML

by the designer as an inheritance of behaviour in

the user-interface diagram). As it can be observed,
the purchase window inherits all the graphical
components (and behaviour) of the query catalogue window and it also adds the add to cart,
show cart and proceed buttons, which are not
activated. From the initial state query catalogue,
the four out coming transitions correspond to
interactions with the buttons of the window and
deal to new states. What the diagram shows from
its initial state is the following: If there is an element selected from the list, then it is activated
the button to add one element to cart.

<<JButton>>
[ selected article ] / Add to cart

If the cart is not empty ([cart not empty]) it

is activated the button to show the content of the
cart and make the purchase of the products in
the cart.

<<JButton>>
[ cart not empty ] / Shopping cart

<<JButton>>
[ cart not empty ] / Proceed

The transition [exit] in the initial query catalogue state, corresponds with the operation of
clicking on the exit button inside the behaviour
diagram of the window (frame): query catalogue
(window inherited by purchase). The same happens with the other two transitions that reach
this initial state from the state Confirm proceed
(window to confirm the purchase).
For this example, there are four modelled
graphical components: labels, text fields, lists,
and buttons. In Table 3, it is shown the correspondence between graphic components of the
user-interaction diagram (stereotypes) and the
code generation.
In Table 3, column stererotype represents
the stereotypes used in the states/transitions of
the user-interactions diagram.
The columns attribute and class represent the name of the attribute and the base class
instantiated in the code generated. The criterion
followed to establish the name generated in
the code is: name_Type. That is, as a name it is
used the same name indicated in the interaction
diagram in capital letters, followed by a hyphen
(i.e., _) and then finished by base type (label,
button, list, etc.). If the original name has blanks

Table 3. Correspondence between user-interaction stereotypes and code generation

UI diagrams

Code generation

Stereotype

Atribute

Class

In/Out

Listener

S/T

Out

No

State

/** @Label */

JTextField

In/Out

No

State

/** @TextField */

JList

In/Out

Maybe

State

/** @List */

In

Yes

Trans.

/** @Button */

In/Out

No

Trans.

None

<<JLabel>>
name _ Label

JLabel

name
<<JTextField>>
name

name _ Label
name _ TextField

Markup

name _ Label
<<JList>>
name _ List
name
name _ ScrollPane
<<JButton>

name _ Label

name

name _ Button

JButton

[condition]

None

None

421

User Interaction and Interface Design with UML

User Interaction and Interface Design with UML

Figure
Figure16.
16.The
Theconfirm
confirmremove
removearticle-label
article-labelstate
stateininquery
querycatalogue
catalogue

in between, they are substituted by hyphens. For

example, the label:

The listener column indicates whether a

graphical component is moved by means of a
listener class. This type of class allows the class
<<JLabel>>
it for listen the behaviour of the graphic
Figure 17. The searching criteria text field state in querycreating
catalogue
Shopping cart
component with the user. It is usual to use listener
in buttons and lists (well study these later).
It is translated to the name of the attribute
The S/T column is referred to whether the
shopping cart button and then it is generated the
graphic component corresponds with a state or a
following Java code:
transition in the user-interaction diagram.
Finally, markup column is referred to the
public JLabel shopping _ cart _
mark prototype created before the translated code.
Button = new JLabel();
This mark helps to identify the content of the text
when the programmer has to rewrite or insert the
the in/out column represents an exit from/to
new code to add and complete the functionality
the user (or window). So, for example, one steof the windows.
reotype label will represent a text shown from
the system towards the user (window) whereas
Mapping States into Code
the button (according to its graphic behaviour)
represents an income action (interaction) from
As previously said, the states of a user-interaction
the user towards the system. The textfields
diagram can be referred to terminal states (label,
and lists stereotypes represent both options: the
textfield, list) and non-terminal states
system shows and modifies the content of these
graphic components and at the same time, the
Non-Terminal States
user can either write (textfield) or choose (list) in
these graphic components. The out components
Let us begin with the user-interaction diagram
have no correspondence with the code, whereas
purchase (see Figure 15). The diagram is comthe in components have correspondence with
posed of three non-terminal states. We should
one listener class.
remember that a non-terminal state is a state

422



User Interaction and Interface Design with UML

Figure17.
17.The
Thesearching
searchingcriteria
criteriatext
textfield
fieldstate
stateininquery
querycatalogue
catalogue
Figure

User Interaction and Interface Design with UML

Figure
The
results
state
query
catalogue
Figure
18.18.
The
results
listlist
state
in in
query
catalogue



423

User Interaction and Interface Design with UML

whose behaviour is defined separately from other

diagram of interaction. For example, the query
catalogue, confirm proceed, and shopping cart
states will be defined in independent user-interaction diagrams having the same name as the corresponding state. In the user-interface diagram,
the query catalogue is established by the designer
as an inheritance for purchase, whereas the other
two states (which are use cases and therefore
windows) are established as conclusions in the
user-interface diagram (i.e., <<include>>). The
translation is as follows:
public class purchase extends querycatalogue {
/** @GUI */
ShoppingCart shoppingcart _ GUI =
new ShoppingCart();
/** @GUI */
ConfirmProceed confirmproceed _ GUI
= new ConfirmProceed();
...

Note that purchase class inherits querycatalogue in the code. The two include relationships
are included in the code in the heading section (one
of the four that frame has). On the code line a @
GUI mark is inserted to trace an inclusion of the
window in the process of translation. Moreover,
the criterion for the name of the GUI variable is
considering the original name of the included
window followed by _GUI.

Terminal States
Terminal states (label, list, textfield) are translated
into code lines in the heading, initiation of Frame
( jbInit) and implementation of methods sections
(only in list). Appendix A contains the Java code
patterns translated from the graphical components
of the user-interaction diagram. Lets see each
of them separately.
For a label case, three basic code lines are
generated: one in the heading to define the type
(JLabel) and two in the initiation ( jbInit) to establish the component. In Figure 16, it is shown

User Interaction and Interface Design with UML

Figure
Figure 19.
19. The
The purchase
purchase window
window

a conditioned transition has been included for

reaching these two windows from purchase: that
424
the cart is no empty. In other words, a conditioned
button-transition corresponds with an operation

of the transition in the user-interaction diagram.

For instance, a button-transition Add to cart
containing a condition [selected articles]
means that the class (frame) purchase enables the

User Interaction and Interface Design with UML

a bit of code generated to the label state confirm

remove article. The code line 3 creates one
variable of JLabel type. Note how the name of
the variable is obtained from the name given to
the <<JLabel>> state in the user-interaction
diagram. The line 7 establishes the text that will
appear in the window. Line 9 adds the label to the
container. Lines 2, 6, and 9 are marks established
previously that are inserted in the code to facilitate
the identification of each part of the code.
Table A.2 of Appendix A shows the Java
code pattern that is translated from the <<JLabel>> graphical components of user-interaction
diagram.
For the text fields, five basic code lines
are generated. Two of them are generated in the
heading to define two graphical components in
the heading: one JTextField and one Jlabel
to name it in the window. The other three code
lines are generated in the initiation section of
frame ( jbInit). For example, Figure 17 shows how
translating the terminal state searching criteria
(in the user-interaction diagram of frame query
catalogue) into a code in two components (lines
2 and 3). It is created a text field area (JtextField) and a text (label) which accompanies
the text field. For the naming of the variables we
use the same criterion established before: the same
text that appears in the state followed by label
or textfield. In line 9, it is established the text in
the window. Lines 11 and 12 add both graphical
components to the container.
Table A.3 of Appendix A showns the Java code
pattern that is translated from the <<JtextField>> graphical components of the user-interaction diagram.
Finally, there are the terminal states of list
type. Graphically these states are somehow special
sincelike a textfieldthey can be translated
into different ways: a label component, other list
or even a scrollpane. The latter can be used as a
container to show the elements from the corresponding list. Several lines are generated in the
translation, in three sections (heading, initiation,

and methods). For example, Figure 18 shows a

translation of the results state.
Lines 5, 6, and 7 are generated in the heading (after /** @GUIcomponents */) defining
the name of the variables of the three graphical
components. Again the criterion used for the
naming of the variables is the same as that one
used in previous examples. In the initiation section ( jbInit) the following code is created. Firstly,
the label text is established in the panel (line 11).
Then, an event of listener type is used to catch
the interactions of the user with the list when he
or she presses the elements of the list. Therefore,
in line 14, there is the container Scrollpane type
where the list is and from lines 15 to 18 there is
the listener event on the list. To do so, the library
javax.swing.event.* is imported (line 1).
Lines 21 and 22 add graphical elements to the
panel. Finally, the method implementation section
(line 26) includes the prototype of the method
used each time the user presses on an element
from the list. The implementation of the method
is a later task of the programmer.
In table A.3 of Appendix A it is shown the
Java code pattern translated from JList graphical
components of user-interaction diagram.

Mapping Transitions into Code

Transitions correspond with buttons (<<JButton>>) or changes in the internal state of the data
logic. They are represented in the diagram as a
condition markup (p.e., [cart not empty]
o [Close]). Table A.1 (Apendix A) shows the
Java code pattern for a graphical component
<<JButton>> translated from the user-interaction diagram.
Transitions represent graphical components
that generate code at several sections of the
program. The translation into code is the more
important issue in the code generation process,
since it reveals the usefully of the user-interaction
diagrams for the traceability of the interaction
of the user with the graphical components of the

425

User Interaction and Interface Design with UML

window and also how that (window) responds to

the user. To explain the translation of states to
code, we will use again the purchase windows
sample. In Figure 19, we show (at the right side)
the two related user-interaction diagrams of the
purchase window (showed at left side).
Remember that the systems designer has
described the purchase window from the query
catalogue window, which represents a state in
the user-interaction diagram of the purchase
window. This state inherits the inner behaviour
of the query catalogue (whose behavioural interaction is shown at the right-lower side in Figure
19). It arrives at two possible states through four
transitions from the state query catalogue: (a) see
the state of the cart (shopping cart state); or (b)
carry out the purchase of the products stored in
the cart (confirm proceed state). Both states are
reached from the purchase window through the
buttons show cart and proceed. In the state,
a conditioned transition has been included for
reaching these two windows from purchase: that
the cart is no empty. In other words, a conditioned
button-transition corresponds with an operation
of activation/deactivation of buttons.
Since these three buttons are conditioned by
a selection action in a list, a listener operation
is generated on the list in the purchase class.
Remember that both the list and the most of the
graphical components of the purchase window
are inherited from query catalogue window
(i.e., public class purchase extends
querycatalogue), and therefore the operation
listener is done on the superclass, which contains
the list.
/** @List */
super.results _ List.addListSelectionListener(
new javax.swing.event.ListSelectionListener() {
public void valueChanged(ListSelectionEvent e) {
selected _ articles _ List _ querycatalogue _ GUI(); } } );

In this way, whenever the user presses on the

list, an event is captured, which calls to the function selected _ articles _ List _ query-

426

catalogue _ GUI(). The name of themethod is

established by the text included in the condition

of the transition in the user-interaction diagram.
For instance, a button-transition Add to cart
containing a condition [selected articles]
means that the class (frame) purchase enables the
button Add to cart in the window if and only
if an element of the list has been selected; in other
case, the button will remain disabled (see Figure
19). In case of exist a selected element in the list,
the button is activated and the transition leads
again to the same state, to include the selected
element (and others) to the list. This generates
the following lines of code for the method on the
listened element.
/** @List */
void selected _ articles _ List _ querycatalogue _
GUI() {
if (results _ List.getSelectedIndex()==-1)
add _ to _ cart _ Button.setEnabled(false);
else add _ to _ cart _ Button.setEnabled(true);
}

The activation of the button (add _ to _

cart.Button.setEnabled) implies the button
has been before declared in the panel, created
and added.
public class purchase extends querycatalogue {
/** @GUIcomponents */
public JButton add _ to _ cart _ Button = new JButton();
private void jbInit() throws Exception {
/** @Button */
name _ Button.setText(Add to cart);
name _ Button.addActionListener(
new java.awt.event.ActionListener() {
public void actionPerformed(ActionEvent
e) {
add _ to _ cart _ Button _ Purchase _ GUI();
} });
/** @Panel */
this.getContentPane().add(add _ to _ cart _ Button, null);
} // end jbInit()
// Methods ...
/** Button */
void add _ to _ cart _ Button _ Purchase _ GUI()
{ ... }
}

User Interaction and Interface Design with UML

Following the trace of the example, the other

two button-transitions that leave the state query
catalogue remain disabled until the cart is not
empty (button Add to cart):
/** @Button */
void add _ to _ cart _ Button _ purchase _ GUI() {
show _ cart _ Button.setEnabled(true);
proceed _ Button.setEnabled(true);
}

Once the buttons is activated, two other states

can be reached from the state query catalogue:
(a) the state shopping cart, whether the button
show cart is pressed; or (b) the state confirm
proceed, whether the button proceed is pressed.
Since both states are not stereotyped, they are
translated into two independent windows (frames)
with independent behaviours in user-interaction
diagrams, also independent (i.e., with an initial
state and another end state).
Nevertheless, this encapsulated behaviour (i.e.,
a user-interaction diagram) can be interrupted
by means of conditioned simple transitions (or
simply simple transitions). Simple transitions
are those transitions that are labelled only by a
condition (p.e., [screen], [accept], [close],
and [exit]). These transitions refer to an action of pressing a button that exists inside the
non-terminal state (a window). For instance, the
button-transitions [accept] and [screen] refer
that, being in the state (frame) confirm proceed
the user presses one of these buttons, and one
returns to the state (frame) query catalogue. The
same for the transition [close].
Nevertheless, there is a special case of simple
transition: that one that does not come from another state, but it departs from the target state.
For example, the simple transition [exit] means
that there has been pressed the button exit inside
the frame query catalogue, whose behaviour is
defined in another independent user-interaction
diagram (in which the button is defined). In any
case, both the simple-transitions of the type [accept] or of the type [exit] the interpretation

in the code is reflected by means of a listener

operation on those buttons of the window (class)
containing the buttons, that interrupts the normal
interaction sequence of the above mentioned
class.
public class purchase extends querycatalogue {
...
private void jbInit() throws Exception {
...
// Listener on children
/** @Button */
shoppingcart _ GUI.close _ Button.addActionListener(
new java.awt.event.ActionListener() {
public void actionPerformed(ActionEvent
e) {
close _ Button _ ShoppingCart _ GUI();
} } );
/** @Button */
confirmproceed _ GUI.accept _ Button.addActionListener( ... );
/** @Button */
confirmproceed _ GUI.cancel _ Button.addActionListener( ... );
}
...
}

Appendix B partially shows the java code of

one of the five classes of the project example: the
class ShoppingCart. This information is available
in the following Web page: http://indalog.ual.
es/mdd/purchase.

CONCLUSION AND FUTURE WORK

In this chapter, we have studied a model-driven
development technique for GUI design. We have
defined a new kind of UML diagram (i.e., the
user-interface diagram), which specialises in
the use case diagram for user interface design.
In addition, we have shown how to describe use
cases by means of specialized activity diagrams
(user-interaction diagrams) in order to specify
the user interface of each use case. Finally, we
have shown how to generate class diagrams for
GUI and how to build rapid prototypes for the
user interface. As future work, we plan to study
the following extensions of our work:

427

User Interaction and Interface Design with UML

1.
2.
3.

Integrate the user interface view with the

data and business logic;
Incorporate our method in a CASE tool in
order to automate it;
Integrate our technique in the whole development process.

ACKNOWLEDGMENT
This work has been partially supported by
the Spanish project of the Ministry of Science
and Technology TIN2005-09207-C03-02 and
TIN2006-06698 under FEDER funds.

REFERENCES
Anderson, D. J. (2000). Extending UML for
UI. UML2000 Workshop on Towards a UML
Profile for Interactive Systems Development
(TUPIS2000), York, UK, October 2/3.

Elkoutbi, M., Khriss, I., & Keller, R. K. (1999).

Generating user interface prototypes from scenarios. Requirements Engineering99 (pp. 150158). IEEE Computer Society.
Harel, D., & Gery, E. (1997). Executable object
modelling with statecharts. IEEE Computer,
30(7), 31-42.
Heumann, J. (2003). User experience storyboards:
Building better UIs with RUP, UML, and Use
Cases (Tech. Rep.). Rational Software.
Ilogix. (2006a). Rhapsody 6.2. Iogix/Telelogic.
Retrieved from http://www.ilogic.com
Ilogix. (2006b). Statemare. Iogix/Telelogic. Retrieved from http://www.ilogic.com
Kovacevic, S. (1998). UML and user interface
modeling. UML98, LNCS 1618 (pp. 253-266).
Lieberman, B. (2004). UML activity diagrams:
Detailing user interface navigation (Tech. Rep.
29 April). Rational Software.

Campos, P. F., & Nunes, N. J. (2005).

A UMLbased tool for designing user interfaces. UML 2004
Satellite Activities, LNCS 3297 (pp. 273-276).

Nunes, N. J., & Falcao e Cunha, J. (2001). Wisdom

a UML based architecture for interactive systems.
DSVIS 2000, LNCS 1946 (pp. 191-205).

CommentEdition. (2000). UIRupture (Tech. Rep.).

UIDesign. http://www.uidesign.net

Nunes, N. J. (2003) Representing user interface

patterns in UML. In In D. Konstantas, M. Leonard, Y. Pigneur & S. Patel (Eds.), Object-oriented
information systems, LNCS 2817 (pp. 142-151).

Conallen, J. (1999). Modelling Web application

architectures with UML. Communications of the
ACM, 42(10), 63-70.
Constantine, L. L., & Lockwood, L. A. D. (2001).
Structure and style in use cases for user interface
design. In M. van Harmelen (Ed.), Object modelling and user interface design (pp. 245-279).
Addison-Wesley Longman Publishing.
Elkoutbi, M., & Keller, R. K. (2000). User interface prototyping based on UML scenarios and
high level Petri nets. International Conference on
Application and Theory of Petri Nets (ICATPN
2000), LNCS 1825 (pp. 166-186).

428

OMG. (2005). Unified modeling language specification (Tech. Rep. No 2). Object Management
Group.
Paterno, F. (2001). Towards a UML for interactive
systems. EHCI 2001, LNCS 2254 (pp. 7-18).
Pinheiro da Silva, P., & Paton, N. W. (2000).
User interface modelling with UML. Information Modelling and Knowledge Bases XII (pp.
203-217). IOS Press.
Schfer, W. (2002). Fujaba documentation (Tech.
Rep.). University of Paderborn.

User Interaction and Interface Design with UML

APPENDIX A. GUI CODE PATTERNS

A.1. Mapping transitions (buttons) into code
User-Interaction Stereotype
<<JButton>
name

Code Generation
import java.awt.event.*;
public class className extends JFrame {
/** @GUIcomponents */
public JButton name _ Button = new JButton();
private void jbInit() throws Exception {
/** @Button */
name _ Button.setText(name);
name _ Button.addActionListener(
new java.awt.event.ActionListener() {
public void actionPerformed(ActionEvent e) {
name _ Button _ nameClass _ GUI(); } });
/** @Panel */
this.getContentPane().add(name _ Button, null);
} // end jbInit()
/***
* Method THIS
*/
/** Button */
void name _ Button _ className _ GUI() { ... }
}

A.2. Mapping labels to code

User-Interaction Stereotype
<<JLabel>>
name

Code Generation
public class className extends JFrame {
/** @GUIcomponents */
public JLabel name _ Label = new JLabel();
private void jbInit() throws Exception {
/** @Label */
name _ Label.setText(name);
/** @Panel */
this.getContentPane().add(name _ Label, null);
} // end jbInit()
}

A.3. Mapping textfields to code

User-Interaction Stereotype
<<JTextField>>
name

Code Generation
public class className extends JFrame {
/** @GUIcomponents */
public JLabel name _ Label = new JLabel();
public JTextField name _ TextField = new JTextField();
private void jbInit() throws Exception {
/** @Label */
name _ Label.setText(name);
/** @Panel */
this.getContentPane().add(name _ Label, null);
this.getContentPane().add(name _ TextField, null);
}// end jbInit()
}

429

User Interaction and Interface Design with UML

A.4. Mapping lists to code

User-Interaction Stereotype
<<JList>>
name

Code Generation
import javax.swing.event.*;
public class className extends JFrame {
/** @GUIcomponents */
public JLabel name _ Label = new JLabel();
public JScrollPane name _ ScrollPane = new JScrollPane();
public JList name _ List = new JList();
private void jbInit() throws Exception {
/** @Label */
name _ Label.setText(name);
/** @List */
name _ ScrollPane.getViewport().add(name _ List, null);
name _ List.addListSelectionListener(
new javax.swing.event.ListSelectionListener() {
public void valueChanged(ListSelectionEvent e) {
selected _ articles _ List _ ShoppingCart _ GUI(); } });
/** @Panel */
this.getContentPane().add(name _ Label, null);
this.getContentPane().add(name _ ScrollPane, null);
} // end jbInit()
/***
* Method THIS
*/
/** @List */
void selected _ articles _ List _ ShoppingCart _ GUI() { ... }
}

430

User Interaction and Interface Design with UML

APPENDIX B. THE SHOPPING CART FRAME

public class ShoppingCart extends JFrame {
/** @GUIcomponents */
public JLabel shopping _ cart _ Label = new JLabel();
public JLabel selected _ articles _ Label = new JLabel();
public JButton remove _ article _ Button = new JButton();
public JButton close _ Button = new JButton();
public JScrollPane selected _ articles _ ScrollPane = new JScrollPane();
public JList selected _ articles _ List = new JList();
/** @GUI */
ConfirmRemoveArticle confirmremovearticle _ GUI = new ConfirmRemoveArticle();
public ShoppingCart() { // Constructor }

private void jbInit() throws Exception {
/** @Label */
shopping _ cart _ Label.setText(Shopping Cart);
...
/** @Button */
remove _ article _ Button.setText(Remove article);
remove _ article _ Button.setEnabled(false);
remove _ article _ Button.addActionListener(new java.awt.event.ActionListener() {
public void actionPerformed(ActionEvent e) {
remove _ article _ Button _ ShoppingCart _ GUI(); } });
...
/** @List */
selected _ articles _ ScrollPane.getViewport().add(selected _ articles _ List, null);
selected _ articles _ List.addListSelectionListener( ... );
/** @Panel */
this.getContentPane().add(close _ Button, null);
...
/***
* Listener on children
*/
confirmremovearticle _ GUI.cancel _ Button.addActionListener( ... );
confirmremovearticle _ GUI.accept _ Button.addActionListener( ... );
} // end jbInit()
/***
* Methods THIS
*/
/** @Button */
void remove _ article _ Button _ ShoppingCart _ GUI() {
this.setEnabled(false); confirmremovearticle _ GUI.setVisible(true);
}
/** @Button */
void close _ Button _ ShoppingCart _ GUI() { this.setVisible(false); }
/** @List */
void selected _ articles _ List _ ShoppingCart _ GUI() {
remove _ article _ Button.setEnabled(true);
}
/***
* Methods children
*/
/** @Button */
void cancel _ Button _ ConfirmRemoveArticle _ GUI() { this.setEnabled(true); }

/** @Button */
void accept _ Button _ ConfirmRemoveArticle _ GUI() { this.setEnabled(true); }

This work was previously published in Visual Languages for Interactive Computing: Definitions and Formalizations, edited
by F. Ferri, pp. 328-356 , copyright 2008 by Information Science Reference, formerly known as Idea Group Reference (an
imprint of IGI Global).

431

432

Chapter 2.5

Design Methods for Experience

Design
Marie Jefsioutine
University of Central EnglandBirmingham, UK
John Knight
University of Central EnglandBirmingham, UK

ABSTRACT

INTRODUCTION

The following chapter describes an approach

to Web design and evaluation where the user
experience is central. It outlines the historical
context in which experience design has evolved
and describes the authors experience design
framework (EDF). This is based on the principles
of user-centred design (UCD) and draws on a
variety of research methods and tools to facilitate
the design, development, and evaluation of user
experiences. It proposes that to design usable,
accessible, engaging, and beneficial Web sites,
effort needs to focus on visceral, behavioural,
reflective, and social factors, while considering
contexts such as the who and why; what and how;
when and where; and with what of Web site use.
Research methods from a variety of disciplines
are used to support exploration, communication,
empathy, and speculation. Examples of the application of the EDF, to various stages of the Web
design process, are described.

Although change is afoot, designers (including

design engineers)still worry that corporate
bosses and clients see them as merely making
things pretty.
Molotch, 2003, p. 28
Producing Web sites is a process that involves a
range of skills and disciplines. Design is not an
add-on to make the screens look good. Design
impacts on what people do, how they do it, and
with whom they do it. Products, whether they are
Web sites, toasters, or services embody a range
of values and our interaction with them is a form
of communication. When interacting with a Web
site, for example, we may be communicating
with other people, real or imagined. As we type
with the keyboard, we hear and feel the keys and
may vent our frustration through the force of our
tapping. While reading the words on a screen we
are also taking in all kinds of nonverbal messages

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Design Methods for Experience Design

from the layoutimages, colours, fonts, icons,

language, and style of language. Our experience is
also affected by our surroundings, our memories
of past actions, our current knowledge, and our
expectations of the future.
Interaction with a Web site is mediated by
a network of people, machines, and systems.
The Internet works because of the design of
protocols, browsers, laws, technical standards,
security mechanisms, machines, communication technologies, and physical infrastructure as
much as it does because of the design of the Web
pages. Our experience of a Web site is therefore
a product of our personal understanding of its
context. Givechi and Velzquez (2004) describe
the positive space of a product, which becomes
more significant than the product itself, the aura
of a product, the sum of its physical attributes plus
its intangible essenceor the meaning it hosts for
each of its users (p. 43). The meanings we attach
to a product change with time, in part due to our
changing experiences, the wider socioeconomic
and political context of a product, and our changing expectations. To build a successful product,
a designer needs to be aware of as many of these
factors as possible, and this is why a consideration
of the whole user experience is important.

FROM SOFTWARE ENGINEERING

TO EXPERIENCE DESIGN
Concepts like user experience and design
of experience are common in the design and
business communities now (Fulton-Suri, 2004,
p.14), but this has not always been the case, and
in some software development circles designer
still has negative connotations. It is important to
recognise the context in which Web design has
emerged, to understand why these concerns are
so topical.
The development of computer software and
interfaces in the 1960s emerged from fields traditionally associated with engineering and science.

This era was typified by optimism for technology,

and even in the traditional design disciplines there
was a movement towards rationalising design
methods. Hailed as the design science decade
(Fuller, 1969, p. 305) the emphasis was on objectivity, rationalism, and technology.
The term software engineering can be traced
back to an international conference convened
by NATO in 1968, to solve the software crisis
(Campbell-Kelly & Aspray, 1996, p. 200). The
crisis emerged from large scale (often military)
software projects that encountered problems in
management and quality. In response to these failings a number of risk management strategies were
advocated including the waterfall development
lifecycle. Here design and development flowed
through a predetermined course of phases. The
formalised nature of the process was predicated
by each phase having specified inputs and outputs,
which could be checked along the way.
A design approach based purely on linear,
logical problem solving did not work for the new
technologies (e.g., computing, software, and solid
state electronics), where problems were nebulous
and constantly evolving through technological
development and changing requirements. Brooks
(1975) experience on IBMs System/360 project
was typical. He describes how software defied
traditional logical engineering approaches. Even
increasing resources did not improve the success
of the approach.
In design disciplines such as product design
and architecture, there was a growing realisation
that some problems were unsolvable by logical
deduction (Cross, 2001). Rittel and Webber (1973),
for example, contrasted the wicked problems of
design with the tame ones of science and engineering. In software design attention was turning
to human factors. Whereas traditional engineering
and industrial design focussed largely on external
and measurable ergonomic factors, computer
interaction required an understanding of internal, cognitive aspects. In 1969 the International
Journal of Man Machine Studies was launched

433

Design Methods for Experience Design

and a new field emerged, looking at what is now

called human-computer interaction (HCI).
The idea that the computer interface could
be modelled on human needs rather than driven
by system capabilities led to the development of
the graphical user interface (GUI) launched with
one of the early commercially available personal
computers, the Xerox Star, in 1981. Although the
product itself was not successful, the concept of
the GUI became almost universally adopted. This
interface had been developed using usability engineering methods, including paper prototyping
and testing on potential users. The design process
was iterative and responsive.
In 1985, Gould and Lewis codified their own
approach to software development, identifying
the following key principles: early focus on users
and tasks; empirical measurement through early
and continual user testing; integrated design; and
iterative design (Gould 1995; Gould & Lewis,
1985). UCD, as it came to be called, is now embodied in international standards for software
development. International Organization for
Standarization (ISO, 1999) 13407, for example,
describes it as:
an approach to interactive system development
that focuses specifically on making systems usable.
Whatever the design process the allocation of responsibilities and roles adopted, the incorporation
of a human-centred approach is characterised by
the following:
a.
b.
c.
d.

the active involvement of users and a clear

understanding of user and task requirements;
an appropriate allocation of function between users and technology;
the iteration of design solutions; and
multidisciplinary design.

Usability became a high profile issue in the

1990s (see Knight & Jefsioutine, 2002) and
much of the HCI literature aimed at supporting

434

interaction designers, focused on cognitive and

behavioural models of human interaction with
attempts to formalise methods (e.g., Harrison
& Thimbleby, 1990) and generate guidelines
about human behaviour, much like the rational
approaches to design.
The relationship between humans and computers is not entirely rational or logical however.
Reeves and Nass (2002), for example, describe
the ways in which people treat computers and
software like real people with feelings and personalities. Jordan (2000) argues that usability is
no longer sufficient as a design goal. He argues
that customers now expect products to be easy
to use and claims that usability has moved from
what marketing people call a satisfier to being
a dissatisfier (Jordan, 2000, p. 3). People no
longer notice when a product is usable, just when
it is difficult to use. Fulton-Suri (2004) points out
established products have become more similar
in technology, functionality, price and quality,
companies have turned to design to differentiate, their offeringsto create stronger emotional
connections with their customers (p. 13). Lastly,
the notion of a problem that can be identified and
solved through logical deduction is often at odds
with how successful products are developed and
adopted by consumers.
The challenge for Web designers is no longer
just to produce a functioning and usable product,
they must now meet the needs of an increasingly
sophisticated and demanding audience and a
competitive market place. The EDF was developed by the authors to address these challenges
by collating useful methods and approaches to
designing products like Web sites and by considering the whole user experience (Jefsioutine
& Knight, 2004).

Design Methods for Experience Design

THE EXPERIENCE DESIGN

FRAMEWORK
The EDF advocates the principles of UCD already
described, and adds another set of considerations:
qualities, dimensions of experience, research
contexts, and research methods. These are described hereafter.

Qualities
Each product will have its own specific user
requirements. The EDF proposes that four fundamental qualities underlie these requirements.

Usable
To go beyond usability does not mean that it is
no longer necessary. The benefits of usability are
well documented (e.g., Bevan, 2000). Usability is
defined as the extent to which a product can be
used by specified users to achieve specified goals
with effectiveness, efficiency and satisfaction in a
specified context of use (ISO, 1998) and the capability of the software product to be understood,
learned, used and attractive to the user, when used
under specified conditions (ISO, 2000). Indeed,
in long-term use of a product, there is evidence

that usability becomes more important than style

in predicting users satisfaction (Maguire, 2004).
Usability cannot be ignored as a design goal.

Accessible
The Web relies on a level of standardisation
such that pages can be accessed by anyone, on
whatever browser or device they use, be it a PC,
Mac, or mobile phone. As the Web has grown,
so have the standards and so have the number of
inaccessible Web sites. Despite the publication of
guidelines for accessible web design by the World
Wide Web Consortium (W3C, 1999), a survey of
Web designers found that difficulty interpreting
guidelines was a major barrier to implementing
them (Knight & Jefsioutine, 2003). Furthermore,
the UKs Disability Rights Commission (2004)
found that nearly half of the usability and accessibility problems were not violations of any of the
WCAGs checkpoints. DiBlas, Paolini, Speroni,
and Capodieci (2004) argue that W3C guidelines
are not sufficient to ensure an efficienteven
less satisfactoryWeb experience (p. 89). It is
important that accessibility is seen as part of the
user experience rather than a series of technical
checkpoints to cover.

Figure 1. The experience design framework (Jefsioutine & Knight, 2004)

435

Design Methods for Experience Design

Engaging
Shedroff (2001, p. 4) suggests an experience
comprises of an attraction, an engagement, and
a conclusion. What attracts someone to a product
could be a need to perform a task, an aesthetic
quality, or an affordance. The engagement is then
sustained over a period of time, beyond the initial
attraction. Csikszentmihalyi (1991) describes
the experience of optimal experience and flow:
Concentration is so intense that there is no attention left over to think about anything irrelevant,
or to worry about problems. Self-consciousness
disappears, and the sense of timing becomes
distorted (p. 71).
Although this level of engagement may not be
appropriate for all products, it is useful to consider
the properties of an experience that make it engaging. Fiore (2003) suggests that an experience
includes a number of dimensions: it is educative
and memorable; whole, unique and nonreproducible; historical; meaningful/aesthetic; contextual;
physical/sensual/embodied; and situated in time
and space. Jones, Valdez, Nowakowski, and Rasmussen (1994) suggest that engaged learning tasks
are challenging, authentic, and multidisciplinary.
Such tasks are typically complex and involve
sustained amounts of time and are authentic.
Quinn (1997) suggests that engagement in learning
applications comes from two factorsinteractivity and embeddedness, where the user perceives
that they have some control over the system, and
it is relevant and meaningful to them.

Beneficial
According to Csikszentmihalyi (1991), an optimal
experience is so gratifying that people are willing
to do it for its own sake, with little concern for
what they will get out of it (p. 71). One might
assume that what they are getting is some degree
of pleasure. Jordan (2000) identifies pleasure as
the ultimate quality of the user experience, and
DeJean (2002) points out that pleasure is a complex

436

concept. Apparently unpleasant aspects of the

user experience, such as difficulty, challenge, and
fatigue can all be pleasurable in certain contexts,
for example, by generating feelings of achievement or superiority.
Bonapace (2002) adapts Maslows (1970) hierarchy of human needs to product use qualities.
His pyramid begins with safety and well-being,
moves up to functionality, and then usability which
leads up to an apex of pleasure as the ultimate
quality. Dunne (1999) explores the aesthetics of
use and argues, The most difficult challenge
for designers of electronic objects now lies not
in technical and semiotic functionality, where
optimal levels of performance are already attainable, but in the realms of metaphysics, poetry and
aesthetics, where little research has been carried
out (p. 7).
Liu (2003) describes qualities of psychosomatic soundness, referring to the extent that
a product contributes to the wholesomeness
or well-being of a person (from harmful to
healthful) and the degree to which it is ethical
(bad/wrong to good/right) (p. 1296). Knight
(2004) too, argues for a reconsideration of the ethics of HCI design. The EDF uses the term benefit
to include such concepts as ethics, psychosomatic
soundness, pleasure, and self-actualisation.
The EDF, therefore, advocates designing for
the four fundamental qualities of product use of
accessibility, usability, engagability, and benefit.
To do this it becomes necessary to widen the
focus of research beyond cognitive and behavioural interaction, typical of usability, and HCI
studies, to include a multi-dimensional approach
to experience.

Dimensions of Experiencing
McDonagh-Philp and Lebbon (2000) suggest that
emphasis must change from hard functionality,
to soft functionality (p. 38). Rather than focussing on what a product does and how it does it,
the focus is on less tangible aspects like emotional

Design Methods for Experience Design

associations, familiarity, aesthetics and taste.

Fiores (2003) framework of experience considers
physical, emotional, and intellectual aspects. In a
similar vein, Norman (2004) describes emotional
design in terms of its visceral, behavioural, and
reflective elements (p. 63). He points out that
these three levels of experience are not discrete
but interact with each other. Spillers (2004) suggests, for example, that a new icon on a screen
could arouse a state of curiosity or annoyance,
producing a change in the users emotional state
which can either propel the user toward a feeling of satisfaction (success) or disappointment
(failure) (p. 2).
Jordan (2000) develops Tigers (1992) concept
of pleasure, and describes four ways in which pleasure can be experienced: Socio-Pleasure arises
from interaction with others or from a product that
represents a social grouping; Pyscho-Pleasure
comes from the satisfaction felt when a task is
successfully completed or from a product that
makes a task more pleasurable. Physio-Pleasure
is derived from the senses; and Ideo-Pleasure
is derived from entities such as books, art and
music or the values that a product embodies (p.
13-14).
The EDF uses Normans (2004) classification
and adds a social dimension. The EDF, therefore,
directs attention to the visceral, behavioural,
reflective, and social dimensions of experience.
Furthermore, it suggests that they be considered
in the context of the use qualities. For example,
considering accessibility in the context of each
dimension may generate design goals such as:

Accessible/visceral: legibility and visual

clarity, text alternatives to audio content
Accessible/reflective: limiting cognitive
overload /clear, simple language use
Accessible/behavioural: keyboard alternatives to using a mouse, shortcuts, voice
input
Accessible/social: culturally inclusive,
secure, private, moderated

Research Contexts
This section describes a set of key questions that
can be asked throughout the design process to develop an understanding of the contexts of product
use. The questions are derived from a number
of models including Rothsteins (2002) model
consisting of the four As of activity, artefacts,
atmosphere and actors (p. 3), and Ortony, Clore,
and Collins (1998) cognitive model comprising
events, agents and objects (p. 63). The EDF
advocates four key contexts: (1) who and why
(users and stakeholders and their motivations); (2)
what and how (content, tasks, task flow, actions,
functionality); (3) when and where (situation,
frequency, environment); and (4) with what (tools,
knowledge, and skills).

Who and Why

In order to be user centred it is necessary to
identify who the users and stakeholders are
and what motivates them to use a product. It is
important to include everyone, however limited
their involvement with a product might be. Users
of a software system, for example, might include
people that buy it, use it at home or at work, communicate through it, people who sell it, or sell on
it, administrate it, repair it, install it or support it
(e.g., Hackos & Redish, 1998).
When the focus of UCD goes beyond usability,
it becomes necessary to collect data that pertain
to the qualities and dimensions of the EDF. This
might include: demographics (such as age, gender,
ethnic origin, and culture); behaviour and skills
(such as computer literacy, typing skills, embedded knowledge of a task or system); knowledge
and experience (such as novice or domain expert,
tacit knowledge); personal characteristics and
motivations (such as personality, learning style,
attitude, aspirations, values, beliefs, tastes, and
preferences); and physical characteristics (such
as dexterity, physical abilities, height).

437

Design Methods for Experience Design

What and How

This refers to the tasks or activities that will be supported, influenced, or affected by the product and
how users carry out these tasks. Tasks are typically
mapped from a behavioural perspective or cognitive dimension. Liddle (1996) suggests, The most
important component to design properly is the
users conceptual model. Everything else should
be subordinated to making that model clear, obvious, and substantial (p. 21). The EDF suggests
that activities be considered in all dimensions.
For example, an airhostess may be performing
complex emotional tasks as well as checking in
baggage (customer relations, anxiety reduction,
risk assessment, etc.) (Hochschild, 1983).

When and Where

The role of context in understanding tasks was
emphasised by Suchmans (1987) notion of situated actions, that activity is conditional on the
situation in which it takes place and is of an
improvised rather than planned nature. What
people say they are doing or how they do it, is
an after-the-event rationalisation. Context should
be considered in terms of all of the dimensions.
For example, what is the users current frame of
mind? What ideological factors will influence
their experience? What aspects of the environment affect their visceral experienceare there
loud distracting noises around or reflections on
the screen? What emotional state is the user in?
What other behaviours are they performing? What
is the social contextis the product being used
with friends to communicate, to play, to learn, or
to work cooperatively?

With What
This refers to objects, artefacts, or tools that are
being used or are influencing use (such as software,
browsers, input devices, assistive technologies),
and to knowledge, expertise, and skills that are

438

used to carry out tasks. Considered in the context

of qualities and dimensions, brainstorming, for example, might suggest that a haptic interface or 3D
glasses (visceral) may improve engagability when
viewing a virtual museum object. Conversely, users knowledge of existing interface conventions
(reflective) may create expectations that reduce
the usability of an innovative interface.

Methods and Tools

Reeves and Nass (2002) point out that people are
rarely aware of their less rational motivations, so
attempts to model what people are thinking or
doing by asking them will not necessarily capture
the reality of an experience. They advocate the
use of methods from the social sciences to establish what people really think. The EDF suggests
casting a wide net across many disciplines to find
appropriate and useful methods.
Rather than prescribing a process or method,
the EDF suggests that a range of tools and techniques can be employed provided they cover four
basic purposesexploration, empathy, communication, and evaluation. Furthermore, by applying
these methods to the dimensions, qualities, and
research perspectives, a better understanding of
the user experience as a whole can be achieved.

Exploration
These methods are about discovery and can be
drawn from demography, ethnography, market
research, psychology, and HCI. They include
surveys, interviews, questionnaires, focus groups,
task analysis, field observation, user testing, affinity diagramming, laddering, and experience
diaries. A key area of exploration is in understanding users mental models of a domain. This can
be explored by, for example, in depth elicitation
techniques and card sorts.
Contextual interviews are conducted during the activity or in the environment in which
the product will be used. Users are able to refer

Design Methods for Experience Design

to artefacts, such as the documents, memos,

and equipment that they normally use in their
workflow, and indeed may be encouraged by the
researcher to describe what they are doing. One
of the most common methods of achieving this
is to use the think aloud protocol (where users
verbalise their actions). This has the advantage
that the user is not relying on memory to describe
his or her actions, and the researcher is able to
note and probe omissions. There are some disadvantages however. The act of conscious reflection
may change the way the task is performed and
may create excessive cognitive load, or compete
for the same cognitive channel as the task.
Both ethnography and ethnomethodology
have been applied to eliciting contextual data.
Ethnography is the study of a culture achieved by
researchers immersing themselves in that culture
in order to understand it, while ethnomethodology
studies the ways in which participants give order to
and make sense of their social worlds (Garfinkel,
1967). The focus of ethnomethodological studies
is often at a very detailed level of interaction, including interaction between people and artefacts,
technologies, and systems. For this reason it is a
useful approach for the study of complex work
situations, for example, air traffic or train control
systems (Heath & Luff, 2000). The researcher may
also use the artefacts produced by organisations,
such as manuals and training materials; policies
and procedures; and forms and documentation to
enrich the detail of the observation. The rationale
for applying these approaches to design is that
the knowledge acquired will enable designers
to work with, rather than against, users ways of
understanding and making sense of the activities
in which they are engaged.
Nevertheless, integrating these methodologies within the design process is not without its
problems. The skills of trained ethnographers
are often underestimated. Done properly, ethnographic methods involve sophisticated and skilful observation and recording techniques. The
data may be far more detailed than the designer

needs, and in a form that is difficult to interpret.

Furthermore it can take many years to do a full
ethnographic study, and for most software development projects, this amount of time is unrealistic.
A simpler version may include shadowing target
users for a short time, noting what they do and
use, taking photographs, asking questions, and
using techniques to elicit their values or emotional
responses (Beyer & Holtzblatt, 1998).

Communicating
Design involves communication with a wide
range of people, from users to software engineers.
Design teams need to accommodate different
viewpoints and share a common language. Curtis
(2002) emphasises the importance of actively
listening to a client and finding out the story
behind a product. Hammer and Reymen (2004)
stress the importance of designers expressing
their emotional as well as rational reflections on
design decisions. Communication methods serve
to clarify and share the goals of stakeholders, the
exploratory research data, design requirements,
and ideas to a multi-disciplinary team who may
not have a common vocabulary. It is important
to ensure good communication throughout the
process to ensure the product itself communicates
the design goals effectively. Methods include story
telling; user profiles and personas; use cases or task
scenarios; scenario-based design; mood boards;
written briefs and specifications; storyboarding;
and prototypes.
User profiles are generated from demographic
data and lifestyle surveys. They can include textual
and visual descriptions of key user groups. They
are used to think through design solutions and for
recruiting users for research. Mood boards are
normally collages of photographic information
that aim to generate a visual personality for the
product or service. Mood boards can be created
by designers or users or be the result of collaboration between the two. Mood boards are useful
because they work at a nonverbal level where

439

Design Methods for Experience Design

people may otherwise have difficulty expressing

their wants and needs. Storyboards are timebased often with a narrative aspect. Storyboards
can be sketches or low-fidelity screen shots of a
users interaction. Prototypes range from paper
sketches to working interactive replicas. Information architecture can be communicated through
formalised diagrams and charts or simple tree
structures and hierarchies.

Empathy
These methods represent an approach aimed at
gaining a deeper understanding and empathy
for users. They include focus groups, diaries,
workshops, participatory design, and immersion.
Diaries can be used to record users interaction
with a product over time while workshops and
participatory design involve users in the development team either directly or through a user
advocate that champions their perspective. Participant observation, or eat your own dog food,
involves taking part in the activity or culture being
observed, where the designer becomes a user.
Molotch (2003) describes a method used by Ford
designers in which they test products dressed in
what they call a third age suit, with glasses and
gloves, to simulate having the body and eyesight
of a 70-year old (p. 49).
Crossley (2004) describes a technique used
to help designers develop an awareness of the
target audience for male grooming products. It
involved:
rapidly immersing the design team into the lives,
hearts and minds of people in a short space of time.
The challenge for this project was to get young men
inspired to tell us their own stories and express
their emotions about a mundane functional activity Character modelling [was used] where
the team and sometimes the user has a kit with
questions, cameras and collages, [that enabled
them] to frame and understand the lifestyle of the
person they are creating for. (pp. 38-39)

440

Personas are used to develop a shared vision

of end users among development teams. They
can be textual, visual, animated, or acted. One
of the most important functions of personas is
to get teams to think differently about people, as
Cooper (1999) notes:
A fully realized, thoroughly defined user persona is a powerful tool. Until the user is precisely
defined, the programmer can always imagine that
he is the user. A completely pronounced user
persona is key to the suppression of any tendency
for the developer to usurp or distort the user
personas role. Long before a single line of code
is written, a well-defined user persona becomes
a remarkably effective tool for interaction design.
(pp. 128-129)
Role-play is often used to encourage designers to empathise with users. Dramatic techniques
can also help teams get into the minds of users
and their values. Carmichael, Newell, Morgan,
Dickinson, and Mival (2005) describe how the
UTOPIA project improved designers understanding of requirements. Using professional actors
the project developed video scenarios of elderly
people using technology. This was then shown
to design teams to encourage them to design for
others rather than themselves.

Speculation
In addition to understanding users wants and
needs, designers also need to speculate about new
solutions and future trends. The Sony Walkman,
for example, introduced an entirely novel mode of
behaviour that no users had asked for. The decision to add text messaging to mobile phones was
based on a speculation of how that functionality
might be needed and by whom. The success of
text messaging, however, was based on its uptake
by an entirely different user group with its own
needs and method of use. Designers may need
to predict how users will adopt and adapt to a
new product.

Design Methods for Experience Design

Here the solution is not tied to a particular

technology or need but to reconceiving it. Good
design requires up-to-date knowledge of what is
possible and the ability to see beyond the obvious. There are a number of methods that achieve
this. Jones (1990) sees speculation as divergence,
transformation and convergence (p. 64-68) and
suggests methods such as brainstorming (search
for alternatives) and synetics (search for alternatives by analogy).
Rosson and Carroll (2002) describe scenariobased design in which scenarios are deployed
throughout the design process to speculate the
future. Scenarios can be used to challenge existing
practices and encourage users and designers to
think beyond the confines of the current situation.
Dunne (1999) describes the use of scenarios to
communicate design ideas and to stimulate thinking beyond the preexistent reality. He suggests that
they can push the viewer towards a more complex, emotional or revolutionary understanding
of the problems posed (p. 75). By testing design
ideas rather than prototypes, the users attention
is shifted from the aesthetics of construction
to the aesthetics of use (p. 73).
Speculation and innovation are, in part, about
predicting the future, and a number of methods
have been developed for this. The UK Cabinet
Office published a Futurists toolkit (UK Cabinet
Office, 2001) describing six key methodologies
for futures work, including quantitative and
qualitative trend analyses, the Delphi survey
method, scenario methods, wildcards, and future
workshops. Cayol and Bonhoure (2004) describe
a methodology which used sociological studies
predicting the evolution of French people for the
year 2025, to identify and evaluate future product
concepts.
De Bono (1995) uses lateral thinking to generate different scenarios and ways of thinking about
problems. De Bonos Six Thinking Hats (pp.
78-79) is a method for creative thinking but is also
useful for developing empathetic intelligence. This
method involves six team members role-playing

different aspects of the design process. Thus the

white hat is neutral and reflects back opinions and
ideas, whereas the black hat is judgemental, cautionary and avoids mistakes. Speculative methods
are important because design teams need to be
able to see beyond user requirements and consider
latent needs and potential opportunities.

Evaluation
Evaluation methods include auditing, standards
compliance, and user testing. Evaluating may
also use similar methods to exploration, with
a shift in emphasis from discovery to checking
outcomes against intentions. Does the product
meet the design goals and/or the user expectations?
Evaluation can be formative, conducted during
the development of a product, or summative, conducted when a product is complete and is being
used. Summative testing of an earlier version or
a similar product can be useful to identify design
goals, while summative testing of the product at
the end of its design lifecycle is usually done for
auditing and verification purposes. Feedback at
this stage is of little use to the designer, the deadline has passed and the money is spent. Clearly
formative testing is most helpful to a designer/
developer. Gould and Lewis (1985) stress the
importance of empirically testing design iterations throughout the design process. Evaluative
tools such as heuristics are often used, although
evidence suggests that they are no substitute for
testing real users (e.g., Lee, Whalen, McEwen, &
Latremouille, 1984). The EDF broadens the test
and evaluative criteria from the traditional focus
on cognitive and behavioural measures. Bonapace
(2002) describes a method aimed at tapping into
the four pleasures described by Jordan (2000),
called the Sensorial Quality Assessment Method
(SEQUAM) applied to the design of physical
products in car manufacturing. User testing can
combine empirical methods of behavioural observation with techniques such as co-discovery,
think aloud and empathic interviewing, to tap into

441

Design Methods for Experience Design

the subjective aspects of experience.

Establishing evaluation criteria at the beginning of a design process helps to focus the evaluation process from the beginning, although it is wise
to allow some freedom for the evolution of design
goals through iteration and to allow evaluative
concerns to emerge (Hall 2005). The EDF can be
used to generate evaluative criteria for qualities
in each dimension and context of use.

Applying the Experience

Design Framework
The authors have used the EDF to adapt and focus
methods for requirements research, brief development, ideation, and testing and have developed
a range of services and training based on it. The
EDF has been particularly useful in generating
design goals, aiding decision making, and developing user-testing scenarios. Some examples of
applications follow.

Visioning Workshops
Visioning workshops usually take place at the
beginning of the design process, preferably before
the brief is finalised. The workshop is structured
around the experience design framework and fulfils a number of functions. Techniques are adapted
to improve communication; to build literacy and
a common language to describe the medium; and
to build an empathic understanding of users and
other contexts of use. Individual and group activities are facilitated by a researcher and recorded
by a scribe. Activities include the following:

442

Participants identify their vision of the

project goals and their personal objectives,
roles, and stories.
Participants share their personal preferences
and tastes by discussing examples of the
product type or medium from visceral, behavioural, reflective, and social perspectives,

and in so doing, build up a shared language

for product qualities.
Participants discuss the nature of the design
problem from different perspectives and
use speculative techniques to generate new
ideas.
Participants identify audience and stakeholders; then develop personas and scenarios
based on different contexts of use.
Participants share desired product qualities,
functionality, and content from the perspective of the stakeholders identified.
Participants identify and prioritise product
goals and develop evaluative criteria.

By the end of the workshop, participants have

prepared the foundations of a creative brief, have
started working together as a team, and have
formed a common language and shared understanding of the project goals.

Contextual Interviews
Contextual interviews take place prior to detailed
requirements specifications and may follow a
visioning workshop. Firstly, stakeholders are
identified and a representative sample recruited.
The aim is to survey a large sample and to iteratively develop knowledge about the design
problem. In this context, as many as 50 users may
be involved and the focus is to gain as full a set
of requirements as is possible in a short space of
time. The interviews are semi-structured over
approximately 30 minutes and are conducted
within the context of use. Their exact format is
dependent on whether the product is a new one or
a refinement of an existing one. In the former case,
the researcher works with low-fidelity prototypes
and in the latter case with the existing product.
Activities include the following:

Stakeholders are asked to provide documents, processes, and artefacts involved in

the tasks.

Design Methods for Experience Design

Interviewees are asked to complete a persona

template with details about themselves, their
interests, and their lives.
Interviewees are asked to identify critical
tasks, events, and work-arounds with the
existing product.
Interviewees identify relationships with
other users/stakeholders in their use of the
product, and what other tools or products
they use in association with the tasks.
Interviewees are asked to describe key tasks
with the product and/or walk through a task.
The interviewer elicits details of any prior
knowledge, expertise, or skills being applied to the tasks and probes for emotional
responses to aspects of the activity.
Interviewees are asked to describe how the
product fits in with their daily life.

The results of interviews inform the production of anonymous personas and use scenarios,
which are used to communicate the requirements
to the development team and build their empathy
with the users.

Conceptual Design Workshops

Conceptual design workshops involve development teams rethinking the design problem and
considering potential solutions. The aim is to
generate a number of alternative design concepts.
These are then evaluated and then worked up
for initial user testing or participatory prototyping
(see what follows). Workshop members usually
include members of the development team plus
representatives of key stakeholders. The workshops are structured around the EDF and activities
include the following:

Participants reconceptualise the design

problem in terms of the EDFs contexts.
Participants brainstorm and develop a number of design concepts.
The concepts are discussed and similar so-

lutions are merged together and evaluated

against the EDFs qualities.
The process is repeated until a number of
distinctly different and viable concepts
have been generated and can be tested on
users.

Participatory Prototyping
Participatory prototyping combines the skills
of the development team with user feedback.
Prototypes are developed of content structures,
interaction flow, and layouts. Initial prototypes
are developed and users are asked to critique
or adapt the prototype with specific reference
to the qualities of the EDF and the dimensions
of experience. Designers interpret this feedback
and develop further prototypes, with the focus on
interaction and structure, and later on look and
feel issues. Activities include the following:

Users are asked to card sort content or to map

out their expectations of content domains
and structures.
Users are asked to walk through prototypes
in order to carry out their key tasks, using
the think aloud method.
Responses to the organisation of the interface
and the terminology used are elicited.
Users perform tasks with the prototype and
are asked to speculate on improvements,
applications, or additional features.
Where trade-off or multiple interfaces exists
users sort the prototypes on the basis of their
preference.

By the end of the process a complete low-fidelity prototype has been developed that has been
iterated around the qualities of the EDF.

Audience Reception Workshops

Audience workshops review the final prototype
design. Users are recruited to represent the key

443

Design Methods for Experience Design

stakeholder groups. The prototype then undergoes a group critique that tests the solution
against the initial requirements and evaluative
criteria gathered by the visioning workshop and
contextual interviews. As well as ensuring that
the prototype is suitable the workshops gauge
barriers and opportunities to take up and adoption of a new product or design. In addition, it
provides the development team with a rationale
and evidence of the suitability of the final design
solution. Activities include the following:

Group walk-throughs of the prototype

Identification of conflicts and trade-offs
Comparison of look and feel prototypes
Quantitative research methods and user
attitude measurement

Post Implementation Research

Post implementation research reviews the application after launch. Unlike traditional usability
that focuses on requirements and development,
this research is necessary to monitor changing
user needs throughout the products lifecycle. A
user cohort is recruited and is asked to provide
regular feedback based on the EDF. As well as
ensuring that maintenance and new features are
accepted by users this research is important to
identify new products and user needs. Activities
include the following:

User diaries
Online discussion groups and surveys
Focus groups and ongoing user testing

Conclusions and the future

of web experiences
The EDF was created in response to the needs
of clients and has generated research, which has
fed back into the design of services. The EDF has
been applied to a number of internal and external

444

products and has been refined through practice.

The EDF is meant to provoke discussion, raise
questions, challenge assumptions, and generate
alternatives.
Every design problem is different and so it is
necessary to deploy a range of research methods
to support the work of the digital media designer.
Nevertheless, these methods should be focused
around the key issues for UCD, which is to understand users, the tasks they undertake, and the
contexts in which they function. They provide
a rich understanding of users from which the
creative designer can create usable and desirable
products. The EDF helps teams focus requirements research and can be used in a workshop
environment with a range of stakeholders and
users. The EDF can be used as a brainstorming
tool that aims to map out the requirement and the
process, and possible solutions from the team and
stakeholders.
If methods are focused around qualities, dimensions, and contexts of experience, they provide
a richer understanding of users from which the
designer can create usable and desirable products.
The EDF provides a reminder of the complexity
of the user experience and can be used throughout
the design lifecycle.

References
Bevan, N. (2000). Esprit Project 28015 Trump:
Cost benefit analysis. Retrieved April 26, 2002,
from http://www.usability.serco.com/trump/documents/D3.0_Cost%20benefit_v1.1.doc
Beyer, H., & Holtzblatt, K. (1998). Contextual
design: Designing customer-centred systems. In
S. Card, J. Grudin, M. Linton, J. Nielson, & T.
Skelley (series Eds.). Morgan Kaufmann series in
interactive technologies. San Fransisco: Morgan
Kaufmann.
Bonapace, L. (2002). Linking product properties
to pleasure: The sensorial quality assessment

Design Methods for Experience Design

methodSEQUAM. In W. S. Green & P. W. Jordan

(Eds.), Pleasure with products: Beyond usability (pp.
187-217). London: Taylor & Francis.
Brooks, F. P. (1975). The mythical man-month: Essays
in software engineering. London; Reading, MA: Addison-Wesley.

Campbell-Kelly, M., & Aspray, W. (1996). Computer: A history of the information machine. New
York: Basic Books.
Carmichael, A., Newell, A. F., Morgan, M., Dickinson, A., & Mival, O. (2005, April 5-8). Using
theatre and film to represent user requirements.
Paper presented at Include 2005, Royal College
of Art, London. Retrieved from http://www.hhrc.
rca.ac.uk/programmes/include/2005/proceedings/pdf/carmichaelalex.pdf
Cayol, A., & Bonhoure, P. (2004). Prospective
design oriented towards customer pleasure. In
D. McDonagh, P. Hekkert, J. van Erp, & D. Gyi
(Eds.), Design and emotion: The experience of
everyday things (pp. 104-108). London: Taylor
& Francis.
Cooper, A. (1999). The inmates are running the
asylum. Why high-tech products drive us crazy
and how to restore the sanity. Indianapolis, IN:
SAMS.
Cross, N. (2001). Designerly ways of knowing:
Design discipline versus design science. Design
Issues, 17(3), 49-55.
Crossley, L. (2004). Bridging the emotional gap.
In D. McDonagh, P. Hekkert, J. van Erp, & D.
Gyi (Eds.), Design and emotion: The experience
of everyday things (pp. 37-42). London: Taylor
& Francis.
Csikszentmihalyi, M. (1991). Flow: The psychology of optimal experience. New York: Harper
Collins.
Curtis, H. (2002). MTIV: Process, inspiration and
practice for the new media designer. Indianapolis,

IN: New Riders.

De Bono, E. (1995). Serious creativity. Using the
power of lateral thinking to create new ideas.
London: Harper Collins.
DeJean, P. H. (2002). Difficulties and pleasure?
In W. S. Green & P. W. Jordan (Eds.), Pleasure
with products: Beyond usability (pp. 147-150).
London: Taylor and Francis.
DiBlas, N., Paolini, P., Speroni, M., & Capodieci,
A. (2004). Enhancing accessibility for visually
impaired users: The Munch exhibition. In D.
Bearman & J. Trant (Eds.), Museums and the
Web 2004: Selected papers (pp. 89-97). Ontario,
Canada: Archives and Museum Informatics.
Dunne, A. (1999). Hertzian tales: Electronic
products, aesthetic experience and critical design.
London: Royal College of Art.
Fiore, S. (2003). Supporting design for aesthetic
experience. Paper presented at HCI, the Arts and
the Humanities, International HCI Workshop,
King's Manor, York, UK.
Fuller, B. (1969). Utopia or oblivion. New York:
Bantam Books.
Fulton-Suri, J. (2004). Design Expression and
human experience: Evolving design practice. In
D. McDonagh, P. Hekkert, J. van Erp, & J. Gyi
(Eds.), Design and emotion: The experience of
everyday things (pp. 13-17). London: Taylor &
Francis.
Garfinkel, H. (1967). Studies in ethnomethodology. Englewood Cliffs, NJ: Prentice Hall.
Givechi, R., & Velzquez, V. L. (2004). Positive
space. In D. McDonagh, P. Hekkert, J. van Erp,
& D. Gyi (Eds.), Design and emotion: The experience of everyday things (pp. 43-47). London:
Taylor & Francis.
Gould, J. D. (1995). How to design usable systems
(Excerpt). In R. M. Baecker, J. Grudin, W. Buxton,
& S. Greenberg (Eds.), Readings in human-com-

445

Design Methods for Experience Design

puter interaction: Toward the year 2000 (2nd ed.,

pp. 93-121). Morgan Kaufmann.
Gould, J. D., & Lewis, C. (1985). Designing for
usability: Key principles and what designers think.
Communications of the ACM, 28(3), 300-311.
Hackos, J. T., & Redish, J. C. (1998). User and
task analysis for interface design. New York:
John Wiley & Sons.
Hall, R. (2005). The value of visual exploration:
Understanding cultural activities with young
people. UK: The Public.
Hammer, D. K., & Reymen, I. M. M. J. (2004).
The role of emotion in design reflection. In D.
McDonagh, P. Hekkert, J. van Erp, & D. Gyi
(Eds.), Design and emotion: The experience of
everyday things (pp. 421-425). London: Taylor
& Francis.
Harrison, M., & Thimbleby, H. (1990). Formal
methods in human-computer interaction. Cambridge, UK: Cambridge University Press.
Heath, C., & Luff, P. (2000). Technology in action.
Cambridge, UK: Cambridge University Press.
Hochschild, A. R. (1983). The managed heart:
Commercialization of human feeling. London;
Berkeley: University of California Press.
International Organization for Standardization (ISO). (1998). Ergonomic Requirements
for Office Work With Visual Display Terminals
(VDTs)Part 11 Guidance on Usability (9241-11).
Geneva, Switzerland: ISO.
International Organization for Standardization
(ISO). (1999). Human-Centred Design Processes
for Interactive Systems (13407). Geneva, Switzerland: ISO.
International Organization for Standardization
(ISO). (2000). Software EngineeringProduct
QualityPart 1: Quality Model (9126-1). Geneva,
Switzerland: ISO.

446

Jefsioutine, M., & Knight, J. (2004, November 17-21).

Methods for experience design: The experience
design framework. Paper presented at Future
Ground, Design Research Society International
Conference 2004, Melbourne.
Jones, J. C. (1990). Design methods (2nd ed.). London: Design Council, David Fulton Publishing.
Jones, B., Valdez, G., Nowakowski, J., & Rasmussen, C. (1994). Designing learning and technology
for educational reform. Oak Brook, IL: North
Central Regional Educational Laboratory.
Jordan, P. W. (2000). Designing pleasurable
products. London: Taylor & Francis.
Knight, J. (2004, September). Design for life: Ethics, empathy and experience. Paper presented at
Design for Life, British HCI Group Conference,
Leeds, UK.
Knight, J., & Jefsioutine, M. (2002, September).
Relating usability research to design practice.
In M. Maguire & K. Adeboye (Eds.), European
Usability Professionals Association Conference
Proceedings (Vol. 3, pp. 2-7). EUPA Conference.
London: British Computer Society.
Knight, J., & Jefsioutine, M. (2003, October 14).
Attitudes to Web accessibility. Usability News.
Retrieved October 2003, from http://www.usabilitynews.com/news/article1321.asp.
Lee, E. S., Whalen, T., McEwen, S., & Latremouille, S. (1984). Optimizing the design of menu
pages for information retrieval. Ergonomics, 27,
1051-1069.
Liddle, D. (1996). Design of the conceptual model;
An interview with David Liddle. In T. Winograd
(Ed.), Bringing design to software (pp. 17-31).
New York: Addison-Wesley.
Liu, Y. (2003). The aesthetic and the ethic dimensions of human factors and design. Ergonomics,
46(13/14), 1293-1305.

Design Methods for Experience Design

Maguire, M. (2004). Does usability = attractiveness? In D. McDonagh, P. Hekkert, J. van Erp, &
D. Gyi (Eds.), Design and emotion: The experience of everyday things (pp. 303-307). London:
Taylor & Francis.
Maslow, A. (1970). Motivation and personality
(2nd ed.). New York: Harper & Row.
McDonagh-Philp, D., & Lebbon, C. (2000). The
emotional domain in product design. The Design
Journal, 3(1), 31-43.
Molotch, H. (2003). Where stuff comes from:
How toasters, toilets, cars, computers and many
other things come to be as they are. London; New
York: Routledge.
Norman, D. (2004). Emotional design: Why we
love [or hate] everyday things. New York: Basic
Books.
Ortony, A., Clore, G. L., & Collins, A. (1988). The
cognitive structure of emotions. Cambridge; New
York; Melbourne: Cambridge University Press.
Quinn, C. N. (1997). Engaging Learning. Instructional Technology Forum (ITFORUM@UGA.
CC.UGA.EDU), Invited Presenter. Retrieved June
19, 2003, from http://itech1.coe.uga.edu/itforum/
paper18/paper18.html
Reeves, B., & Nass, C. (2002). The media equation:
How people treat computers, television, and new
media like real people and places. Stanford, CA:
Center for the Study of Language and Information Publications.
Rittel, H., & Webber, M. (1973). Dilemmas in a
general theory of planning. Policy Sciences, 4,
155-169. Amsterdam: Elsevier Scientific Publishing Company Inc., 155-169.

Rosson, M. B., & Carroll, J. M. (2002). Usability

engineering: Scenario-based development of
human computer interaction. San Francisco:
Morgan Kaufmann.
Rothstein, P. (2002, September 5-7). a (x 4):
Combining ethnography, scenario-building, and
design to explore user experience. In D. Durling
& J. Shackleton (Eds), Common Ground Proceedings of the Design Research Society International
Conference 2002 (pp. 945-960). Stoke-on-Trent:
Staffordshire University Press.
Shedroff, N. (2001) Experience design 1. Indianapolis, IN: New Riders.
Spillers, F. (2004). Emotion as a cognitive artifact
and the design implications for products that
are perceived as pleasurable. Paper presented at
Design and Emotion 2004, Ankara, Turkey.
Suchman, L. A. (1987). Plans and situated actions:
The problem of human machine communication.
Cambridge, UK: Cambridge University Press.
Tiger, L. (1992). The pursuit of pleasure. Boston:
Little, Brown and Company.
UK Cabinet Office. (2001). A futurists toolbox:
Methodologies in futures work. United Kingdom:
Performance and Innovation Unit, Strategic
Futures Team. Retrieved June 12, 2006, from
http://www.strategy.gov.uk/downloads/files/toolbox.pdf
UK Disability Rights Commission. (2004). The
Web: Access and inclusion for disabled people.
London: TSO.
World Wide Web Consortium (W3C). (1999). Web
Content Accessibility Guidelines 1.0. Retrieved
August 26, 2004, from http://www.w3.org/TR/
WAI-WEBCONTENT/checkpoint-list.html

This work was previously published in Human Computer Interaction Research in Web Design and Evaluation, edited by P.
Zaphiris, pp. 130-147, copyright 2007 by Information Science Publishing (an imprint of IGI Global).

447

448

Chapter 2.6

Integrating Usability, Semiotic,

and Software Engineering into
a Method for Evaluating User
Interfaces
Kenia Sousa
University of Fortaleza, Brazil
Albert Schilling
University of Fortaleza, Brazil
Elizabeth Furtado
University of Fortaleza, Brazil

Abstract

Introduction

We present artifacts and techniques used for user

interface (UI) design and evaluation, performed
by professionals from the human-computer interaction (HCI) area of study, covering usability
engineering and semiotic engineering, which can
assist software engineering (SE) to perform usability tests starting earlier in the process. Tests
of various interaction alternatives, produced
from these artifacts, are useful to verify if these
alternatives are in accordance with users preferences and constraints, and usability patterns, and
can enhance the probability of achieving a more
usable and reliable product.

In a software development process (SDP), it is

crucial for developers, customers, and users to
interact in order to specify, generate, and evaluate the software. From software specification to
its delivery, various kinds of tests must be performed, involving aspects such as: functionality,
portability, performance, and usability. This work
focuses on the context of usability, communicability, and functionality tests (e.g., appropriateness
of a chosen interface design alternative to user
preferences, consistency to a visual pattern, efficient execution of interactive tasks on interface
objects, etc.).

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Integrating Usability, Semiotic, and Software Engineering into a Method for Evaluating User Interfaces

Through our researches on tests in HCI and SE,

and through our experiments on their integration
in a SDP, we verified that HCI concepts facilitate
the UI evaluation work performed by the test team
of an interactive system under development. More
specifically, by means of UI generation based
on HCI models (e.g., task model), it is possible
to evaluate the UI earlier (e.g., its functionality),
independent of having the entire noninteractive
specification ready. Prototypes, for instance,
can represent UI design alternatives that may be
tested early by HCI experts to verify if they are
in accordance with user preferences, usability
patterns, and so on.
This work presents a SDP to design and
evaluate UIs, based on the integration of concepts,
models, and activities of usability, semiotic, and
software engineering.
This chapter is structured as follows: The
User-Interface Evaluation section shows the
contribution of each engineering area to UI
evaluation; The Process section describes the
UI design process; The Evaluation Strategy
section describes the UI evaluation process, showing which concepts are used to perform tests and
when they are performed; the Case Study section
describes the case study in which we designed
and evaluated UIs for the Brazilian System for
the Digital Television (SBTVD); and, finally, the
Findings and Future Works section describes
findings and future works, and the Conclusion
section concludes this work.

User-Interface Evaluation
In this section, we present concepts and evaluation
techniques from usability engineering, software
engineering, and semiotic engineering.

Usability Engineering
Usability engineering is a set of activities that
ideally take place throughout the lifecycle of

the product, with significant activity at the early

stages even before the UI has been designed. The
need to have multiple usability engineering stages
supplementing each other was recognized early in
the field, though not always followed in development projects (Gould & Lewis, 1985).
In usability engineering, techniques and methods are defined aiming to assure a high usability
level of the interactive UIs. Among them, we
emphasize the application of ergonomic criteria
in the UI design. Verification of these criteria
in designed UIs is called heuristic evaluation,
performed by usability experts without user participation. Evaluators examine the IS searching
for problems that violate general principles of
good UI design, diagnosing problems, obstacles
or barriers that users will probably encounter
during their interaction. In addition, methods to
capture usability requirements attend to user preferences, restrictions, and use-context. A usability
requirement can be derived from an interaction
restriction; such as if part of the system needs to
be implemented for palm-top devices.
The evaluation approaches from usability
engineering suggests a structured sequence of
evaluations based on usability inspections
methods and on usability tests.
Some inspection methods are: (1) heuristic
evaluation, verification of usability heuristics
(Nielsen, 1993); (2) review of guidelines, verification if the UI is according to a list of usability
guidelines (Baranauskas & Rocha, 2003); (3)
consistency inspection, verification of the consistency among the UIs related to terminology,
color, layout, input and output format, and so
on; and (4) cognitive walkthrough, simulation
of the user walking through the UI to execute
typical tasks.
Some usability test methods are: (1) thinking
out loud, we request the user to verbalize everything he or she thinks while using the system, and
we expect that their thoughts demonstrate how the
user interprets each UI item (Lewis, 1982); and
(2) performance measures, quantification of some
evaluated items to make future comparisons.
449

Integrating Usability, Semiotic, and Software Engineering into a Method for Evaluating User Interfaces

Software Engineering

Semiotic Engineering

Software engineering is composed of technologies

and practices that are used in the development of
software products, enhancing software productivity and quality, by providing a more systematic
and controlled SDP (Sommerville, 2001).
In software engineering, there are various
types of tests to be performed in each test stage,
such as: usability tests to ensure that access and
navigation through functionalities are appropriate
for users; UI tests, to ensure a good functionality
of the UI components and verify conformity to
corporate patterns; and functionality tests, which
are responsible for verifying if the generated
software achieves all the proposed functionalities
according to the customers requests. Test cases
(Myers, 2004) are normally generated and comprise procedures to be followed in test activities
in order to deal with all possible situations when
using the software, including basic flows, as well
as error treatment and invalid data verification.
According to Pressman (1995), the main goal of
test cases is to derive a set of tests that will probably reveal errors in the software. To achieve this
goal, software engineering basically proposes two
test categories: white-box and black-box tests.
For Pressman (1995), the white=box test must
verify the internal part of the product, tests can
be performed to guarantee that the components
are integrated, and the internal operation achieves
the performance level as specified in the requirements.
Functional tests, or black box tests, represent
a test approach in which tests are derived from
specifications of the system. In this kind of test,
the evaluator is concerned with the functionality,
not with the software implementation (Sommerville, 2001).
In these two test categories, software engineering defines four main types of tests: unit tests
that are generally white-box tests; acceptance
and regression tests that are usually black-box
tests, and integration tests that blend this two
categories.

Semiotic engineering is an HCI theory that emphasizes aspects related to the metacommunication
designer user(s) via user-system communication,
which passes through the UIs of interactive
applications. The system is considered to be
the deputy or a representative of the system
designer (Souza, Barbosa, & Silva, 2001). The
content of messages is the application usability
model. Its expression is formed by the set of all
interaction messages sent through the UI during
the interaction process. The user plays a double
role: interacting with the system and interpreting
messages sent by the designer.
Semiotic engineering is essentially involved in
test procedures with final users (empiric evaluation), aiming at system communicability analysis
based on qualitative evaluation, in which
there are four phases: test preparation; labeling;
interpretation, and formatting and elaboration
of the semiotic profile of the application to be
evaluated. The techniques used in theses phases
are: system-user observations, questionnaires,
(somative-evaluation) inspections, interviews,
filming,and so on.
Semiotic engineering is essentially present in
tests with final users (e.g., empiric evaluation),
aiming at analyzing the system communicability.
A UI has a good level of communicability when it
is able to successfully transmit the designer message to the user, allowing him or her to understand
the system goal, the advantages of using it, how it
works, and the basic UI interaction principles.

450

Evaluation
After studying about evaluation techniques, artifacts and approaches from software, usability, and
semiotic engineering we are able to conclude that
an evaluation process can be seen under various
perspectives.
Concerning software engineering, we noticed
the importance of software quality concerning

Integrating Usability, Semiotic, and Software Engineering into a Method for Evaluating User Interfaces

functionality, performance, portability, and other

nonfunctional requirements. Its artifacts and
techniques include the evaluation of these aspects
in an objective manner.
Usability engineering focuses in providing
more ease of use, ease of learning, and efficiency
to interactive systems.
Semiotic engineering includes procedures that
allow the evaluation of the quality of the interactivity of systems by observing the communication
through messages of the user to the system.
Based on these perspectives, we believe that
an approach for UI evaluation of interactive
systems that integrates these approaches is able
to guarantee a system with quality concerning
functionality, usability, and interactivity, derived
from software, usability, and semiotic engineering, respectively.
Next, we will describe a lightweight development process for interactive systems, called UPi
(Sousa & Furtado, 2004), which integrates HCI
and SE activities, artifacts, and professionals.

The Process
UPi can serve as a guide, providing useful steps
and artifacts that can be tailored and customized
when organizations intend to develop usable
interactive systems. One of the best advantages
of UPi is the idea to focus on activities, artifacts
and guidelines that add value to the UI generation.
With this approach, it can be integrated with any
other process and inherit activities that are vital
to the entire process, but that are better defined
and solidified in other processes. For instance,
project management, configuration and change
management, implementation, and deployment
activities are very well detailed in the RUP
(Kruchten, Ahlqvist, & Bylund, 2001). Besides
the RUP, UPi can also be applied in conjunction
with ISO 13407 (ISO 13407, 1999), which already
has other activities defined and validated, such as
project planning, testing, and so on.

UPi is composed of activities that aim at designing UIs. These activities are based on RUP
activities, but they follow different guidelines that
take into consideration usability aspects.
In this work, we are integrating UPi with
UPi-Test (to be presented in the next section) in
order to guide professionals that are developing
interactive systems to evaluate them throughout
the entire development process.

Phase I: Inception
The main goal in this phase is to elicit requirements from users in order to develop an interactive system that best suits their needs through the
execution of some activities (presented as follows).
These requirements are documented through
certain artifacts: use-case models, task models,
usability requirements, and paper sketches.
Use-case models represent a well-established
manner to define the system functionality, while
task models can be used to detail use cases by
breaking them down into tasks. Usability requirements represent users preferences or constraints
that can be part of a usable interactive system.
Paper sketches focus on the interaction, UI
components, and on the overall system structure,
keeping the style guide secondary, without being
too abstract.
The purpose of the Elicit Stakeholder Needs
activity is to understand users, their personal
characteristics, and information on the environment where they are located that have a direct
influence on the system definition, and to collect special nonfunctional requirements that the
system must fulfill, such as performance, cost,
and device requests.
The purpose of the Find Actors and Use Cases
and Structure the Use-case Model activities is
to define the actors (users or other systems) that
will interact with the system and the functionality of the system that directly attend to users
needs and support the execution of their work
productively.

451

Integrating Usability, Semiotic, and Software Engineering into a Method for Evaluating User Interfaces

The purpose of the Detail a Use Case activity

is to describe the use cases tasks using the task
model, to describe any usability requirements
related to the use case, to define the system
navigation based on the task model hierarchical
structure, and to create paper sketches.
The purpose of the Review Requirements
activity is to verify, with usability experts, if the
paper sketches are in accordance to the task model
and validate, with users, if the requirements are in
conformance with their needs by showing them
the elaborated paper sketches.

Phase II: Elaboration

The main goal in this phase is to transform the
requirements in a representation that can be
understood by UI designers and programmers.
These representations are provided by the following artifacts: system architecture, UI Definition
Plan, and drawing prototypes.
System Architecture is composed of smaller
components that represent the main functionality
of the entire system. The UI Definition Plan is
a new artifact that aims to define which visual
objects should be part of the UI. Drawing prototypes produce an accurate image of the system
and they are useful to demonstrate patterns and
style guides.
The purpose of the Define and Refine the
Architecture activity is to (re)design the classes
that represent the data that are handled by users
while performing certain tasks.
The purpose of the Define UI Plan activity is
to define which visual objects and which usability
patterns can be part of the UI according to the
nonfunctional requirements defined in the Elicit
Stakeholder Needs activity.
The purpose of the UI Prototyping activity is
to design a UI prototype in drawings following the
description specified in the task models, in the UI
definition plan and in the system architecture.

452

The purpose of the Evaluate Prototype activity

is to verify if the UI prototypes are in accordance
to usability principles and to validate with users
if the UI prototypes are in conformance with
their needs.

Phase III: Construction

The main goal of this phase is to implement and
verify the accuracy of the components implemented and of the UI designed.
The purpose of the Implement Components
activity is to develop the classes previously designed and implement the UI prototyped.
The purpose of the Evaluate the Version of the
System activity is to verify if the functionality
of the interactive system is in accordance with
users requirements.

Phase IV: Transition

The main goal of this phase is to deliver to the
customer a system with high level of quality and
usability.
The purpose of the Deploy the System activity is
to make the system available for the customer.
The purpose of the Evaluate the System activity
is to validate with users (by using the system in
the deployment site) if the system conforms with
their view of the system.
All of these phases are supported by processes
concerning configuration and change management and project management, such as the RUP.
This support is provided for the Manage Change
Requests activity, which aims at evaluating the
impact of change requests, deciding if they are
to be included in the current iteration, and, if
they are accepted, manage the changes in the
appropriate artifacts.
Concerning the evaluation activities performed in each phase, they will be more thoroughly
explained in the next section.

Integrating Usability, Semiotic, and Software Engineering into a Method for Evaluating User Interfaces

The Evaluation Strategy

Talk with Users and/or Customers

The unified process for evaluating interactive

systems UPi-Test (Schilling et al., 2005) has the
same phases as the RUP (inception, elaboration,
construction, and transition) (Kruchten, Ahlqvist,
& Bylund, 2001). This process is based on the
Unified Process for Interactive Systems, called
UPi (Sousa & Furtado, 2004) and follows the
approach to design UI prototypes dependent on
the device (Coyette, Faulkner, Kolp, Limbourg,
& Vanderdonckt, 2004).
Each phase is directly related to a specific
area. This way, usability engineering supports
the verification and validation in the inception
and elaboration phases, software engineering
supports verification in the construction phase,
and semiotic engineering supports validation in
the transition phase.
Figure 1 illustrates these four phases in the
UPi-Test, each one with its flow of activities,
artifacts, and techniques.
UPi-Test includes the verification and validation of usability and functionality of interactive
systems UIs. Nonfunctional requirements are not
in the scope of this work, such as aspects related to
database (e.g., connection, integrity, etc.), security,
and architecture. These aspects can be supported
in future versions of this process.

This activity consists of the first contact with users

or customers, in which system analysts understand
their profiles, objectives, and the scenario where
they are included. In this activity, it is necessary
to use a technique to elicit requirements.
We propose an initial informal talk in order to
better understand users environment. Then, we
suggest the performance of interviews, with the
use of a questionnaire that aids in the identification of users and/or customers and their intended
goals, preferences, and possible constraints.

Phase I: Inception
The inception phase is important in guaranteeing
that the following phases achieve results to attend
to users and customers usability goals. This phase
has the constant participation of users in order
to understand their requests, which are verified
and validated according to usability engineering
to allow the development of these requests. The
description of the activities and artifacts in this
phase is presented as follows.

Obtain Users and Customers Preferences and Constraints

In this activity, system analysts aim at eliciting
users and customers preferences and constraints
in order to design UIs that attend their needs and
also to help usability experts during evaluation
and to help UI designers during prototyping.
Some examples of preferences are: design colors,
font styles, navigation schemes. Some examples
of constraints are: technology constraints, such
as platform, device, and so on.

Consider HCI Models and Use Cases

Use cases represent an artifact from software
engineering that identifies the functionality of
interactive systems; users interactions, expressed
through users tasks; and the systems responses
to perform these interactions (systems tasks).
Task models detail a use case or a group of
related use cases by specifying users tasks and
systems tasks. It is useful to support UI designers in the elaboration of prototypes because it is
easy to identify the necessary views and objects
in the prototype from the task model hierarchical structure.

453

Integrating Usability, Semiotic, and Software Engineering into a Method for Evaluating User Interfaces

54 Sousa, Schilling, & Furtado

1. Evaluation
process of UIs
Figure 1.Figure
Evaluation
process of UIs

454

Integrating Usability, Semiotic, and Software Engineering into a Method for Evaluating User Interfaces

Analyze Existing Systems

In this moment, it is important to analyze existing systems. This involves a comparative study
of similar systems. These systems can be used
as a reference in order for system analysts to
propose new functionality or to choose design
patterns, which are all useful for the design of
UI prototypes.

Consider Paper Sketches

After talking with users and customers, the UI
designer designs paper sketches. This prototype
is verified and validated, as presented in the next
activity.

Verify with Expert

This activity consists of the verification of paper
sketches by the usability expert. The expert is
concerned with verifying if users goals, in terms
of functionality, were included in the prototype,
as well as if usability principles were used.
We propose that experts use the heuristic evaluation approach (Nielsen, 1993) as an inspection
method. The usability heuristics will guide the
usability expert in the process of verifying UIs
quality of use. We also suggest the use of the task
model to verify if all the specified functionality
was designed in the prototype.
When the expert notices that a functional
requirement or any usability principle is missing, change requests can be made, which leads
to changes in the models and new proposals of
prototypes.

Validate with Users and Customers

After the verification with the expert, we propose
a validation with users and customers so they can
approve the generated artifacts. If the prototypes
do not attend users needs, change requests can
be made, which leads to changes in the models

(if new functionality is requested) and new proposals of prototypes (if changes in the navigation
are requested). This process is repeated until the
generated prototype attends users preferences
and needs.
This activity early in the process provides
flexibility for users and customers to evaluate the
evolution of the system, therefore, designers and
users feel more confident with the UI design.
After the conclusion of the inception phase,
the resulting artifacts are verified and validated
as paper sketches.

Phase II: Elaboration

The elaboration phase is concerned with designing
and evaluating drawing prototypes. In this phase,
we use verification and validation techniques,
such as heuristic evaluation and validations with
users. After this phase, the resulting artifacts
are drawing prototypes validated according to
usability requirements and patterns. The description of the activities and artifacts in this phase is
presented as follows.

Use Design Patterns

In order to guarantee the quality of the product
and efficiency of the project, we suggest the use of
design patterns for graphical UIs. These patterns
will guarantee that we elaborate and develop UIs
following already verified and validated parameters, which can be incremented by the reports
generated in the end of the transition phase.

Consider Drawing Prototypes

UI designers are responsible for designing drawing prototypes based on paper sketches previously
validated, and on usability patterns. These prototypes are verified and validated by the following
two activities.

455

Integrating Usability, Semiotic, and Software Engineering into a Method for Evaluating User Interfaces

Verify with Expert

Implement and Integrate

This activity consists of verifying the usability

of drawing prototypes by experts. We propose
that experts use the heuristic evaluation approach
(Nielsen, 1993) as an inspection method to verify
whether or not certain usability principles are
present in the prototype.
When the expert notices that any usability
principle is missing, change requests can be made,
which leads to new proposals of prototypes. At
this moment, it is not necessary to make changes
in the models because this approach evaluates the
quality of use, not functionality aspects, which
were evaluated in the previous phase.

These activities consist of developing the UI and

integrating it with the application. The integrated
product, either a prototype or the final system, can
be useful for evaluating the navigation, interactivity, and functionality aspects.

Validate with Users and Customers

After verification with the expert, a validation is
proposed to users and customers so they can approve the generated prototypes. At this moment,
users and customers evaluate the used usability
patterns and the style guide. This process is
repeated until the generated prototype attends
users preferences and constraints. If the prototypes do not attend users needs, change requests
can be made, which leads to new proposals of
prototypes.
After the conclusion of the elaboration phase,
the resulting artifacts are verified and validated
drawing prototypes, which support development,
tests, and deployment activities.

Phase III: Construction

In this phase, the UI is developed and the application is integrated with it. Considering software
engineering, we propose functionality tests of an
executable prototype (i.e., a product with some
functionality) or the final product (i.e., a product
with all the functionality), using functional test
cases.
The activities and artifacts in this phase are
presented as follows.

456

Elaborate Test Cases

Test cases can be elaborated starting in the inception phase, using paper sketches, they can then be
updated in the elaboration phase, using drawing
prototypes, and finished in this activity. This
artifact focuses on the system functionality, not
on nonfunctional aspects.
The technique used to define test cases includes
the following topics: association to a use case,
specification of the item to be tested, preconditions
to execute before testing, identification of valid
and invalid inputs, and the expected outputs. The
actual outputs are compared with the expected
outputs described in the test cases and this comparison is used as validation of the use case.

Verify with Expert

This activity consists of the verification of the
functionality of the product by usability experts
and developers. Examples of aspects that are verified are: consistency of the outputs, navigation,
existence of error messages, results after clicking
on objects, as well as other aspects identified in
the test cases.
After this verification, developers and experts
can generate change requests to correct the errors;
which leads to the repetition of the implementation and integration activities.
After the conclusion of the construction phase,
the resulting artifact is the integrated product,
which is tested with consideration to usability
aspects in the transition phase.

Integrating Usability, Semiotic, and Software Engineering into a Method for Evaluating User Interfaces

Phase IV: Transition

This phase comprehends, in general terms, the
preparation of the test environment, which can
be a test laboratory or the environment where the
system is used. With consideration to semiotic
engineering, we use some validation techniques
and artifacts. The description of the activities and
artifacts in this phase is presented as follows.

Prepare the Environment

To prepare the environment, we suggest the installation of the system, software for capturing
the systems use, and equipment, such as video
cameras and necessary hardware devices. We
also suggest the creation of questionnaires and
checklists.
The test laboratory must be similar to the
real user environment with consideration to
physical structure, climate, sound aspects, and
equipment in order to allow users to live the
same conditions of the real environment. There
should be a room where the test takes place and
another one for observation.

Install the Product

In this activity, the product is installed, either a
partial version or the final version of the system,
in order to allow users to use the system in their
real environment. This installation allows the
tests to be performed.

Usability Tests with Users

This evaluation is performed with users. In it,
evaluation techniques, proposed by semiotic engineering, are used, such as: recording, observation,
questionnaires, and so on.
Before starting the tests, the usability expert
talks with the user in order to: clarify that the
system is under evaluation, not him/her; present
the scenario used for the test; and make the user

feel comfortable; which are aspects that influence

the final results.
Observers that are in the observation room
should fill out the questionnaires and checklists.
This activity can be divided in two moments,
the first one, when the user interacts with the
system to perform a task of his/her own interest;
the second one, when the expert requires the user
to perform a specific task.
In this activity, navigability, interactivity, and
acceptability will be evaluated.

Use Observations/Checklist/
Questionnaires
Observations, questionnaires, and checklists are
artifacts and techniques proposed by the semiotic
engineering in order to verify the user-system
interactivity and communicability. Experts and
observers will use these artifacts during the
tests, which result in the definition of the quality
of the interactive system. These results can lead
to change requests for developers to correct the
detected mistakes.
Users comments and the actual execution of
the tests will be recorded to help in the analysis
of the results of the questionnaires and of users
observations.

Make Corrections
In this activity, developers make corrections proposed by experts after the tests. After the changes
are made, users validate the product.

Deliver the Evaluated Product

As a result of the process, we have the results of
evaluations, which are useful for future versions;
and we also have a verified and evaluated product
according to a set of techniques proposed by usability, software, and semiotic engineering.
If the product is a final version of the system,

457

Integrating Usability, Semiotic, and Software Engineering into a Method for Evaluating User Interfaces

it is ready to be delivered for use. If it is a partial

version (e.g., executable prototype), the professionals need to perform the activities in the construction phase, then in the transition phase, until the
product reaches its final version.

Elaborate Reports
The results obtained will be used as a basis for
the elaboration of evaluation reports, which
propose adaptations in the used patterns and in
the creation of new patterns that can be used in
future iterations.

Case Study
In this chapter, we describe the case study of this
research work, which is concerned with the evaluation of UIs for the SBTVD project, focusing on
the applications: electronic portal, insertion of
texts, and help.

Introduction
The digital TV represents digital and social inclusion for a great part of the Brazilian population,
especially for people less privileged, who do not
have access to computers, and therefore, cannot
access the Internet.
The SBTVD must be adapted to the socioeconomic conditions of the country, as well as allow
the use of conventional TV sets already in large
use in the country in order to decrease risks and
costs for the society.
The digital TV creates various possibilities of
interaction between the TV and the user, such as:
exchange of text or voice messages, virtual chats,
searches for a favorite show, access to information about the government, and so on. These
possibilities are different from the characteristics
of the conventional TV, in which the user plays
a passive role.

458

In the following section, we describe the

performance of the activities proposed by UPi
and UPi-Test. It is important to point out that the
SBTVD is still under development. That is the
reason why we cannot demonstrate all the activities of the process.

Phase I: Inception
Talk with Users and/or Customers
This activity was difficult to perform in this project
because there is an almost unlimited number of
users and/or customers. Fortunately, there were
specific and clear specifications, established by the
Brazilian government. Such specifications were
used as requests from users and customers.
To define the scope of the application under
our responsibility (access portal), we had meetings with representatives of the government and
with other institutions that are participating in
the project. These meetings were supported with
brainstorming and the resulting decisions were
analyzed and were shared with all the institutions
through e-mails and discussion lists.
After these meetings, we decided that the portal
will consist of a main application that allows the access to all other applications in the SBTVD, which
can be: electronic mail, electronic commerce, EPG,
help, electronic government, and so on.

Obtain Users and Customers Preferences and Constraints

The SBTVD project is concerned with various users profiles, including the ones who are and those
who are not used to technology, but not including
the ones with any kind of disabilities.
In order to identify their possible preferences
and constraints, we studied existing systems; we
had many meetings and workshops. The opinions
of all the participants in the project were taken into
consideration because we can also be considered
to be potential users.

Integrating Usability, Semiotic, and Software Engineering into a Method for Evaluating User Interfaces

Our main goal in the usability workshops

was to choose the best usability pattern for each
requirement based on the evaluation of positive
and negative aspects of each proposed usability
pattern, as specified in the UPi activity define
UI plan.
These proposed usability patterns were selected from a list, such as the one available in
Welie (2005), which are organized in the following
format: problem to be solved, solution, context in
which it can be used, and graphical illustration.
For the participants to evaluate the usability
patterns, we provided a set of guidelines. After
they read the guidelines, each group evaluated
the positive and negative aspects of each usability pattern suggested for the requirement. Then,
the participants discussed and reached the final
decision as to what was the best usability pattern
for the requirement under discussion.
The personalization group considered the
guidelines as they evaluated the positive and
negative implications of each usability pattern,
aiming at achieving Nielsens usability goals. For
demonstration purposes, we focus on presenting
the personalization of colors.
As a result of the evaluation, the personalization group decided to use color templates because
of the greater impact of positive implications over
the negative ones. The result of this work was the
generation of a document associating usability
patterns with their positive and negative implications. For instance, the selected pattern with more
positive implications was Provide predefined
color templates to change font/background colors
instead of the pattern Offer a list of colors from
where the font/background colors can be chosen
because of the following positive implications: efficiency of use, compatibility of the system with the
real world, and clarity of information. The second
option had more negative implications than the
first one, such as difficulty in use by beginners
and the constant need to resize the space reserved
to the presentation of information.

We defined that the applications in the SBTVD

need to have a high level of usability. There was
a list of technology constraints, especially the
ones related to memory and processing capacity.
Besides that, graphical representations need to
be simple because conventional TV sets do not
support images as computer monitors do, and the
TV luminosity is very different from the one in
monitors; consequently, colors and texts appear
differently.

Consider HCI Models and Use Cases

In this activity, analysts and usability experts
elaborated use case and task models, and then
changed them, when change requests were made
after the evaluation of users in the end of the Inception phase. The requests resulted in changes
in the models because they were changes in the
functionality, such as: do not consider access to
multiple applications simultaneously (technical
constraint) and include personalized help. These
changes reflected in the models and will be considered in the prototypes.

Analyze Existing Systems

In this activity, we researched on the Internet in
order to find digital TV systems already in use. It
was very difficult to find them because there are
few systems available for access in the Internet.
This analysis was used to identify some UI
design patterns for the digital TV, such as: upper
and bottom options bar, menu on the left, navigation icons, interaction buttons, and so on.

Consider Paper Sketches

After the analysis of existing systems, UI designers elaborated paper sketches (Figure 2). However,
they needed to be redesigned after the evaluation
of users in the end of the inception phase, which
resulted in change requests.

459

Integrating Usability, Semiotic, and Software Engineering into a Method for Evaluating User Interfaces

Figure 2. Paper sketch: Personalization

Portal

Watch TV

Color

Font

Help

Validate with Users and Customers

In meetings with the institutions participants of
the project and with representatives of the Government users evaluated the elaborated prototypes.
Various change requests, as well as users preferences were identified. These requests resulted
in updates in the models and in redesign of the
paper sketches.
As examples of preferences, we point out the
following: Include the TV and Portal in the
remote control; the bottom options bar should
overpass the application; and make available a set
of options for inserting text, such as the one used
in computer keyboards, the one used in mobile
phones, and in alphabetical order.

Phase II: Elaboration

Use Design Patterns
Some change requests resulted in the following
changes in the prototypes: transfer the bar from the
top to the bottom of the screen, include the TV
button in the bar when accessing an application or
in the portal; include the Portal button in the bar
when accessing an application, take off activated
icons (because of technical restrictions), and give
a preview of the requested options of personalization. However, other requests resulted in changes
in the models as well as in the prototypes, such
as the ones mentioned in the activity Consider
HCI models and use cases.

Verify with Expert

In this activity, we performed heuristic evaluations with three experts who considered the
usability of the paper sketches. They observed
aspects, such as: navigation between views and
if usability principles were applied. Besides that,
the experts verified if all users goals, in terms of
functionality (in the task models), were included
in the prototypes.

460

In this activity, the UI design patterns, identified

while analyzing existing systems, were evaluated
and the UI designers started to design drawing
prototypes using the design patterns.

Consider Drawing Prototypes

The UI designers used image editors, such as
Photoshop and Corel Draw, in order to design
the drawing prototypes (Figure 3), which followed
Figure 3. Drawing prototype: Personalization

Integrating Usability, Semiotic, and Software Engineering into a Method for Evaluating User Interfaces

the selected patterns, the updated task models, and

paper sketches. In this prototype, the small view
on the right shows the preview of the required
personalization about the color of the screen
before applying it.

Verify with Expert

The drawing prototypes were evaluated by experts
(Figure 4), who observed various aspects, such
as: layout, colors and fonts, UI design patterns,
and usability principles applied.

Validate with Users and Customers

This activity is yet to be fully performed in this
project, currently; we have finished the document
that specifies the SBTVD graphical UI, which is
going to be evaluated by representatives of the
Brazilian Government.
We scheduled a workshop with the other institutions participants of the project to evaluate the
drawing prototypes (Figure 5). They requested us
to elaborate other alternatives of UIs (including
association of options in the UI with options on
the remote control), increase the size of the options and fonts in the upper bar, and change the
way to differentiate the selected option.

Figure 4. Verification with expert

Phase III: Construction

Implement and Integrate
The implementation of the portal has not started
yet. However, the developers have been studying
and implementing simple applications for the
digital TV, such as three possibilities of insertion of text (the one used in computer keyboards
(qwert), the one used in mobile phones, and in
alphabetical order).

Elaborate Test Cases

The elaboration of functional test cases has
started since the inception phase, when use cases
and paper sketches were elaborated. Functional
requirements were selected and the associated
test cases were elaborated: structure applications,
access applications, access help and personalization (Table 1).

Verify with Expert

In this activity, the programmers and the experts
verify the applications using the test cases for
guidance. Some of these evaluations were done
in the small applications developed for insertion
of text.

Figure 5. Validation with users

461

Integrating Usability, Semiotic, and Software Engineering into a Method for Evaluating User Interfaces

Table 1. Test case: Personalization

Findings and Future Works

Test case

Personalization

Test items

Changes in font and background color

Changes in font size

Pre-conditions

The user must be accessing the portal

Inputs

The user must select a type of personalization: Font color The user selects green

Expected results

The portal must present the personalized

content with a green background color

Phase IV: Transition

After the tests, we evaluated the checklists and

generated reports with solutions to the problems
encountered during the tests. These reports contain comments about the icons, the navigation,
and the help module of the portal application.

No user realized the possibility of navigation

through the numbers associated to each icon,
which represents an interactive application
(see Figure 6).
When they were told about this, they used
it and said it was very practical.
For them, the icons were not a natural representation, but they did not want to give
any opinion about possible solutions.
Some users did not notice the application
that was selected. They said it was better to
have a big square around the icon instead of
the current selection.
When the users were confused and wanted
some help, they did not look for it in the
portal (the blue option in the bottom bar
that activates the help module). Some users
activated the menu button of the remote
control and others waited for the evaluator
to tell them what was necessary to do to
obtain the help.

The usability experts and the developers have

prepared the test environment. For that, they have:
prepared the physical structure (e.g., TV, video

camera, couch, computer to simulate the set-top
box, etc.) and installed the necessary software
(i.e., software to capture the user interaction with

the TV, Linux Operating System, Java Virtual
Machine, and applications to be tested).
A group of three usability experts were
responsible for performing the following activi
ties before the tests started: First, they defined
a questionnaire to apply with users in order to
understand their characteristics and familiarity
to the DTV technology. Second, they selected appropriate metrics (e.g., number of errors, number
of access to the help, etc). Third, they created a
checklist based on Nielsens usability goals and on
the metrics from the previous step. Fourth,
they Schilling, & Furtado
68 Sousa,
prepared the environment with a DTV, a couch,
and a center table in order to make users
feel 6.
at The portal application
Figure
Figure 6. The portal application
home. Fifth, they selected ten users with different
profiles between the ages of 18 and 26.
When the users arrived, each one at a time
was taken to the DTV room, where a usability
expert explained the goals of the test; applied the
questionnaire; and defined a specific goal to be
achieved while interacting with the DTV. While
the user interacted with the application, usability
experts filled out the checklist in the visualization
room, where we monitored the environment and
the user with a camera and captured the interaction
with the DTV using specific software.

462

No user realized the possibility of navigation through the numbers associated to

each icon, which represents an interactive application (see Figure 6).

Integrating Usability, Semiotic, and Software Engineering into a Method for Evaluating User Interfaces

As the bottom bar is used as a design pattern

to put the navigation options, and as each
option is different depending on the situation
(the application being executed), the users
got very confused. We realized they looked
at this bar only once, and then they did not
look at it any more. After memorizing the
options, they wanted to use them all the
time, but the color was different, so they
made many navigation errors.

From the execution of these tests, we were able

to quantify the results: 85% of the users found
the system easy to use, easy to read, and easy to
navigate, but on the other hand, 50% had difficulties in identifying that an option was selected and
in getting out of an unexpected situation.
Our next step is to make the necessary changes
that are related to improving the layout, color
contrast, and icon selection. In addition, the next
tests will be done with the elderly in order to
investigate people who are not used to interaction devices.

Conclusion
In this chapter, we present a new method, which
focuses on integrating different techniques of
usability, semiotic and software engineering. The
aims are to design usable UIs following a modelbased UI design process and to facilitate the test
process by using the evaluated HCI models. In this
manner, we hope to contribute to the development
of interactive systems that are easy for users to
learn and use, and to help testers in performing
their usability tests in an efficient manner.
As main contributions, we focus on evaluating the usability of UIs with the constant participation of users and customers. Besides that,
the integration of various approaches results in
positive outcomes for the prototypes, as well as
for multidisciplinary team members, who are
better integrated and can have their knowledge

enhanced, since they are continuously exchanging

information and experiences.

References
Baranauskas, M. C. C., & Rocha, H. V. da. (2003).
Design e Avaliao de Interfaces Humano-Computador, NIEDNcleo de Informtica Aplicada
Educao, UNICAMPUniversidade Estadual
de Campinas.
Coyette, A., Faulkner, S., Kolp, M., Limbourg,
Q., & Vanderdonckt, J. (2004). SketchiXML:
Towards a multi-agent design tool for sketching
user interfaces based on UsiXML. In P. Palanque,
P. Slavik, & M. Winckler (Eds.), 3rd Int. Workshop
on Task Models and Diagrams for User Interface
Design (pp. 75-82). New York: ACM Press.
Gould, J. D., & Lewis, C. (1985). Designing for
usability: Key principles and what designers think.
Communications of the ACM, 28(3), 300-311.
ISO 13407. (1999). Human-centred design processes for interactive system teams.
Kruchten, P., Ahlqvist, S., & Bylund, S. (2001).
User interface design in the rational unified process. In M. Van Harmelen (Ed.), Object modeling
and user interface design (pp. 45-56). New York:
Addison-Wesley.
Lewis, C. (1982). Using the thinking-aloud
method in cognitive interface design (IBM
Research Rep. No. RC9265, #40713). Yorktown
Heights, NY: IBM Thomas J. Watson Research
Center.
Myers, G. J. (2004). The art of software testing.
New York: John Wiley & Sons.
Nielsen, J. (1993). Usability engineering. Boston:
Academic Press.
Pressman, R. S. (1995). Engenharia de Software.
So Paulo: Makron Books.

463

Integrating Usability, Semiotic, and Software Engineering into a Method for Evaluating User Interfaces

Schilling, A., Madeira, K., Donegan, P., Sousa, K.,

Furtado, E., & Furtado, V. (2005). An integrated
method for designing user interfaces based on
tests. In Proceedings of the ICSE 2005 Workshop
on Advances in Model-Based Software Testing,
(pp. 27-31). St. Louis, MO: ACM.

evaluating end-users programming environments.

Interacting with Computers, 13(4), 467-495.

Sommerville, I. (2001). Software engineering.

New York: Addison-Wesley.

Sousa, K. S., & Furtado, E. (2004). UPi A

unified process for designing multiple UIs. In
IEEE Computer Society, Proceedings of the
International Conference on Software Engineering, Edinburgh, Scotland (pp. 46-53). Edinburgh,
Scotland: ACM.

Souza, C. S. de, Barbosa, S. D. J., & Silva, S. R.

P da. (2001). Semiotic engineering principles for

Welie, M. van (2005). Retrieved September, 15,

2005, from http://www.welie.com

This work was previously published in Verification, Validation and Testing in Software Engineering, edited by A. Dasso, pp.
55-81, copyright 2007 by IGI Publishing, formerly known as Idea Group Publishing (an imprint of IGI Global).

464

465

Chapter 2.7

Guidance in the Interface:

Effects of Externalizing Information

During Problem Solving
Christof van Nimwegen
Utrecht University, The Netherlands
Hermina Tabachneck-Schijf
Utrecht University, The Netherlands
Herre van Oostendorp
Utrecht University, The Netherlands

Abstract
How can we design technology that suits human cognitive needs? In this chapter, we review
research on the effects of externalizing information on the interface versus requiring people to
internalize it. We discuss the advantages and
disadvantages of externalizing information.
Further, we discuss some of our own research
investigating how externalizing or not externalizing information in program interfaces influences problem-solving performance. In general,
externalization provides information relevant to
immediate task execution visibly or audibly in the
interface. Thus, remembering certain task-related
knowledge becomes unnecessary, which relieves
working memory. Examples are visual feedback
aids such as graying out nonapplicable menu

items. On the contrary, when certain needed

task-related information is not externalized on
the interface, it needs to be internalized, stored
in working memory and long-term memory. In
many task situations, having the user acquire
more knowledge of the structure of the task or
its underlying rules is desirable. We examined
the hypothesis that while externalization will
yield better performance during initial learning,
internalization will yield a better performance
later. We furthermore expected internalization to
result in better knowledge, and expected it to provoke less trial-and-error behavior. We conducted
an experiment where we compared an interface
with certain information externalized versus not
externalizing it, and measured performance and
knowledge. In a second session 8 months later,
we investigated what was left of the participants

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Guidance in the Interface

knowledge and skills, and presented them with a

transfer task. The results showed that requiring
internalization can yield advantages over having
all information immediately at hand. This shows
that using cognitive findings to enhance the effectiveness of software (especially software with
specific purposes) can make a valuable contribution to the field of human-computer interaction.

Introduction
Humans interact with information in the world
around them by taking it in, processing it, and
outputting reactions. To process information, they
use cognitive skills such as thinking, learning,
reasoning, recognizing, and recalling, as well as
metacognitive skills, which entail thinking about
cognition (for instance, planning, strategizing,
or choosing between reasoning or calculation
types). Cognitive science studies these domains
of human thought. Much research in this field is
done through the analysis of subject reactions
to presented information. This makes cognitive
science a source of knowledge that couldand
doesguide interface and system designers toward a more effective presentation of information
in computer systems. We believe that utilizing
cognitive findings to enhance the effectiveness
of software can make a valuable contribution.
Increasingly humans exchange information with
the aid of computers, for instance, in education,
entertainment, office tasks, information search, email, and many other domains. Advances in computer and multimedia technology ensure that the
format of this information is increasingly diverse
using multiple media. Moreover, programs can
have hundreds of functions. However, progression
becomes difficult with this complexity of choices
and representations. Harnessing this complexity
to make it manageable for humans gave rise to the
domain of usability. Soon, among other things,
the importance of minimizing user memory load
became apparent. This resulted in recommenda-

466

tions to simplify the interface, restricting available options to those needed to carry out the task
action at hand, and to keep options visible on the
interface so users could interact on the basis of
recognition rather than recall (Nielsen, 1994). In
other words, the aim was just-in-time delivery of
just the right information, obviating the need for
memorization and extensive search in memory.
Our research does not aim at uncovering
more principles that make systems even more
usable, intuitive, or appealing. It goes beyond
plain usability and focuses on how to shape interfaces that induce a user to learn cognitive and
metacognitive skills, and thereby learn about the
domain underlying the interface. We would like
to find patterns of human behavior occurring with
computer use, to find out what kind of behavior
certain decisions in interface design provoke,
not only during interaction, but also after delays
and in transfer situations. In this, we feel that one
continually has to consider the real purpose of
the system. If a system ought to teach material to
students or children, or needs to make sure that
users do not mindlessly follow interface cues
because the task to perform is of a certain crucial
nature, then we should know what it is about an
interface design that induces people to think and
learn. In this chapter, the focus is on internalization and externalization of information, and how
this may lead to different behavior on the users
part. In the following sections, we explain the
different terms used in this context. After this,
we will look at the pros and cons of externalizing
and internalizing information, and some effects
of varying interface elements on learning and
metacognitive processes. In the next sections we
discuss an experiment on users behavior that two
interface styles (internalization and externalization) provoke, and more specifically, the amount
of planning and learning from the users side. In
the concluding section, we discuss our findings
and lay out our future plans.

Guidance in the Interface

Externalization vs.
Internalization Information:
Related Fields

Visualization of task-specific information, thus
minimizing user memory load, is often called
externalizing the information (Zhang & Norman, 1994). Externalization of information can
be contrasted with internalization of information,
whereby certain task-related information is less
directly available and needs to be internalized
(inferred and stored in memory).
Early work of Simon (1975) examining advanced chess skills and strategies to solve the
Tower of Hanoi puzzle had noticed the interdependence of external memory (perception, cueing
recognition memory), and internal memory (recall
memory). Norman (1988) argued for the need of a
similar analysis for human-computer interaction.
Tabachneck-Schijf, Leonardo, and Simon (1997)
created a model in which distributed cognition
was reified. For an example of distributed cognition, showing the interaction between internal
and external memory in creating a Microsoft
PowerPoint page, see Tabachneck-Schijf (2004).
The question is, how much information should be
internalized and how much should be externalized
for an optimal task execution? Some research
shows that the more information is externalized,
the easier executing a task becomes (e.g., Zhang
& Norman, 1994). Other research shows that
externalizing all needed information seduces
the user into mindlessly following the interface
cues and learning or planning little (e.g., Mayes,
Draper, McGregor, & Oatley, 1988). Yet other
research shows that giving users incentives to
plan and learn (i.e., internalizing information)
pays off in better performance (e.g., OHara &
Payne, 1999). Internalizing and externalizing
information are related to, but not the same as,
internal and external memory, internal and external cognition, and plan-based and display-based
problem solving. All relate to having information in the head, be it in long-term or in working

memory, versus information that is available via

(direct) perception. Simon and colleagues refer
to external memory as all information that can
be made available for perception (e.g., Greeno
& Simon, 1974). In practice, this comes down to
all information that, within the task execution,
can be made available for direct perception and
working memory. Internal memory is all the information within long-term memory that can be
made available to working memory.
External cognition is a term coined by Scaife
and Rogers (1996) and refers to the interaction
between internal and external representations (the
latter being external memory) when performing cognitive tasks (e.g., learning). Scaife and
Rogers provide a useful overview on the use of
graphical representations in educational environments and outline a theoretical framework. They
conceptualize how different kinds of graphical
representations (e.g., diagrams, animation, and
multimedia) are used during cognitive activities.
The framework presents a set of core properties,
of which the central one is computational offloading (the extent to which different external
representations reduce the amount of cognitive
effort required for cognitive activities). This property was also mentioned by Greeno and Simon
(1974) and Larkin (1989).
Plan-based and display-based problem solving
relate to which source is used to draw information from. In plan-based problem solving, one
uses detailed problem-solving strategies from
(internal) long-term memory. Display-based
makes little use of learned knowledge but relies
on (external) interface information (Larkin, 1989;
OHara & Payne, 1999).
Externalizing and internalizing information,
finally, refer to making information available on
external memory versus not making it available
on external memory, thus requiring recording it
on internal memory. We feel that studying the relationship between internalizing and externalizing
information in order to find effects of interface

467

Guidance in the Interface

information on behavior and problem solving is

an important first step.

Pros and Cons of

Externalizing Information
In many instances, and for many purposes,
externalizing information and thereby reducing working memory load is indeed a fine idea.
Externalizing interfaces take the user by the
hand, limit choices, and provide feedback. What
information should a computer system offer? If
one is working with a dynamic visual display on a
problem-solving task, problem states are directly
visible and elements that are normally stored
in working memory can now be read from the
display. A way to facilitate reduction of memory
load is by visually showing what possible steps
can be taken to go from one state to the other
in the problem space (for a problem space, see
Figure 4). It prevents users from internalizing a
great deal of information. Other ways to reduce
memory load and support explorative behavior
are, for instance, guidelines such as keep the
number of possible operations small, make the
possible operations distinguishable, make clear

what the consequences of every action will be,

make the effects of actions visible once they have
been executed, or make actions easily undoable
to make it safe to experiment (Lewis & Polson,
1990; Van Oostendorp & De Mul, 1999).
Systems such as cash dispensers, automated
airline check-in systems, and so forth should be
designed in such a way that users can simply and
easily follow the interface cues. Other examples
are wizards and help options. Graying-out menu
items that cannot be used at that moment is also
an example of externalizing information. For
example, in Word, you cannot select paste
from the edit tab in the menu when nothing
is copied or cut first. Paste is shown in gray,
indicating that the command exists, but that using it is impossible at the moment. In various
lists of user interface guidelines similar issues,
such as visibility status and visual feedback
are mentioned.
Interfaces that externalize do minimize working memory load. For instance, Larkin (1989)
considered the role played by differences in
external displays in cognitive problem solving,
finding that making information visible, enabling
display-based problem solving, helped people
recover from interruptions in work. As mentioned,

Figure 1. Examples of externalization by graying-out items

468

Guidance in the Interface

Zhang and Norman (1994) showed that the more

information was externalized, the easier solving
a problem became.
Van Joolingen (1999) studied the idea of learners actively engaging in order to learn about the
properties of what is underlying a certain domain.
He mentioned the reduction of memory load in
the context of research on the role of learner
support in discovery learning. Discovery learning assumes that learners take an active role and
construct their own knowledge base by interacting with a domain, and inferring rules from the
results of these interactions. Active involvement
of the learner would result in a better base of
knowledge in the learner as opposed to more
traditional ways of learning. The question is, of
course, how to accomplish this behavior from the
users side. He addressed the issue of supporting
discovery learning by means of cognitive tools.
Cognitive tools were defined as instruments that
support cognitive processes by relieving working
memory and which can aid the learner to direct the
process, to perform a part of it, or to externalize
part of processes and memory.
However, externalization of information also
appears to have disadvantages. There is some
evidence that during computer-based problem
solving, with most information externalized users simply follow the yellow brick road. Users
search the interface for cues and reliably find
them, perform an action that alters the interface,
thus enabling another search for cues and so,
applying simple perception-based means-ends
analysis, eventually complete the task. If the
task appears to be completed correctly, then it
follows that users will not be triggered to plan,
learn, or look for underlying rules or reasons, and
also, that people can only carry out a sequence
of actions if the interface supplies the needed
cues. In a study comparing memory of users for
a sequence of actions, researchers found just that
(Mayes et al., 1988). Users using a command-style
interface (e.g., LaTeX) that requires internalization could recall a sequence, while users using

a direct-manipulation interface (e.g., Microsoft

Word) with most cues externalized could not.
Also, external cues can easily lead subjects to
incorrect actions. Tabachneck (1992) found that
when the needed information was externalized
on a graph so that simple perception yielded the
correct answer, subjects gave the correct answer.
However, in different representational formats
simple perception yielded an incorrect answer;
here subjects also followed the interface cue and
gave the incorrect answer. In summary, externalizing information does decrease memory load,
but drawbacks are that people are likely to perform continual trial-and-error recognition-based
search and not learn or plan. Also, the external
information may lead to the wrong answer and
thus to errors. Task execution via externalized
information may also be very resistant to speeding up. We feel that there are many tasks for
which, during task execution, internalization
of information is desirable. Examples are tasks
with the purpose to acquire knowledge or skills,
be they cognitive or metacognitive; tasks one
does often, and consequently speeding up and
reducing errors is desirable (such as booking a
trip for a travel agent), or tasks where the cost
of errors is high (e.g., controlling nuclear power
plant or flying an airplane). When looking more
specifically at problem-solving tasks, planning
and learning are thought to be essential factors.
It is therefore desirable to do more research on
how interface design, internalizing, and externalizing information can stimulate people to do more
internalization. Not externalizing all information
may stimulate self-directed search, more active
planning, and learning.

Varying Elements and

(Meta)Cognitive Processes
OHara and Payne (1998) and Trudel and Payne
(1996) found that when people are induced to
plan, too strong a reliance on external information

469

Guidance in the Interface

leads to negative effects regarding planning and

transfer of skills. Inducing people to do plan-based
problem solving would lead, they hypothesized,
to internalization of metacognitive and cognitive
knowledge, in turn improving task performance
and easier recovery from errors (see also Payne,
Howes, & Reader, 2001). OHara and Paynes
(1998) experiments showed that plan-based
activity did lead to shorter solution routes and
easier recovery from errors, while a display-based
strategy involved more steps because of more
(trial-and-error) searching. OHara and Payne
used an interesting methodology to get people to
plan more. In one of two conditions, they made
the interface harder to use by imposing delays on
operators. They reasoned that subjects plan more
when there is something at stake: here, reducing
costly delays. A similar observation was made
by Svendsen (1991). Using the Towers of Hanoi
problem, a high-cost interface yielded improved
understanding of problems. In the studies of
Payne and colleagues, the amount of information externalized was not varied. We felt that
not externalizing all information will similarly
stimulate students to self-initiate a search for the
missing information to plan and to learn.
In the first session of our experiment, discussed
next, we researched whether varying the amount of
externalized information would have an effect on
performance, planning, and learning while solving a problem. In the second session we assessed
what was left after 8 months of what subjects
had learned in the first session and whether this

Figure 2. Hypotheses

470

still had an effect on task performance. Details

follow.

Experiment Session 1
Hypotheses
1. Externalization will initially yield better task
performance than internalization. Knowledge is yet to be acquired, so guidance by
indicating the legality of moves will help
externalization subjects.
2. Internalization yields better task performance later, especially after a severe distraction.

After a while, subjects will have learned to
solve the puzzle. Internalization subjects had
no guidance and had to acquire the solving
skill by themselves. They will have stored
the rules in long-term memory more solidly,
and have the needed information more readily available and thus perform better later,
especially after an interruption erased working memory. Because of the guiding nature
of the interface, externalization subjects will
plan and think less than the internalization
subjects, therefore work more on the basis
of trial and error and consequently display
worse performance.
3. Internalization yields better knowledge. Not
having externalized information available
will motivate a subject to start planning on
the basis of self-acquired rules. After the

Guidance in the Interface

experiment, we expect the explicit knowledge of rules to be memorized better by

internalization subjects. There may also be
a difference in implicit procedural knowledge, but we expect it to be smaller since
both groups will have played the puzzle a
similar number of times.
An experiment was conducted in which subjects solved a puzzle on a PC nine times in two
conditions: internalized and externalized. In terms
of performance, we expect the outcomes to take
the following course:

Phase 1: Three trials. Subjects do not have

any needed knowledge available yet, externalization offers more interface cues, and
will result in better performance.
Phase 2: Three trials. Performance in the
two conditions will be equal by this time,
because internalization subjects acquired
internal knowledge that compensates for
the interface cues.
Interruption: This task is sufficiently demanding that all of the working memory is
erased.
Phase 3: Three trials, after the interrupting
distraction task. Internalization subjects
perform better because of better internalized
knowledge and a more elaborate plan.

Materials
Our problem-solving task, Balls & Boxes
(B&B), is an isomorph of the classic puzzle
Missionaries and Cannibals (M&C). Five
missionaries and five cannibals are standing
on a riverbank, and all have to reach the other
bank by boat. Constraints are that the boat only
holds three creatures, and the minimum to sail
is one, because the boat cannot move by itself.
Furthermore, the cannibals can never outnumber
the missionaries at any place (except when there
are zero missionaries), or the missionaries will be

eaten. Our B&B problem (Figure 3) uses exactly

the same problem space (Figure 4), but is more
abstract.
We made the puzzle more abstract to reduce
the many common knowledge elements in the
classic M&C: cannibals eat people, boats cross
rivers, boats cannot sail empty, and so forth.
Using boxes, blue and yellow balls, and a dish
instead, we offer a semantically less rich problem,
avoiding the rules to be learned and remembered
too easily.
The rules of B&B were as follows:
1. Balls should be transported using the
dish.
2. You can transport at most three balls at a
time.
3. To move, the dish must contain at least one
ball.
4. The dish has to reach the boxes in turn.
5. There can never be more blue than yellow
balls in the dish (except when there are zero
yellow balls).
6. There can never be more blue than yellow
balls left in either of the boxes (except when
there are zero yellow balls).
Rules 5 and 6 are the most problematic rules,
making no semantic sense whatsoever. In our
puzzle, it was possible to consult all the rules,
but having access to these does not imply that
one knows how to solve the problem, and surely

Figure 3. The Balls & Boxes puzzle

471

Guidance in the Interface

not how to solve it in the most efficient manner.

Below we depict the formal problem space of
the puzzle (Figure 4). The numbers indicate the
number of blue and yellow balls in the left and the
right box, and the * indicates on which side the
dish is (see the legend of Figure 4). The shortest
path from A to N takes 11 moves. There are
also states of the puzzles that we call dead-end
states, such as states D, E, G, J, K, and M. These
states are not on a direct solution path, and force
backtracking.
The interface controls of the puzzle were
straightforward (Figures 5 and 6). To get balls
into or out of the dish, blue or yellow up-arrows
or down-arrows had to be clicked. To move the
pink dish horizontally, one had to click a pink

arrow (left or right). After the dish was moved,

the balls automatically dropped into the box. To
consult the rules one clicked on the rules tab.
The rules screen overlaid the puzzle screen. The
puzzle was designed in two versions:
1. Internalized: All arrows were always colored (Figure 5) and clickable, providing
no information about the legality of moves
(performing illegal moves is possible). One
could click all buttons at all times; however,
illegal moves would be carried out only
partially. For instance, if one wanted to
transport the dish empty from left to right
and clicked the right pink arrow, the empty
dish would move to the right. Then, an error

Figure 4. Problem space of the Balls & Boxes problem

472

Guidance in the Interface

notification would pop up saying, This is not

possible. By clicking OK in the dialogue
box, the program would undo the move.
2. Externalized: The arrows are colored (and
thus clickable) only when an action is legal.
The arrows are grayed-out (unclickable, as
in Figure 6) when a move is illegal. For instance, in Figure 6, moving the dish empty
from left to right is illegal (violating rule
3), externalized by graying-out both pink

arrows. It is now only legal to move balls

up into the dish.

Subjects and Design

Thirty subjects of age 1928 (mean age 21.8),
experienced with PCs, were recruited at Utrecht
University. Subjects were paid 5. The puzzle, a
Java-applet, ran on a Pentium IV 2.0 GHz PC with
a 17 monitor. Our design has one random-as-

Figure 5. Internalized version: No information whether an action is possible (no gray-out)

Rules-tab

Pink arrows

Yellow arrows/balls

Blue arrows/balls

Figure 6. Externalized version: Information whether an action is possible (grayed-out)

Greyed out buttons

473

Guidance in the Interface

Performance measures (logged)

Number of puzzles subjects solved per

phase (maximum three per phase)
Speed: time needed to solve the puzzles
Reaching dead-end states. These are
puzzle states that are off a direct solution path, indicating trial-and-error
search (states D, E, G, J, K, and M in
Figure 4)

Knowledge test: After the puzzles, we

measured implicit and explicit knowledge
of the rules and shortest-path solutions of
the problem, using a paper questionnaire.
Attitudes: Likert-scale questions concerning
perceived amount of planning, feeling lost
during interaction, and other such measures.

Procedure
The experiment was conducted in the usability
lab at the Center for Content and Knowledge
Engineering, Utrecht University. We informed
the subjects of the course of the experiment, and
gave a brief oral explanation of the interface and a
short demonstration. The experiment consisted of
nine puzzle trials, divided into three equal phases,
and a 10-minute distraction task between phase 2
and 3. The maximum time for each trial was set
at 7 minutes. Slightly different starting situations
of the puzzle were used to avoid subjects simply
repeating actions (states A, B, and C in Figure
4). Also, in the second phase, the playing direction of the puzzle was reversed to right to left.
In the third phase, the playing direction was set
to left to right again. After the last trial, subjects
filled out a knowledge test (score 08) consisting

474

of four multiple choice and four open questions

with screenshots of puzzle situations. They had
to judge and explain whether and why certain
actions were possible (implicit knowledge), and
recall the rules (explicit knowledge). Subjects
also rated how clear the rules were for solving
the problem.

Results Experiment, Session 1

Number of Puzzles Solved per Phase
Subjects could attempt to solve three puzzles per
phase. A MANOVA (analysis of variance) showed
a significant main effect of phase on the average
number of puzzles solved, F(2,56) = 53.74; p <
.001 (Figure 7) (pointer to figure). The number
of puzzles solved (out of three) improved in later
phases, indicating a learning effect. Post-hoc comparisons showed that only the difference between
performance in phase 1 and phase 2 was significant
(p < .05) (Figure 8) (pointer to figure). Although
the graph suggests an overall better performance
for internalization (M = 6.93, SD = 2.05 vs. M =
6.33, SD = 3.09), this did not reach significance.
There were no interaction effects.

Figure 7. Puzzle trials solved per phase, per

version
3 .0
2 .5

T ria ls s o lve d p e r p h a se

signment between-subjects independent variable:

internalization versus externalization. Among the
dependent variables were the following:

2 .0
1 .5
1 .0

Inte rfa ce style

.5
0 .0

Inte rna liza tio n

1

2
P ha se

E xte rna liza tio n

Guidance in the Interface

Time Needed for the Puzzle Trials

A MANOVA showed a main effect also for phase
on average solving time, F(2,42) = 35.16; p < .001.
The time subjects needed to solve puzzles lessened
in later phases, also a learning effect. Post-hoc
comparisons showed that all subjects needed
significantly more time in phase 1 than in phase
2, and also in phase 2 than phase 3.
The graph suggests that internalization subjects need some more time on average, but ANOVA
showed that this was just a trend (F(1,21) = 2.79,
p = 0.11). Looking closer, it was only in the beginning, in phase 1, that internalization subjects
needed more time (M = 238.77, SD = 62.20) to
solve the puzzles than externalization subjects
(M = 186.77, SD = 67.08). This difference was
significant, t(21) = 1.94, p = 0.03 (one-sided). After
this first phase, the differences were minuscule
and no longer significant.

Dead-End States

dition, there is a trend for an interaction effect

of phase and version (F(2,56) = 2.11; p = 0.13).
Internalization subjects significantly improved
from phase 1 to 2 (M = 11.4, SD= 4.70 and M =
3.4, SD = 3.18), t(14)=5.80, p = 0.00, one sided).
This is an improvement of eight states. They also
improved from phase 2 to 3 by almost two states,
nearly reaching floor level (M = 3.4, SD= 3.18 and
M = 1.47, SD = 2.07), t(14)=1.96, p=0.04, one sided.
Externalization subjects improved by 6.2 states
from phase 1 to 2 (M = 12.67, SD= 6.91 and M =
6.47, SD = 7.11), t(14)=2.74, p = 0.02, one sided.
But after the interruption, it was different. From
phase 2 to 3, they did not further improve. On the
contrary, in phase 3 they reached more dead-end
states than before, although not significantly so.
Here in phase 3 the difference between internalization (M = 1.47, SD = 0.53) and externalization
(M = 7.73, SD = 10.64) was significant, t(28)=
2.24, p = 0.04. Externalization subjects reached
more than six more dead-end states in phase 3
than internalization subjects.

Knowledge Test

Figure 8. Time needed per solved trial, per version

Figure 9. Dead-end states per phase, per version

300
250
200
150
100

Inte rfa ce style

Inte rna liza tio n

50
0

2
P ha se

E xte rna liza tio n

Before measuring knowledge of the rules and

states, we confirmed that there was no difference
in the number of times subjects in both conditions consulted the rules. The knowledge test that

Nu m b e r o f d e a d -e n d -sta te s re a ch e d

A ve ra g e tim e (se co n d s) p e r so lve d tria l

Below is the number of times per phase subjects

reached dead-end states.
A MANOVA showed an interesting nearly
significant main effect of interface style (Figure
9). Overall, externalization subjects reached more
dead-end states (F(1,28) = 3.58; p = 0.06). In ad-

14
12
10
8
6
Inte rfa ce style

Inte rna liza tio n

2
0

E xte rna liza tio n

P ha se

475

Guidance in the Interface

subjects received after the experiment consisted

of several procedural knowledge questions and
one explicit knowledge question. The Balls &
Boxes puzzle contained more than 30 legal states
(Figure 4). There were seven procedural implicit
knowledge questions about seven of those states
(open and multiple choice) in which subjects were
visually presented with a puzzle state. They had to
judge whether certain moves led to the solution,
and explain why. The scores on those seven questions were high for both internalization (M = 6.3,
SD = 0.96) and externalization subjects (M = 5.8,
SD = 1.20), considering that the maximum score
was seven. There was a trend that internalization
on average scored higher than externalization
on the procedural knowledge questions (t(28) =
1.17, p = 0.12).
To test the explicit rule knowledge, the possibilities were limited. The puzzle had only few
rules, some of which (rules 1 to 4) were extremely
easy to grasp and remember. Rules 5 and 6 are
more crucial and define the difficulty of the puzzle.
They can be merged to just one rule: Blue balls
can never be in the majority at any place, except
when there are only blue balls (zero yellow balls).
Subjects were asked about this rule with a multiple-choice question. All internalization subjects
answered this question correctly, whereas of the
externalization subjects only 60% answered it
correctly. This difference was significant (Phi =
-.50, p < 0.01). Furthermore, subjects had to estimate the clarity of the rules. They rated the question the given rules were clear enough to solve
the problem (score range 15). Internalization
subjects found the rules clearer than the subjects
from the externalization condition (M = 4.13, SD
= 0.52 and M = 3.53, SD = 1.25), t(28)=1.72, p =
.04 (one-sided).

Summary Experiment Session 1

Our first hypothesis stating that initially externalization yields better performance was only partly

476

supported in terms of the time subjects needed.

Internalization subjects did take more time in the
first phase. Other measures were in the expected
direction, but not significantly so. Unexpectedly,
externalization yielded no better performance in
terms of puzzles solved in the beginning, as we
expected.
The second hypothesis stating that internalization yields better performance in a later stage was
partly supported. We saw that the time subjects
needed stabilized, and the same was true for the
number of trials solved per phase. However, we
also looked at performance in a more delicate
manner, not in terms of time or correctness, but
at how, via what route, subjects reached their goal.
We introduced the measure dead-end states to
inform us as to how subjects behaved, in terms
of the insight they had, the smartness of their
route. We assumed that internalization subjects
do some smarter, more elaborate planning, while
externalization subjects are expected to solve more
by trial and error and on the basis of interface cues.
It showed that internalization subjects performed
better after the interruption and reached those
dead-end problem states less often in all three
phases (indicating less lostness). Furthermore,
there was also the trend-like interaction that after
the interruption, internalization subjects kept
improving, while externalization subjects fell
back, reaching more dead-end states than they did
before. This confirms our expectation that after
the interruption, internalization subjects would
continue to work on the basis of the plan-based
strategy as they did before. Externalization subjects, on the other hand, perform no better after
the interruption. They fell back to depending on
the interface, having a less elaborate plan.
The third hypothesis in which we expected
that internalization would result in having better knowledge was supported. We also assumed
that internalization subjects, who could rely less
on interface information, had to build a stronger, more elaborate plan. Though when testing
implicit knowledge both groups scored equally

Guidance in the Interface

high, with a trend advantage for internalization

subjects, when asked about explicit knowledge
of the crucial rule that defines the difficulty of
the puzzle there was a significant difference.
All the internalization subjects could correctly
answer this question, whereas only 60% of the
externalization subjects couldin spite of having the rules readily available for consultation
and consulting them as often as internalization
subjects. Furthermore, there was also the tendency
that internalization subjects rated the clarity of
the rules higher. This is intriguing, because in
the externalization version of the puzzle subjects
had interface feedback and were able to consult
the rules. Internalization subjects, who only had
the rules and no interface help found the rules
clearer. We carefully interpret the latter two findings as indicators of better understanding in the
internalized condition.

Experiment Session 2
We were curious to see how stable this better
knowledge provoked by internalization was, and
therefore decided to conduct a second session
after a delay of 8 months. We invited the same
subjects of experiment 1; 14 of the 30 were available. There were two reasons for this rerun. First,
to see whether the better knowledge measured
in the internalization subjects had endured, we
asked subjects to solve B&B five times (experiment 1 showed three to four puzzles suffice for all
subjects to be able to solve the puzzle within the
allotted time). Second, to see whether the better
knowledge might result in better performance on
a transfer task, we also confronted subjects with
a transfer problem. Transfer problems require
subjects to apply acquired skill on a different
task of the same nature. To be able to measure
differences in performance between the two initial groups (internalization and externalization)
on the same material, we presented all subjects
with the same material this time, one interface

style, namely, externalization. Note that the internalization subjects had to make a change in
interface style.

Hypotheses
1.

2.

Because internalization subjects still have

better memory of the rules and solution, they
will perform better. We expect the internalization subjects to solve the puzzle faster
initially because of their better knowledge
of the rules. They will be faster in recollecting knowledge needed to solve the puzzle.
After a while, we expect the two groups to
perform more or less equally on this puzzle,
like in session 1.
Internalization subjects will perform better
on the transfer tasks. After five times of
B&B, we expect the two groups to perform
at the same level. But when confronted with
transfer tasks (that have similarities but
also a few differences), we expect internalization subjects to perform better, again
because they possess better knowledge of
the rules.

Materials
To test knowledge retention, all subjects first
solved the B&B puzzle in the externalized version five times. To test transfer performance, we
used another puzzle of the same M&C family,
but with varying characteristics. We first used a
quite literal version of M&C, which was further
away in terms of transfer.

Balls & Boxes

The only difference between this and the previous B&B was that the rules were not consultable
anymore, so we could obtain a more pure measure of retention. All subjects received the same
externalized version. This meant that attempting
illegal moves (moves that violate rules) was not

477

Guidance in the Interface

possible, as the externalized interface only allowed legal moves.

1.

Missionaries and Cannibals

The second problem was a transfer task, ironically
the original version of the M&C. It was a game
that literally shows missionaries, cannibals, a
river, and a boat. The solution algorithm to this
problem was the same as B&B, but there were
some difficulties that did not exist before. The
most important one is that there were not five of
each entity as before, but three, and the playing
direction was always from right to left (Figure
10). Although it might seem that one can solve the
three-creature problem faster, this is not sothe
shortest path to the solution is also 11 moves.
Also, unlike in B&B, in this game attempting
illegal moves does not lead to an error message,
but subjects would die, as in a computer game,
and the game would start over. Errors thus had
more severe consequences, and subjects needed to
exercise caution. Subjects had to solve the problem
as many times as they could in 8 minutes.

2.

After completing the experiment, subjects

received a 5 reward.

Results Experiment Session 2

Subjects and Procedure

Fourteen of the 30 subjects from session 1 were
willing to participate a second time. Luckily, seven
of them had worked with internalization, and seven
with externalization in session 1. We confirmed
that these 14 subjects were subjects who performed
normally (no extreme scores) in session 1. The
experiment consisted of two parts:

B&B (five times). The maximum time for

each trial was set at 7 minutes. Slightly
different starting situations of the puzzle
were used to avoid subjects simply repeating
actions.
M&C (8 minutes).

Balls & Boxes. Encouraging results were

found concerning solving the puzzle correctly again the first time. After not having
seen the puzzle for 8 months, it took the
internalization subjects only half the time
used by the externalization subjects to solve
the first B&B puzzle (M = 432 sec, SD =
314 and M = 778 sec, SD = 397). This difference was significant, t(12)= 1.81, p =
.05. There were no further significant differences between the two groups. After the
first puzzle, as expected, all subjects solved
the remaining four trials of B&B puzzle
equally well, just as 8 months ago.
Missionaries & Cannibals. This puzzle
was a bit further in transfer. The graphical characteristics of this puzzle differed
considerably. Still the algorithm to solve
it in itself was the same, but the number of
creatures to transport was different, and also
the maximum number of creatures allowed
in the boat. Although the shortest solution

Figure 10. The initial state of the Missionaries & Cannibals puzzle

478

Guidance in the Interface

path is the same length as B&B, the problem

does have a smaller problem space. The same
basic concept had to be applied to a situation
that differed at face value, and also underneath. Out of our 14 subjects, 10 managed to
solve this puzzle one or more times. Just as
in the B&B puzzle, internalization subjects
solved it the first time faster (M = 176 seconds, SD = 72.5 vs. M = 302.8, SD = 202.7),
though it was just a trend, t(8) = 1.44, p <
0.10 (one-sided). Moreover, internalization
subjects managed to solve the puzzle three
times more often (M = 4.83, SD = 1.94) in
the 8 minutes than externalization subjects
(M = 2.5, SD = 1.73). This was significant,
t(8)= 1.94, p < .05 (one-sided).

Summary Experiment Session 2

Eight months is a long time, but still the interface
style subjects received in session 1 appeared to
make a difference. Subjects were not told at the
first session that they would be asked to play the
puzzle again, and when we approached them, we
explicitly told them it was not the same experiment
they had participated in. Upon being confronted
with the puzzle again, the subjects had to recollect the rules and solution strategy from their
long-term memory.
The two groups both worked with an externalized version this time and had exactly the same
information. We looked at how long it would take
subjects to remember how to do this correctly. It
showed that in solving the first B&B again, internalization subjects were indeed significantly faster,
almost twice as fast, in fact. This is all the more
remarkable given that these subjects received,
to them, an unfamiliar interfaceexternalized
rather than internalized. After that first success,
the performance of both groups equalized. This
is coherent with the results from the first session,
where stabilization in performance also occurred.
The above finding supports our first hypothesis of

experiment 2 stating that internalization subjects

would still have a better memory of the rules and
solutions of the puzzle.
The M&C puzzle that subjects received was
the same for everyone. It can be regarded as an
internalization version (no direct interface clues
were available, all controls were available all
the time). Also here there were some interesting
results. The subjects that worked with the internalization version of B&B managed to solve this
puzzle three times more often than externalization
subjects, and this was significant. Furthermore,
just like in B&B, internalization subjects needed
less time to solve the problem the first time, although this was just a trend. It supports the second
hypothesis stating that internalization subjects
will perform better on a transfer task. In summary,
we demonstrated that the interface style subjects
worked with 8 months ago still appeared to be of
influence months later, both in solving the same
puzzle again as in solving a far transfer puzzle.
We take it as encouraging support for the better
memory of knowledge provoked by the interface
style subjects worked with.

General Discussion on the

Two Sessions
We analyzed the influences that externalizing
certain information on the interface, thereby
making it visible, or not externalizing it has on
performance and knowledge acquisition in a
problem-solving task. Furthermore, we looked
at the long-term effects after 8 months, and at
the issue of transfer. Both in session 1 and in
session 2, the delayed session, NOT externalizing information led to enduring advantages in
performance and knowledge, while externalizing
information merely led to a fleeting advantage at
the very start of session 1.
According to Zhang (1997), the more information is externalized, the easier performing a task
becomes. We hypothesize, however, that being

479

Guidance in the Interface

led by the interface also incurs a cost, namely, a

lessening of metacognitive activity such as planning, and consequently, less learning. In such
situations as where learning is required, transfer
is at stake, or the task is prone to interruptions,
planning, learning, and in general, getting to know
more about the structure of the task at hand are
desirable outcomes. Surprisingly, performance
measures such as time needed to solve the puzzles
and number of correctly solved puzzles were
not much influenced by the interface styleif
anything, the internalization group was at an
advantage. Zhangs (1997) prediction was not
confirmed. We feel that the attention of users in
the externalization group was immediately taken
by the guiding nature of the interface, and as they
were not confronted with actual mistakes (one
could not make illegal moves, only inefficient legal
moves were allowed), they simply kept on solving without feeling a need to figure the problem
out more. Subsequently, less active learning took
place. This idea of attention taken by an interface
fits with Carroll and Rossons (1987) paradox of
the active userusers of computer systems are
so consumed with immediate productivity that
they are not motivated to take time to learn better
ways of accomplishing a task.
Contrary to the externalization group, the
internalization group was confronted with errors
for which an explanation was not immediately
available. These subjects incurred a cost when
making errors: a dialog box popped up which they
had to click away, and their attempted move was
reversed. Though both groups consulted the rules
sheets equally often, this cost probably contributed to motivating the internalization subjects to
study the rules better in order to avoid incurring
the cost. We found that the internalization group
possessed better explicit knowledge of the rules
and engaged in more planful problem solving.
Applying more metacognition to avoid a cost
concurs with the findings of OHara and Payne
(1998, 1999).

480

During session 2 we found that the effects

found in session 1 were still present. Again, only
the internalization subjects from months ago
showed advantages. This group was both faster
in solving the same problem again, and also faster
in figuring out the solution to a transfer task. It
is remarkable that the influence of working with
one interface or the other has effects even after
such a long time.
In general, it was surprising for us to see
that externalization showed even less advantage
than we thought, almost none, to be precise. We
expected externalization at least to be of help in
the beginning when users were not familiar with
the system and the problem. It was, but only in
time taken on the first few trialsand this was
just a trend that was not confirmed in the number
of puzzles solved correctly. This very small and
fleeting advantage did not concur with Zhangs
(1997) findings. Please note that our type of
externalization parallels Zhangs more physical
type: both restrict the number of illegal moves
that can be made, but do not alter the problem
space or give cues as to what the shortest path is
through the problem space.
We find the results so far encouraging in that
lessening the amount of externalized knowledge
apparently can encourage cognitive and metacognitive behavior. Moreover, one can be very
precise about which information not to externalizein other words, this way the information to
be learned can be manipulated. Therefore, we
feel that the issues of manipulating the amount of
externalization in interfaces deserve more attention. In the context of current technology and the
widespread use of computer systems in virtually
all domains, we are convinced that understanding how the systems users will perceive, reason,
and interact with the system can be crucial. The
implications of the findings so far and the ones
in the future can be valuable for development
of applications where active engagement and
learning from users is the aim. Examples are
systems to teach material to students or children,

Guidance in the Interface

or situations where it is important that users are

not kept stupid by the software. In certain situations, users of a system need to understand the
underlying structure/rules of the system because
the tasks are of a crucial nature. To accomplish
this we have to look beyond plain usability and
common sense and better scrutinize the interaction
between internal memory and cognition on the
one hand and external memory and recognition
and perception on the other hand. We believe this
can, for certain types of systems, enhance the
effectiveness in achieving their targets.

Future Trends
There are still many challenges in human-computer interaction and many issues that need to be
explored. Understanding how users will react to
interface information (on the basis of cognitive
findings) is one important issue in attuning software to its purpose, thereby allowing it to achieve
its goal. In the future, we will further investigate
this issue by exploring behavior in different types
of more realistic planning-related tasks. As a more
realistic planning task, we think of, for example,
spreadsheet or drawing applications where actions
are less repetitive, more complex, and could be
part of a real job. We are currently designing such
an environment.

References
Carroll, J. M., & Rosson, M. B. (1987). The
paradox of the active user. In J. M. Carroll (Ed.),
Interfacing thought: Cognitive aspects of humancomputer interaction (pp. 80-111). Cambridge,
MA: MIT Press.
Greeno, J. G., & Simon, H. A. (1974). Processes
of sequence production. Psychological Review,
81, 8798.

Larkin, J. H. (1989). Display based problem solving. In D. Klahr & K. Kotovsky (Eds.), Complex
information processing: The impact of Herbert
A. Simon (pp. 319-341). Hillsdale, NY: Lawrence
Erlbaum.
Lewis, C. H., & Polson, P. G. (1990). Theory-based
design for easily learned interfaces. Special issue:
Foundations of human-computer interaction.
Human Computer Interactions, 5, 191220.
Mayes, J. T., Draper, S. W., McGregor, A. M.,
& Oatley, K. (1988). Information flow in a user
interface: The effect of experience and context
on the recall of MacWrite screens. Proceedings
of the Fourth Conference of the British Computer Society on People and Computers, IV (pp.
275289).
Nielsen, J. (1994). Usability engineering. San
Francisco: Morgan Kaufmann.
Norman, D. A. (1988). The psychology of everyday
things. New York: Basic Books.
OHara, K., & Payne, S. J. (1998). The effects of
operator implementation cost on planfulness of
problem solving and learning. Cognitive Psychology, 35, 3470.
OHara, K. P., & Payne, S. J. (1999). Planning
the user interface: The effects of lockout time
and error recovery cost. International Journal of
Human-Computer Studies, 50, 4159.
Payne, S. J., Howes, A., & Reader, W. R. (2001).
Adaptively distributing cognition: A decisionmaking perspective on human-computer interaction. Behaviour & Information Technology,
20(5), 339346.
Scaife, M., & Rogers, Y. (1996). External cognition: How do graphical representations work?
International Journal of Human-Computer Studies, 45, 185213.
Simon, H. A. (1975). The functional equivalence
of problem solving skills. Cognitive Psychology,
7, 268288.
481

Guidance in the Interface

Svendsen, G. (1991). The influence of interface

style on problem solving. International Journal
of Man-Machine Studies, 35, 379397.
Tabachneck-Schijf, H. J. M. (1992). Computational
differences in mental representations. Effects
of mode of data presentation on reasoning and
understanding. Doctoral dissertation, Department of Psychology, Carnegie Mellon University,
Pittsburgh, PA.
Tabachneck-Schijf, H. J. M. (2004). How the
format of information determines our processing:
An IPP/CIP perspective. In H. van Oostendorp,
L. Breure, & A. Dillon (Eds.), Creation, use and
deployment of digital information (pp. 123144).
Mahwah, NJ: Lawrence Erlbaum.
Tabachneck-Schijf, H. J. M., Leonardo, A. M., &
Simon, H. A. (1997). CaMeRa: A computational
model of multiple representations. Cognitive Science, 21(2), 305350.

Trudel, C. I., & Payne, S. J. (1996). Self-monitoring during exploration of an interactive device.
International Journal of Human-Computer Studies, 45, 723747.
Van Joolingen, W. R. (1999). Cognitive tools for
discovery learning. International Journal of Artificial Intelligence in Education, 10, 385397.
Van Oostendorp, H., & De Mul, S. (1999). Learning by exploration: Thinking aloud while exploring an information system. Instructional Science,
27, 269284.
Zhang, J. (1997). The nature of external representations in problem solving. Cognitive Science,
21(2),179217.
Zhang, J., & Norman, D. (1994). Representations
in distributed cognitive tasks. Cognitive Science,
18, 87122.

This work was previously published in Cognitively Informed Systems: Utilizing Practical Approaches to Enrich Information
Presentation and Transfer, edited by E. M. Alkhafia, pp. 74-101, copyright 2006 by IGI Publishing, formerly known as Idea
Group Publishing (an imprint of IGI Global).

482

483

Chapter 2.8

Ergonomic User Interface

Design in Computerized
Medical Equipment
D. John Doyle
Cleveland Clinic Foundation, USA

Abstract

Introduction

Current statistics suggest that preventable medical

error is a common cause of patient morbidity and
mortality, being responsible for between 44,000
and 98,000 deaths annually, and resulting in
injuries that cost between $17 billion and $29 billion annually. An important approach to tackling
this problem is to apply system design principles
from human factors engineering (ergonomics). By
doing so, systems and equipment become easier
for people to work with, ultimately reducing the
frequency of errors. In particular, in the case of
medical equipment, the design of the user interface
can impact enormously on its successful use. In
this chapter we consider some of the elements of
good and bad medical equipment design, using
examples drawn from the literature and elsewhere.
The concept of ecological interface design is also
discussed, and some practical design guidelines
are provided.

American statistics suggest that preventable

medical error is the eighth leading cause of death,
being responsible for between 44,000 and 98,000
deaths annually, and resulting in injuries that
cost between $17 billion and $29 billion annually
(Kohn, Corrigan, & Donaldson, 1999). Experts
have often stated that an important approach to
improving patient safety is to apply system design principles from human-factors engineering
(ergonomics; Kohn et al.; Leape, 1994). Humanfactors engineering is a relatively young scientific
discipline that focuses on those factors that affect
the performance of individuals using systems or
equipment (Kroemer, 2001). The product may
be as simple as a spoon or an office chair, or as
complex as an aircraft carrier, but in all cases the
goal is to design products to conform to human
nature rather than merely expect people to adapt to
technology. By doing so, systems and equipment
become easier for people to work with, ultimately
reducing the frequency of errors.

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Ergonomic User Interface Design in Computerized Medical Equipment

In the case of medical equipment, the design

can impact enormously on its successful use. In
particular, errors in operating such equipment
are often caused, at least in part, by the design of
the user interface. Of course, such errors can not
only hamper patient care, but in some cases can
even lead to injury or death. It is obviously important that medical equipment be designed with
special consideration given the impact of design
on safe operation. Thus, the user interface for
medical equipment should be straightforward and
intuitive: If its operation is excessively complex
or counterintuitive, safety can be compromised.
Human-factors techniques have been applied to
other industries, such as nuclear power and aviation, and have been very successful in reducing
error and improving safety in these contexts.
Note also that in addition to increasing safety,
an added benefit of using good ergonomic design
practices is the likelihood that training costs will
be reduced.

Bad Design Examples

Examples of perplexing, arcane, and hazardous
designs produced in violation of ergonomic principles are not hard to find. For instance, Michael J.
Darnells Web site www.baddesigns.com offers a
collection of frequently humorous examples. But
when bad designs in medical equipment lead to
injury or death, the situation can be far from amusing. This is sometimes the case for computerized
medical equipment.
In one case reported on the U.S. Food and
Drug Administration Web site (http://www.fda.
gov), a patient was overdosed after a nurse read
the number 7 as a 1 in the drug-infusion pump
display. Because the flow-rate display was recessed, the top of the 7 was blocked from view
at many viewing angles.
In another case report from the same source,
a physician treating a patient with oxygen set
the flow-control knob between 1 and 2 liters

484

per minute, not realizing that the scale numbers

represented discrete, rather than continuous,
settings. Unbeknownst to the physician, there
was no oxygen flow between the settings, even
though the knob rotated smoothly, implying that
intermediate settings were available. The patient,
an infant, became hypoxic before the error was
discovered. One solution could have been a rotary
control that snaps into discrete settings.
In yet another case, a patient on a ventilator
died following the accidental detachment of the
ventilator tubing from the humidifier. Unfortunately, an alarm did not sound because the pressure-limit alarm setting had been set so low that
it was essentially nonfunctional.
Finally, Figure 1 illustrates a less hazardous
example drawn from personal experience.
A series of reports from the laboratory of
Dr. Kim Vicente of the University of Toronto
have looked at user-interface issues for patientcontrolled analgesia (PCA) equipment (Lin,
Isla, Doniz, Harkness, Vicente, & Doyle, 1998;
Lin, Vicente, & Doyle, 2001; Vicente, KadaBekhaled, Hillel, Cassano, & Orser, 2003). PCA
is a computer-based medical technology used
to treat severe pain via the self-administration
of analgesic agents such as morphine. Potential
benefits include superior pain control, automatic
documentation, and improved utilization of nursing resources.
Unfortunately, however, one of these units (Abbott Lifecare 4100 PCA Plus II) has been linked
to a number of overdose deaths. This machine
is easily set up incorrectly by caregivers, who
must manually enter the PCA parameters, and a
number of patients have received drug overdoses
as a result of user errors when using this product:
the insertion of a 5 mg/mL morphine cartridge
when the machine is expecting a 1 mg/mL concentration, or the acceptance of the default (initial)
drug concentration when the correct action is to
scroll up to the correct value, among other errors.
In the latter case, when nurses program the drug
concentration, the Lifecare 4100 display shows a

Ergonomic User Interface Design in Computerized Medical Equipment

Figure 1. The user interface for medical equipment should be straightforward, friendly, and intuitive.
Also, rarely is the operating manual available to the end user, which makes the labeling of the controls
especially important. Consider then the user controls shown for this operating-room table in use at
the authors facility. The top and bottom left-hand controls lower the head and feet, respectively, while
the right-hand controls raise the head and feet. But what if the entire operating table is to be raised or
lowered, which is by far the most common request from the surgeon? It turns out that the entire table is
raised by pushing both right-hand buttons, while the entire table is lowered by pushing both left-hand
buttons. This arrangement makes sense if one thinks about it for a while, but an intuitive interface should
not require a lot of thinking. Furthermore, there is plenty of space available on the control panel to add
two extra buttons.

particular concentration (e.g., 0.1 mg/mL). Nurses

can either accept this initially displayed value or
modify it using the arrow controls. The critical
flaw in the design is that in this situation, the
Lifecare 4100 offers the minimal drug concentration as the initial choice. If nurses mistakenly
accept the initially displayed minimal value (e.g.,
0.1 mg/mL) instead of changing it to the correct
(and higher) value (e.g., 2.0 mg/mL), the machine
will run as if the drug is less concentrated than
it really is. As a result, it will pump more liquid,
and thus more of the drug, into the patient than
is desired.
Aware of the dangers of the Lifecare 4100, Lin,
Isla, et al. (1998) and Lin, Vicente, et al. (2001)

studied the unit using cognitive task-analysis

techniques. Based on this analysis, the interface
was then redesigned to include a dialog structure
with fewer steps, a dialog overview showing the
users location in the programming sequence,
better command feedback, easier error recovery,
and clearer labels and messages (Lin, Isla, et al.,
1998, p. 253). Studies of the new interface showed
significantly faster programming times, lower
mental workload, and fewer errors compared to
the manufacturers original interface. Regrettably,
the improved interface design was not used by
the manufacturer.

485

Ergonomic User Interface Design in Computerized Medical Equipment

Design Guidelines
The U.S. Food and Drug Administration has
offered a number of guidelines to help with the
design of medical equipment, such as the following (adapted from http://www.fda.gov):

Make the design consistent with user expectations; both the users prior experience
with medical devices and well-established
conventions are important.
Design workstations, controls, and displays
around the basic capabilities of the user,
such as strength, dexterity, memory, reach,
vision, and hearing.
Design well-organized and uncluttered
control and display arrangements. Keys,
switches, and control knobs should be sufficiently apart for easy manipulation and
placed in a way that reduces the chance of
inadvertent activation.
Ensure that the association between controls
and displays is obvious. This facilitates
proper identification and reduces the users
memory load.
Ensure that the intensity and pitch of auditory
signals and the brightness of visual signals
allow them to be perceived by users working
under real-world conditions.
Make labels and displays so that they can
be easily read from typical viewing angles
and distances.
Use color and shape coding to facilitate the
identification of controls and displays. Colors
and codes should not conflict with industry
conventions.

Ecological Interface Design

Ecological interface design (EID) is a conceptual
framework developed by Vicente and Rasmussen (1990, 1992) for designing human-machine
interfaces for complex systems such as are often

486

found in process-control applications or in computer-based medical equipment. The primary goal

of EID is to aid operators, especially knowledge
workers, in handling novel or unusual situations.
Studies suggest that the use of EID methods can
improve operator performance when compared
with classical design approaches (Vicente, 2002).
The basis for EID is Rasmussens (1986) skillsrules-knowledge model of cognitive control, and
in this model, critical incidents can result from
errors at any of the skills-rules-knowledge levels
of human cognition.
The first form of incident includes skill-based
errors involving the faulty execution of an otherwise correct plan. Here, behavior is unconscious,
nonverbal, and automatic. An example would
be inadvertently turning on the wrong switch.
Even the most experienced clinicians are prone
to skill-based errors as they often occur during
highly routine procedures such as in reading a
drug label or adjusting a control.
A second category is rule-based errors and
involves the failure to apply a rule, such as stopping at a stop sign when driving a car, or not
administering a drug to which the patient is allergic. At this level, one step up in Rasmussens
(1986) cognitive model, people use stored (or
precompiled) rules acquired with training or with
experience on the job.
Lastly, there are knowledge-based errors in
which the initial intention is itself wrong, often
due to inadequate knowledge or experience. A
clinical example would be administering 10 mg
of morphine to an infant. This corresponds to
Rasmussens (1986) highest cognitive level, and
is most suited for unfamiliar environments where
prior experience is unavailable to provide a system
of rules, such as troubleshooting a new piece of
medical equipment for the first time.
Under Rasmussens (1986) skills-rules-knowledge model, human behavior moves along a ladder
as experience increases. Early on, when one is
placed in an unfamiliar environment, problemsolving behavior will be at the knowledge level.

Ergonomic User Interface Design in Computerized Medical Equipment

As experience is gained so that rules can be

formed, the rules level takes over. In some situations, further experience may lead to even further
automation (skills level).
For each of the three cognitive levels, the way
in which information and environmental cues are
perceived differs. Signals guide skill-based behavior, while symbols apply to knowledge-based
behavior. Signals supply time-space information
only; they have no meaning at higher levels, and
they cannot be verbalized. Signs may trigger rules
(stop, start, etc.) or may indicate the state of the
system (valve open or closed), but they do not
express functional relationships (e.g., the consequences of an open valve). Finally, symbols refer
to concepts that support analytical reasoning, such
as modeling the system to allow one to determine
the consequences of an open valve.
What does all this have to do with avoiding
human error when operating complex medical
equipment? The answer lies in the following.
Rasmussens (1986) three levels of cognition can
be grouped into two broader categories: (a) analytical-based behavior (knowledge-based behavior)
and (b) perceptual-based behavior (rule and skill
based). Such a categorization is helpful because
perceptual processing has important advantages
over analytical-based behavior: Analytical behavior is slow, demanding, and serial in nature,
whereas perceptual behavior is fast, effortless,
parallel, and less error prone. Thus, the goal of
design should be to help people avoid situations
requiring them to work at the knowledge-based
level, while supporting the use of analytical
problem solving for use in unfamiliar situations.
And, as emphasized above, design guidelines that
match the environment to the people involved is
known as ecological interface design.

References

system. Washington, DC: National Academy

Press.
Kroemer, K. H. E. (2001). Ergonomics: How to
design for ease and efficiency (2n d ed.). Upper
Saddle River, NJ: Prentice Hall.
Leape, L. L. (1994). Error in medicine. The
Journal of the American Medical Association,
272, 1851-1857.
Lin, L., Isla, R., Doniz, K., Harkness, H., Vicente,
K. J., & Doyle, D. J. (1998). Applying human
factors to the design of medical equipment:
Patient-controlled analgesia. Journal of Clinical
Monitoring and Computing, 14, 253-263.
Lin, L., Vicente, K. J., & Doyle, D. J. (2001). Patient
safety, potential adverse drug events, and medical device design: A human factors engineering
approach. Journal of Biomedical Informatics,
34, 274-284.
Rasmussen, J. (1986). Information processing and
human-machine interaction: An approach to cognitive engineering. New York: North-Holland.
Vicente, K. J. (2002). Ecological interface design: Progress and challenges. Human Factors,
44, 62-78.
Vicente, K. J., Kada-Bekhaled, K., Hillel, G.,
Cassano, A., & Orser, B. A. (2003). Programming
errors contribute to death from patient-controlled
analgesia: Case report and estimate of probability.
Canadian Journal of Anaesthesia, 50, 328-332.
Vicente, K. J., & Rasmussen, J. (1990). The ecology of human-machine systems II: Mediating
direct perception in complex work domains.
Ecological Psychology, 2, 207-250.
Vicente, K. J., & Rasmussen, J. (1992). Ecological
interface design: Theoretical foundations. IEEE
Transactions on Systems, Man and Cybernetics,
22, 589-596.

Kohn, L. T, Corrigan, J. M., & Donaldson, M. S.

(1999). To err is human: Building a safer health

487

Ergonomic User Interface Design in Computerized Medical Equipment

KEY TERMS
Cognitive Task Analysis (CTA): A family of
methods and tools for understanding the mental
processes central to observable behavior, especially those cognitive processes fundamental to
task performance in complex settings. Methods
used in CTA may include knowledge elicitation
(the process of obtaining information through
in-depth interviews and by other means) and
knowledge representation (the process of concisely
displaying data, depicting relationships, etc.).
Ecological Interface Design (EID): A conceptual framework for designing human-machine
interfaces for complex systems such as computerbased medical equipment. The primary goal of
EID is to aid operators, especially knowledge
workers, in handling novel or unanticipated
situations.
Ergonomics: A synonym for human-factors
engineering, especially in the European literature.

Human-Factors Engineering (HFE): The

branch of engineering devoted to the study of
the interactions between humans and systems,
especially complex systems, with the goal of
designing systems that are safe, comfortable,
effective, and easy to use.
Patient-Controlled Analgesia (PCA): A
computer-based medical technology used to treat
severe pain via the self-administration of analgesic
agents such as morphine.
Skills-Rules-Knowledge Model: A multilevel model of cognitive control developed by
Rasmussen that is especially helpful in examining
human error and critical incidents.
User Interface: The components of a system
that the operator uses to interact with that system.
For example, in the case of a computer, the operator interacts with it using the monitor, keyboard,
mouse, and so forth.

This work was previously published in Handbook of Research on Informatics in Healthcare and Biomedicine, edited by A.
A. Lazakidou, pp. 390-395, copyright 2006 by Information Science Reference, formerly known as Idea Group Reference (an
imprint of IGI Global).

488

489

Chapter 2.9

Development Of E-Government
Services For Cultural Heritage:
Examining the Key Dimensions
Carugati Andrea
Catholic University of Lille, France
Hadzilias Elias
Catholic University of Lille, France

Abstract
This article is aimed at defining a framework for
the design of e-government services on cultural
heritage. Starting from an analysis of three cases
on digitization of different types of cultural objects, we highlight the problems existing in the
creation of e-services on cultural heritage. These
cases show the existence of four key issues in the
development of this kind of information systems:
digitization, requirement engineering, standardization, and interoperability. The proposed framework addresses these issues, focusing on the user
requirements on one side, and the cultural object
representationwhich is the key to interoperabilityon the other. In the cultural domain, the EU
Lisbon strategy pushes for the compatibility of
shared content across multiple, locally generated
contents. Dynamic content exchange requires the
use of a prescriptive framework for the develop-

ment of cultural heritage Web sites. This article

provides such a framework, using observation
from concrete applications, knowledge of information systems development methodologies, and
the IDEF0 modelling method.

Introduction
The Lisbon strategy for eEurope (EU Report,
2002) and the following eEurope 2002, eEurope
2005, eEurope 2010, drafted as results of the
activities of the European Council, are aimed to
make the European Union the most competitive
and dynamic knowledge-based economy with
improved economy and social cohesion by 2010.
In concrete terms, this means broadband and
high-level Internet-based services for the entire
population of the European Union. The means
envisioned to achieve this goal are largely based

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Development Of E-Government Services For Cultural Heritage

on increasing both demand and offer of e-services

respectively, from the public/users and the providers. The problem has been framed as a chicken
and egg problem and the solution has therefore
been to address both ends: increase governmentside services and create a friendly legislation for
the implementation and sale of broadband connections (EU Report, 2002). This article focuses
on the demand side, that is, on the development
of the public electronic services.
On the demand side, electronic government
initiatives involve providing services in e-government, e-learning, e-health, and e-business (EU
Report, 2002). While the efforts of e-government
are focusing on providing services to citizens in
order to achieve higher efficiencies through automation (tax filing, certification, electronic voting,
information provision, and so forth), one other
important area of investment regards the facilitation of access to cultural resources. The regional
and local cultural heritage (to be defined in a broad
sense, from museums to regional gastronomy and
folklore) is one of Europes greatest economic
assets, and ICT and other advanced technologies
can dramatically increase the possibility of its
exploitation. Until now, the government initiatives for divulging electronic material on the local
cultural heritage have been varied in nature and
include the creation of portals for information on
cultural events, which is the most common model
of exploitation today, including the digitization of
artwork for archives, the creation of virtual tridimensional museum visits with tri-dimensional
digitization of art works, and the rendering of
archaeological visits in digital formats (Carugati,
Hadzilias, & Demoulin, 2005).
Nevertheless, the potential of using electronic
services for cultural heritage applications is far
from being fully exploited and many initiatives
have remained at the stage of pilot projects. Of
these pilot projects, few are completed and most
contain only one or very few examples of art
digitization. Until now, experiences of use of ICT
in cultural heritage sectors too often fail in provid-

490

ing valuable economic results due to a number

of problems, and have generated disappointment
among the potential players and beneficiaries. The
main problems have been:

In general, at regional and local level, there

is shortage of experience and expertise about
the use of ICT in cultural heritage areas.
Therefore, local and regional administrators
have to rely on ICT vendors and consultants,
and these in general, are mainly able to
suggest general purpose solutions, nonoptimised for the specific sector, since even
large IT consulting companies have limited
expertise in the cultural heritage area.
If we consider the conventional cultural
heritage, such as museums and galleries,
this sector lacks expertise and experience
in marketing and business promotion on
electronic media, which makes it difficult to
develop credible business models and plans,
and to attract investments.
There are analogous problems also in the
cultural tourism sector. There are hundreds
of projects and initiatives related to cultural
tourism in Europe, but they often have been
designed on supply oriented thinking,
without systematic investigation into what
the customer, the cultural tourist, is looking for. The issue of user diversity should
be considered in the system development
process, as pointed out by Klawe and Shneiderman (2005), since it is a critical success factor to steer user preferences. This
is valid not only for business purposes, but
also and foremost at the community level
(Carroll, 2001). Finally, as e-govenrment
services are web-based, they are available
to different users. Even though the expectations, in terms of services, might differ,
Nielsen (2005) states that usability factors
influence, for both professionals and other
categories, the success of the service.

Development Of E-Government Services For Cultural Heritage

More generally, the problem of mutual

understanding between the cultural sector,
service industry, and ICT experts, due to their
different ways of thinking, makes it difficult
to achieve a fruitful cooperation in areas
ranging between education and multimedia
production, where the availability of content
from the cultural heritage could open very
promising development perspectives. In order
to be able to develop innovative and marketable products and services with a real cultural
added value, the personal direct involvement
of experts from the cultural area is absolutely
required.

Despite the above-mentioned issues, in Europe as well as elsewhere, there are a number of
interesting examples and initiatives, on various
scales, of successful economic promotion of the
cultural heritage (some of these mentioned above).
Unfortunately, they have often only a local or regional visibility, and their positive (and negative)
experiences cannot be fully exploited and shared
by other projects.
To resolve these issues, the article proposes
an integrated framework for developing egovernment services on cultural heritage. The
framework emerges from the study of multiple
cultural heritage electronic initiatives and the in
depth investigation of three case studies. This
framework is represented using the activity modelling method IDEF0, where all necessary steps,
inputs, outputs, rules, and roles are described in
a hierarchical manner.
The article is organized in the following way:
first, the research methodology is explained,
secondly, we present the case studies and analyse
them to highlight key issues, we describe the
IDEF0 modelling technique, and then we continue
presenting the development framework, focusing
on user segmentation and interoperability. Finally,
we conclude with a discussion of the proposed
framework, highlighting the need for a systemic
approach to the activities.

Research Methodology
The research was carried out following the constructive paradigm of the case study approach
(Yin, 2002). In this context, the research used
a practice-based lens to identify processes and
problems in the development of e-government
services on cultural heritage. The process was
comprised of three steps:
1.
2.

3.

Identify representative cases of e-government cultural services;

Identify the different functionalities for the
expected users and the problems incurred in
these specific cases of information systems
development;
Propose a way to consider these problems
given the constraint of limited resources.

As a first step, we carried out an extensive

survey to record most electronic initiatives in
the culture domain. The survey identified 142
initiatives, which should cover the bulk of such
services for the eight languages1 spoken by the
authors. The languages cover the vast majority
of projects in this domain. Among these initiatives, we have selected three that are the most
representative because they fit specific criteria
that we set out in our research: 1) they continue
to exist after the end of the pilot project; 2) they
are complete in terms of the cultural content they
were supposed to cover; 3) they provide additional
functionalities with respect to similar services;
4) they use more advanced technologies; and
5) the informative material about the case was
comprehensive.
As a second step, we analyse the three cases in
a structured manner looking at a) functionalities
provided to the users; b) functionalities provided
to other institutions; and c) the development
process. These three elements are compared and
integrated to indicate key leverage points of these
initiatives. In this step, we include usability testing for the two initiatives available online (the

491

Development Of E-Government Services For Cultural Heritage

Piero della Francesca Project and the incunabula

manuscripts in the Herzog August library) in order
to see if the users find the functions of the Web
site friendly and value-adding. For doing this, we
carry out the testing with five users, representing
different categories (Nielsen and Landauer, 1993).
The layout of the testing follows the indication
of Molich (1999, p. 109), with a testing room and
an observation room. According to the testing
protocol, the users were given different tasks to
carry out and were required to speak over their
actions.
In the third step, we employ the IDEF0 modelling method to describe a high level design framework for implementing e-government services
for cultural heritage. The framework expands on
treating the key points identified in step two.
The proposed solution is therefore the result
of deductive reasoning and is prescriptive in
nature.

Case Studies
According to the selection process, we consider
the following case studies: the Electronic Beowulf
Project at the British Library (UK), the Piero
della Francesca Project in Arezzo (Italy), and the
digitization of the incunabula manuscripts in the
Herzog August Library in Cologne (Germany).

Beowulf: The Pioneer Project

The Electronic Beowulf Project represents a
pioneering effort in the digitization of cultural
heritage. Beowulf is both the first English literary
masterpiece and one of the earliest European epics written in the vernacular, or native language,
instead of literary Latin. The story survives in
one fragile manuscript copied by two scribes near
the end of the 10th or the first quarter of the 11th
century. Until quite recently, most scholars thought
that this surprisingly complex and poignant poem
was written in the 8th century or earlier, but Ki-

492

ernan (1981) stirred up controversy asserting that

the work was composed in the 11th century, and
that the manuscript itself may have even been
the authors working copy. The manuscript has
been damaged over time and the original that is
surviving today is in very poor condition ( Why
is Beowulf important?, 2005).
The Electronic Beowulf Project was carried
out in the period 1993-1999 in order to digitize the
original Beowulf manuscript, of which the original
is now in the British Library, and has been made
accessible to everyone in electronic format.

Development Process
Since the Beowulf represents one of the first attempts to digitize a cultural heritage object, many
difficulties were encountered in different aspects
of the project, as the work in preparing the digital
edition was very complex in the period considered. Technical difficulties concerned scanning
technologies, storing media, and managing the
transfer across large distances of large quantities
of data or moving 24-bit colour images across
different technical platforms. In some cases, it
was necessary to devise innovative technical
solutions to achieve the desired end result. The
following is a quote from Prof. Kiernan2s Web
site, highlighting the technical difficulties encountered in 1993:
The equipment we are using to capture the images is the Roche/Kontron ProgRes 3012 digital
camera, which can scan any text, from a letter or
a word to an entire page, at 2000 x 3000 pixels in
24-bit color. The resulting images at this maximum resolution are enormous, about 21-25 MB,
and tax the capabilities of the biggest machines.
Three or four imagesthree or four letters or
words if that is what we are scanningwill fill
up an 88 MB hard disk, and we have found that
no single image of this size can be processed in
real time without at least 64 MB of RAM. In our
first experiments in June with the camera and its
dedicated hardware, we transmitted a half-dozen

Development Of E-Government Services For Cultural Heritage

images by phone line from the Conservation

Studio of the British Library to the Wenner Gren
Imaging Laboratory at the University of Kentucky,
where identical hardware was set up to receive
the data. Most of these images are now available
on the Internet through anonymous ftp or Mosaic
(Electronic Beowulf: A Guide, n.d.)
Besides the technical challenges, there were
also issues in the coordination and collaboration
of the human resources, due to the variety of
disciplines necessary to carry out this task. The
domain experts comprised scholars, curators, conservators, photographers, and technical experts
(Prescott, 1997), who had to agree on how the
activities had to be carried out in order to record
the maximum information without damaging the
precious document.
In the case of Beowulf, the development was
mainly oriented towards the utilization of information technology for capturing cultural content.
This activity was not only done to preserve the
artifact, but also to enhance the comprehension
of the text. From the multiple sources about this
project, we can see that, given these requirements,
the main activity was to create digital images of
the text with different scanning techniques (B&W,
24 bit color, ultraviolet, and high level fiber optic
light). This was performed once in order to minimize damage and retain the maximum information
digitized because of the extensive erasure and
correction in the original manuscript.
The development process followed was a pure
waterfall approach (Boehm, 1988). In the situation, the requirements had to be set at the offset
of the project and the digitization was done only
once per page. Afterwards the content was used
to populate a database.

Functionalities for Users

The Electronic Beowulf Project is available to
users on CD format that can be purchased from
the British Library. The interface provides access
to the digital content of a database that contains

both pictures and transcribed text. The CD contains multiple transcription of the Beowulf from
different accredited authors. Different search
criteria are available to query Line (Edition), Folio
(Edition), FolioLine (Transcript), Fitt (Edition or
Transcript), Scribe (Edition or Transcript). Images
can be zoomed to a very high detail in order to be
studied by paleographers and calligraphers (see
Figure 1). Alternatively, the layout of the screen
can show both the original manuscript and the
transcription, which is intended to be used by
researchers interested in the content of the poem
(see Figure 2).
The electronic version enables readers to place
the original manuscripts leaves side by side, to
examine the color and texture of the vellum leaves
by magnifying the images and to explore the work
of the early scribes (see Figure 3). Building on
this material, the CD features multiple versions of
restoration of which the most recent one was made
possible by the different scanning techniques.

Functionalities for Institutions

This project was part of the strategy of the British
Library to create the capability to

increase access
to its collections by use of imaging and network
technology. As such, the Beowulf project was
not directly geared towards providing content,
services, or functionalities to other institutions
and remains a stand alone product.
One of the collaborations that emerged from
this project was the inclusion of the Thorkelins
transcript of the Beowulf catalogued in the Royal
Library of Copenhagen.

Piero della Francesca: From

Restorers to Children
The project of Piero della Francesca3 regards the
digitization of the series of frescos Leggenda
della Vera Croce situated in the Bacci chapel in
the Basilica of St. Francis of Arezzo (Italy). The
project was started in the mid 80s to repair the

493

Development Of E-Government Services For Cultural Heritage

Figure 1. Zoom function

Figure 2. Texts comparison

Figure 3. Scanning under multiple light sources

severe damages that the frescos suffered in the

early 80s. The project was lead by the cultural
and artistic heritage authority of Arezzo (Soprintendenza per i Beni Architettonici, Artistici
e Storici di Arezzo). The causes of the damage
were not clear and therefore the Soprintendenza
decided to begin a series of deep and comprehensive scientific and historic research studies to
determine the reasons for the frescos detrimental
conditions. The aim was the preservation of the
masterpiece for future generations, and in order
to achieve this, it was necessary to find the right
technique to restore the painted plaster as well as
the building structure itself.
As the project evolved
and the restoration
work finished, the initial goal was complemented
by the creation of a permanent documentation
center for future study and diffusion of information regarding the study and conservation of the

494

mural paintings of Piero della Francesca and to

construct, with the use of computer technology,
an organic database for the Project and an interactive multimedia system to be hooked up to the
international electronic network and available for
public consultation.3 The project is considered
one of the most advanced and demanding efforts
of the entire scientific restoration field in Italy
and abroad.

Development Process
Originally, this project was carried out to provide
professional restorers with new tools for restoring
the artworks of Piero della Francesca. For this
reason, the initial requirements were set by the
restorers and as a result, the quality and detail of
the digitized material is very high. For example,
they integrated text, vectorial information (CAD

Development Of E-Government Services For Cultural Heritage

drawing), and raster information (images) which,

from a data processing point of view, are of a
completely different nature and very complicated. Figure 4 shows the process followed by the
project team from preliminary study, to material
collection and data processing, and finally the
development of software applications.
The four basic activities in this project: data
capture, CAD drawing, and raster images (overlaid) are presented in Figure 5.
As in the Beowulf case, the development
approach followed the waterfall model. Restorers, together with state institutions, and private
and public research institutes, carried out a long
preliminary research to set out the requirements.
Leading scholars and expert technicians used state
of the art information technology to develop the
applications that were used to successfully define
the restoration approaches to solve the fresco
complex problems.
After the end of the restoration work the development continued with the creation of a Web

site that used the same collected material from the

professional applications, but provided functionalities that were interesting for the wider public.
In this second cycle of development, a new set of
requirements was created to address the goal of
creating a permanent documentation center.

Functionalities for Users

For the restorers, the developed applications provided functionalities to store and
search
thousands
of documents; illustrations of the church choir
were compiled in a basic cartographic model and
in a variety of thematic mappings with reference to
historical studies, visual surveys, diagnostic studies, and scientific analyses and complemented by
about 1000 historical/chronological records. The
catalogued images could be called up individually
from a special menu where they were arranged
according to distinct categories. This technology
was particularly effective in the phases of correlation, control, and updating of the various studies

Figure 4. Activities in the project of Piero della Francesca

Figure 5. Data capturing and sample images in the project of Piero della Francesca

495

Development Of E-Government Services For Cultural Heritage

conducted. The images could also be called up

from a relational database, which allowed the
efficient consultation of records.
Furthermore, the software allowed keeping
record of the restoration in process (work site
diary, graphs and charts of the work performed,
technical specifications, and so forth) as well as
adding new cultural content.
To help restorers in their work, the applications operate in the complete respect of the
iconographic and technical value of the artist
s
work, providing a documentation of the various
painting techniques employed by the artist: from
the transfer of the preparatory design through
the use of stencilled cartoons, etched lines and
colored threads, to the application of color to the
fresh plaster (a fresco), partially dried plaster (a
mezzo fresco), and finally to the dried plaster with
special temperas and organic blending agents. A
further example of the system

s capabilities is that
it provides a chronological reconstruction of the
artist

s production.
Connected to this, the system
allows the study of the relationship between the
saline deposits and the presence of fixatives added
during earlier restorations or, alternatively, to view
the deposits in relation to the various painting
techniques used by the artist.

To satisfy the need for rapid consultation in

the service of monitoring the state of conservation and the progress of the restoration, system
designers developed a thematic glossary with a
unified terminology of key words for research
and access to the data, organized first in separate
technical files and then electronically translated
to permit interactive consultation.
Finally, the applications were online with the
work site, in order to manage the traditional paper
documentation kept by the project director and
the restoration technicians and allowing them
to have immediate access to and use of the data
stored in the system.
In the public Web site, one of the services
provided is the classic function of archive. The
archive allows searching for historical events by
date or by key word and presents the results combining digital images with metadata. The result
of one search is presented in Figure 6.
Other interesting functionalities provided
are games based on the thematic

mappings
. The
Web site proposes two games: one that consists
of adding colours to a black and white painting
and the other where parts of the paintings have
been divided (electronically) into pieces that the
player is supposed to reassemble like a puzzle.

Figure 6. Public archive from the Web site http://www.pierodellafrancesca.it

496

Development Of E-Government Services For Cultural Heritage

These two games have an underlying pedagogic

reason and would not be possible without the
existence of the digitized material specifically
created and mapped for the restorers. The Web
technologies deployed for this purpose were active
images and java applets. An example of the two
games is presented in Figures 7 and 8.

Usability Testing
The usability testing was carried out by five
people: two laymen, one professional restorer,
and two children aged seven and eight years old.
The adults got the task of finding the painting of
La Madonna del Parto, starting from the home
page. The children had the task of solving the
two games. All subjects reported poor navigation

between the different pages due to non-existent

text explanation of the links and unconventional
graphics and structure. Both the laymen and the
professional restorer found the painting in approximately three minutes and using 30/32 mouse clicks
(against one minute and eight clicks that it would
take for an experienced user of the system). The
information presented pleased the professional,
but not the laymen that reported it as too complex
and dull. All three cases showed a critical design
error when the user had to pass from the classic
Web site view to the database of the art works,
because we had to intervene in order to guide the
users on how to continue their search.
In consideration of their age, the children
were not asked to navigate through the Web site
and were presented directly with the games. The

Figure 7. Puzzle game

Figure 8. Painting game

497

Development Of E-Government Services For Cultural Heritage

children did not understand what they had to do

and also complained that there was no evaluation of their performance. Using the words of the
children, these games were reported to be: boring
and meaningless.
Despite the interesting use of the digitized
material for a wider audience, the outcomes were
not value-adding, neither for the laymen nor for
the children.

Functionalities for Institutions

The system provides two levels of access to the
project material. At the first level, the information is available on the Web site that is hosted by
the sponsor of the project (Banca Etruria)and
at the second level, which is more complete and
contains the full record of the restoration project,
the system is available at the offices of the cultural
and artistic heritage authority of Arezzo. This
local system can be accessed only under official
authorization. Data can be processed through a
cartographic system based on photogrammetric
surveys of the painted walls. The relational database is structured in a series of indexes. Files
are divided into didactic, diagnostic, and intervention sections. The local system is addressed
to institutions that seek in depth knowledge, not
only of the artworks, but also of the restoration
process, the mineralogical analysis, and the chemi-

Figure 9. System functionalities

498

cal analysis performed. The left side of Figure 9

shows the local cartographic system and the right
frame shows the functionalities for professionals.
The green dot indicates the access to the public
archive via the Web.

Distributed Digital Incunabula

Library: An Architecture for Sharing
The Distributed Digital Incunabula Library
(inkunabeln.ub.uni-koeln.de/vdib/) is a project
aimed at the digitization of all German incunabula. Incunabula, Latin for cradle, are books
printed between 1454 and 1500. Worldwide,
approximately 550,000 incunabula have been
preserved. The Distributed Digital Incunabula
Library project was organized as a joint effort of
the Koelner Universitaets und Stadtbibliothek
and the Herzog August Bibliothek Wolfenbttel
libraries. The two libraries involved, hold together
approximately 5,800 copies incunabula. Out of
these, 350,000 pages (approximately 2,000 3,000
titles) were selected for this project, of which 40
% were digitized in the first ten months of the two
year effort. When completed, the 350,000 pages
could represent as much as six percent of the titles
of incunabula preserved world wide (Loebbekke
and Thaller, 2005).
As a means of achieving global integration
while simultaneously keeping institutional in-

Development Of E-Government Services For Cultural Heritage

dependence, this project proposed as reference

architecture a new tool called Digital Autonomous Cultural Objects (DACOs). DACOs are
characterized by a common behavioral code
that provides information about cultural objects
on request. The code involves the object wrapping itself in mark-up language. It also provides
a persistent addressing scheme, allowing for
control of the individual institution holding the
good as well as a mapping scheme. Thus, even
if the technology (e.g., HTML, XML, SGML,
VRML objects, Flash movies, SVG constructions) can be chosen by the individual institution,
the structure of basic elements integrated into a
DACO is predetermined.

Development Process
In order to deal with the vast amount information
to be digitized, a standard development process
was followed, which is depicted in Figure 10.
The first stage of the development was the
raw digitization of the single pages, which was
conducted using a Nikon DXM 1200 and a camera
server that stored the images with total capacity
of 80 GB. A book cradle for photos in an angle
of 45 degrees (see Figure 11) allowed treating the
incunabula carefully while simultaneously saving

Figure 10. Development process and system

landscape of Incunabula Library

time and money. The pages were scanned in a 24bit color scheme using a resolution of 3800x3072
pixel leading to a data file size of 34 MB/page.
With such a digitization procedure, more than 80%
of incunabula were digitized leading to about 12
TB raw data volume.
At the second step, the digitized data was
transferred through a local area network to a preprocessing server with 250 GB storage capacity.
This server was responsible for the transformation
of the output files to the size that is efficient for
Web site access and for the different users.
Finally, the files were forwarded to the Web
server environment and the digitized files were
available on the Internet, after passing a quality
control at the Metadata Cache and being transformed to DACOs. These versions of the original
files were not more than 10% of the original data
size, so that they can be accessed quickly from the
Internet. During this phase, the data supervision
and the production of the copy-referred development data took place. The archive server offers a
storage area for disaster recovery purposes.
The entire architecture was based on cost-efficient Linux servers highly performing for the
hierarchical storage organization.
Obviously, beyond tackling the technical challenges, cost played a major role in pushing such

Figure 11. Book cradle for 45 degrees photos

Camera Server
80 GB
(Raw Digitization)

Pre-processing Server
250 GB
(File Downsizing)

Metadata Cache
250 GB
(Quality Control)

Webserver
1 TB
(WWW Interface)

Archive Server
1000 TB
(Long-term backup)

499

Development Of E-Government Services For Cultural Heritage

a digitization effort forward. The most important

cost factors were:

Content and subject classification based on

the already existing Incunabula Short Title
Catalogue (ISTC);
Workflow/data importation;
Raw digitization;
Storage capacity;
Development and maintenance of the WWW
server.

Considering all cost components in this project,

the costs for one page amount to roughly 0.75
(plus metadata and work environment) (Loebbecke and Thaller, 2005).

Functionalities for Users

The users are able to run queries in the incunabula
resources in two levels: text-based and imagebased. In the former, there are query criteria for
locating the page or pages under consideration,
containing all metadata and the digitized images.
In the latter, there is an image archive with three
sublevels of incunabula description: the Incu-

nabula Short Title Catalogue (ISTC) classification,

the original image, and the transcription of the
image. Furthermore, additional tools are provided
allowing for browsing through the manuscripts
like in traditional libraries, as shown in Figure 12.
A native dynamic XML database administers
all codicological manuscript descriptions. For
some codices, those descriptions amount to the
equivalent of fifty printed pages.
There are four different viewing levels of the
scanned images suitable for the categories of users
who will study the cultural object. Figure 12 shows
the query result with the overview level (notice
the eye, the glasses, the loop, and the microscope
icons above the scanned image). Figure 13 shows
a higher level of detail.
Another e-service offered is that the users
can download several tools, which measure the
performance of typical types of codicological/
palaeographic work (e.g., measurement of single
characters). As far as available without copyright
restrictions, the main scientific literature to work
with the codices is also made available in digital
form.
Using DACOs, one URL of the library dissolves into about 600,000 URLs of individual

Figure 12. Query results

( http://inkunabeln.ub.uni-koeln.de/ visited on 25/7/05)

500

Development Of E-Government Services For Cultural Heritage

references to a specific cultural heritage object

that can be referenced directly from any other
cultural heritage resource (see, for example,
the code ig00227000 above the scanned page in
Figure 12). This is due to the various views of the
metadata and different resolutions.

Usability Testing
Given the content of this site, we had to carry
out a careful selection of the five users. Finally,
we selected three librarians and two history researchers. Even though the test subjects did not
research specifically the incunabula, they were
selected for their knowledge in the research and
management of historical documents. They were
given two tasks: to find the document
Mensa
philosophica and the complete production of
Gerson. The users completed the first task in less
than two minutes and with 10/14 mouse clicks
(compared to an experienced user that requires
seven mouse clicks). They completed the second
task, restarting from the main page, with seven
mouse clicks in less than one minute. The users
reported the site to be easy to navigate, even
though it was in German language. However,
they evidenced two major failures in the database
operation: the field tables were not linked and
the navigation functions in the index, instead

of visualizing the records alphabetically, were

providing erratic results.

Functionalities for Institutions

The highest added value of this project lies in
the inter-institutional, cultural content sharing.
Cultural heritage brokers, connected to servers
recognizing DACOs, integrate many DACOs from
different institutions into specialized interfaces,
thus serving the interests of laymen and researchers. The communication between the servers
takes place via the DACO protocol consisting of
several XML-codes transmitted via Hyper Text
Transfer Protocol (HTTP). When addressed by a
broker, the DACO server describes itself through
providing a list of supported access venues (e.g.,
titles, authors, years, and so forth). The access
venues can be used by the broker to develop a
search mask.

Case Analysis
After examining thoroughly the three case studies of e-services in the cultural sector and with
the knowledge of the other existing initiatives,
we observed that the greatest challenge lies on
the management of four dimensions: digitization,

Figure 13. Multiple levels of details for the users

501

Development Of E-Government Services For Cultural Heritage

requirement engineering, standardization, and

interoperability.
The survey we carried out shows that several
European countries are making efforts to comply
with the Lisbon strategy and are moving, albeit at
different paces, towards the digitization of their
cultural heritages. However, given the sheer size
of the material, the variety of the applications
developed, and approaches taken by the different
governments, it appears clear that the seamless
integration of the cultural heritages of multiple
countries is a very complex task and it is addressed
in multiple ways. The three cases showed that
digitization is the starting point of these efforts,
aiming at the preservation of the cultural content
and its availability to a large spectrum of users.
All three cases show a very similar approach to
the digitization problem, using various scanning
technologies to create digital facsimile of cultural
objects.
The second dimension concerns the requirements engineering. In the three cases observed,

the requirements were set initially by the most demanding groups, namely the professionals working with cultural objects in their daily working
life. Researchers were in contact with the technical
developers in order to discuss and define the initial
set of requirements. This helped to set very high
quality standard for what later became material
available to laymen. The process of creating the
Web sites available to the public was carried out
following a top down approach that provided the
digitized content to the public without examining
the specific needs of the different categories of
users. A notable exception is the functionality
offered in the Piero della Francesca project where
the pedagogic games addressed the needs of the
specific target group of children. However, this
effort followed the same push strategy as the other
two and our usability tests proved the limitation
of this approach.
As far as standardization is concerned, there
seems to be a trend towards finding a common way

Table 1. Overview of the findings

Digitization

Piero della Francesca

(1991)

Incunabula
(2001)

Flat documents partly

damaged with different light sources.
Major issue: technology.

Three and two dimensional objects: data

capture, CAD drawing,
and raster images.

Flat documents in
different resolutions. Major issue:
cost.

Requirements
engineering

Top-down approach
from professionals
perspective

Top-down approach
from professionals
perspective

Top-down approach from

professionals
perspective with
special focus on
interoperability.

Standardization

Ad-hoc approach.
Standards applied to
the image format.

Ad-hoc approach using standard rules for

relational database and
graphic file formats.

Use of XML
standard. DACO
typology candidate
for becoming a
European standard.

Not addressed. Available on CD-ROM.

The archive is partially

available online, but
the content is not
accessible from other
Web sites.

Content shared
with other institutions thanks to
specially designed
XML objects
(DACO).

Interoperability

502

Beowulf (1993)

Development Of E-Government Services For Cultural Heritage

to represent cultural objects. In the early cases, the

standardization regarded the format of the digitized images (e.g., jpg, tiff, bmp formats). Instead,
the incunabula project shows a more structured
approach to the standardization problem, using
the DACOs where, not only the images, but also
the metadata, follow a specific standard.
Finally, we have also observed an evolution towards higher interoperability of cultural
information systems. The early example of the
Beowulf was designed as a stand alone project
and it is only available to users on CD-ROM to
be bought from the British Library. The Piero
della Francesca project started as an independent
solution for professionals, but it was later made
available on the Web with some of the original
functionalities activated. The incunabula project
wasfrom the offsetborn as a Web-based information system to execute queries on the pages
of digital incunabula. This last effort is directly
geared towards making the material and the tools
available to other cultural Web sites. In order to
increase the interoperability of existing or future
applications, there should be distributed, localized
development of content that can be integrated as
it is created. The summary of the case analysis
is shown in Table 1.
In an era where content exchange is becoming the norm, and access to cultural content is
mandate by law (eEurope strategy), it becomes
paramount for the different governments to develop their content, respecting a framework that
will allow different Web sites to freely exchange
content. The framework will have to address the
four dimensions just described and embed them
into a process of information system development.
The development of such a framework will be the
focus of the next part of the article, beginning
with the description of the modelling technique
usedIDEF0.

IDEF0: the modelling

technique
In order to model our framework, we selected a
modelling technique that was suitable for easily
conveying a message and, at the same time, did
not impose methods or techniques on the system
modelled. Among the possible techniques (e.g.,
Data Flow Diagram, ARIS, UML, and so forth),
we have selected IDEF0 for its simplicity, well
knowing that experts easily translate it into any
other modelling tool.
The IDEF0 modelling technique is intended
to model the decisions, actions, and activities of
a system. It is not only the most widely used, but
also the most field proven function modelling
method for analysing and communicating the
functional perspective of a system (Malhotra
and Jayaraman, 1992). IDEF0 was derived from
a well-established graphical language, the Structured Analysis and Design TechniqueSADT
(Marca and McGowan, 1988). The IDEF0 modelling method establishes the scope of analysis
either for a particular functional analysis or for
future analyses from another perspective. As a
communication tool, IDEF0 enhances domain
expert involvement and consensus decision-making through simplified graphical devices. As an
analysis tool, IDEF0 supports the identification
of the functions performed and what is needed
to perform them.
The basic activity element of an IDEF0 model
diagram is represented by a simple syntax. A verbbased label placed in a box describes each activity.
Inputs (I) are shown as arrows entering the left
side of the activity box, while the outputs (O) are
shown as exiting arrows on the right side of the box.
Controls (O) are displayed as arrows entering the
top of the box and mechanisms (M) are displayed as
arrows entering from the bottom of the box. Inputs,
Controls, Outputs and Mechanisms (ICOMs) are all
referred to as concept. An IDEF0 model diagram is

503

Development Of E-Government Services For Cultural Heritage

then composed of several activity boxes and related

concepts to capture the overall activity. IDEF0 does
not only capture the individual activities, but also
reveals the relationships among activities through
the activities related concepts. For example, the
output of one activity may become the input,
control, or even a mechanism of another activity
within the same model.
A strategy for organising the development of
IDEF0 models is the notion of hierarchical decomposition of activities (see Figure 14). A box in
an IDEF0 model represents the boundaries of an

activity. Inside that box is the breakdown of that

activity into smaller activities (child diagrams),
which together comprise the box at the higher
level (the parent). This hierarchical structure helps
the analyst keep the scope of the model within
the boundaries represented by the decomposition of the activity. This organisation strategy is
also useful for hiding unnecessary complexity
from view until a more in-depth understanding
is required.
One of the most important features of IDEF0 as
a modelling concept is that it gradually introduces

Figure 14. Hierarchical decomposition of an IDEF0 model diagram

0
A0

A-0

M ore G eneral
1
2

M ore Detailed

3
4
A4
A0

Thi s box is the parent of

this diagram.

1
2
A42
3
A4

1
2
3
A42

504

NOT E: Node numbers shown

here indicate that the box has
been detailed. The C-number
or page number of the chi ld
diagram could have been used
instead of the node number.

Development Of E-Government Services For Cultural Heritage

greater and greater levels of detail, through the

diagram structure comprising the model. In this
way, communication is enhanced by providing the
reader with a well-bounded topic with a manageable amount of detail to learn from each diagram.
The decomposition is graphically indicated with
codes; the highest level is called A0, and the activities which are included are indicated with A1,
A2, A3, and so forth. (see Figure 14).
An IDEF0 model starts by presenting the whole
subject as a single unita box with external-arrow
boundary conditions connecting it to functions
and resources outside the subject. The single
box is called the top box of the model or the
context diagram. Since the single top box of an
IDEF0 model represents the subject as a whole,
the descriptive name in the box is general. The
same is true of the external arrows of the model,
since they represent the complete set of external
boundary conditions of the subject as a whole,
including access to mechanism support that supplies additional means of performance.

4.

According to these four ideas and the four

dimensions identified in the analysis, we developed a framework which contains the following
five phases: collection of user requirements,
digitization of cultural content, system interoperability design, cultural content organization, and
e-government cultural portal development. These
phases are expressed as activities in the IDEF0
model of Figure 15.
The mechanisms that we have identified are:

Description of the Proposed

Framework
Based on the analysis of the cases, the exhaustive survey of the cultural heritage initiatives,
and our theoretical and practical knowledge on
information systems development, we propose a
framework for the development of e-services for
cultural heritage based on four ideas:
1.

2.

3.

E-services for cultural heritage are, in

essence, information systems. As such,
existing methodologies can be adopted and
adapted for their development.
IDEF0 is a proven modelling technique for
the development activities of information
systems.
The impact of Web sites for cultural heritage
increases with the enhancement of content
integration from multiple sites: a one-stop
cultural heritage site can be imagined as the

ideal goal and therefore compatibility of the

finished local products is paramount.
Multiple types of public have to be served by
these sites; satisfying each group is important to serve the goal of increasing demand
for services in the eEurope strategy.

Users. This mechanism includes art specialists (like the restorers in the Piero della
Francesca case), historians, researchers, and
the public in general. In Figure 15, it is shown
that the mechanism users (M1) supports
the collection of users requirements (A1),
the organization of cultural content (A4),
and the development of e-service (A5). The
participation of users in these activities ensures the user-centered evaluation that the
usability testing proved essential to avoid
critical errors.
Technical project team. It includes the project
manager, the system architects, the system
analysts, the database designers, and the
developers, including the Web designers.
In figure 15, it is shown that the mechanism
technical project team (M2) supports the
collection of users requirements (A1), the
digitization of cultural content (A2), the
design of system interoperability (A3), and
the development of e-service (A5). The
participation of the technical project team
in these activities ensures that the users
needs can be supported by the technology
and the resources available.

505

Development Of E-Government Services For Cultural Heritage

Figure 15. IDEF0 model of the proposed framework Level A0

x

DATE: 26/03/06
REV:

AUTHOR:
PROJECT:

USED AT:

WORKING
DRAFT
RECOMMENDED
PUBLICATION

NOTES: 1 2 3 4 5 6 7 8 9 10

testing
protocol
C2

I1

cultural
heritage
e-service
plan

user
segmentation
C1

Collect
Users
Requirements
A1

MINERVA-Digicult
best practices

System
Specifications

C3

ISO 21127
standard
C4

DATE CONTEXT:

READER

ICOM-CIDOC
standards
C5

ISO 9126
standards
C6

Digital
fax-simile

Digitize
Cultural
ContentA2

System
Specifications
System

Specifications
Design
System
Interoperability
A3

Organize
Cultural
ContentA4
Develop
e-service
System
Specifications

users M1
NODE:

A0

technical
project team
TITLE:

M2

O1

A5

cultural
M3
organization

cultural heritage e-service development

The cultural organization. It includes all

the experts in the specific art works that
are part of the cultural heritage e-service
(curators, conservation scientists, archivists,
museum technicians). This mechanism is
discrete from the users because they provide information about the art works and
are not necessarily using the service. The
participation of the cultural organization in
these activities ensures that the art work will
be treated in the proper way, that the needs
of other cultural organizations, in terms
of exchange of the digitized materials, are
upheld, and that the metadata is correct and
complete.

The specific characteristics of each activity are

presented in detail in the following sections.

506

cultural
heritage
e-service

NUMBER:

P. 2

Collection of User Requirements

This activity is aimed at the creation of the requirements document for the development of the
e-service. To design an efficient cultural heritage
e-service that meets the public needs, it is first
of all necessary to define the target group users
and understand the requirements that have to be
addressed. The survey of the cultural heritage
initiatives highlighted different groups of users.
One classification (see Dipartimento dei Beni
Culturali, Italy) identified users as internal, private (customer market), or corporate (business
market). For example, the internal public might
be the law enforcement agencies accessing the
artwork database for research on fakes or stolen
material. The differentiation between private
and corporate is quite unusual in a governmental
service, since every citizen should be equal in
front of the state. This might be the indication of
a trend to adopt models from e-commerce into

Development Of E-Government Services For Cultural Heritage

e-government, or it might be the result of using

technology providers used to dealing with the
private sector instead of with the government. The
three cases analysed above show that segmentation
is deemed important by the creators of cultural
heritage e-services.
The input of this activity is a plan for the
creation of an e-service for cultural heritage. The
plan may originate from a specific need like in
the case of Piero della Francesca, from a research
interest like in the Beowulf case, or from a long
term plan for the distribution of content like in
the incunabula project.
User segmentation (Control C1 in Figure 15)
guides how the requirements will be created.
The users can be classified according to their
profession, location, affiliation, age, and interests
(Sen, Padmanabhan and Tuzhilin, 1998). Region
and language can be used to make sure that the
service provided is culturally acceptable and
the language is understandable by the audience.
Among demographical variables, age may be
useful to present cultural heritage accurately at
every group age4. Occupation or community of
practice of the user is a strong determinant of the
requirements for the e-service.
As these services are rather new, it is difficult
to fathom what would be the requirements of the
different groups. Therefore, it is important to
develop prototypes for the requirement definition. The user-centered evaluation, following the
testing protocols of Nielsen and Landauer (1993)
and Molich (1999), should be carried out to assure
that the technical project team has well interpreted
the user requirements. The testing protocol is
represented by the control C2 in Figure 15.
If the information is missing because the
service is new than the cases indicate that it is
convenient to serve the most demanding group
first, like the restorers in the Piero della Francesca
case, and then reduce the complexity as other user
segments are identified.

The groups involved in this activity are the

e-service project team (the technical developers),
the cultural organization that owns or manages the
art works, and the users in their different declinations. The users are necessary because they will
be able to state what the needs are. Users should
be representative of the categories identified in
the segmentation exercise. The technical team is
necessary because of the preparation for the technical work ahead. They have to make sure that the
requirements can be technically satisfied. Finally
the cultural organization is also necessary because
they know how the art works can be handled and
what are the aspects of their digitization that are
most valuable to cover. As an example, the case
of the Beowulf showed that it was necessary to do
the scanning under UV light and only specialists
could point that out.
Despite a careful segmentation, the actual
collection of the requirements can result in a
quite challenging process. Most constituencies,
especially layman, will in fact be unaware of what
they want, since innovative e-services are still very
limited. While the mental model of a museum or
an archaeological visit is shared among people of
different countries, there might be requirements
that no user can state, simply because one cannot ask what is not known. For the elicitation of
requirements from the final users, it would then
seem appropriate to engage with the different constituencies in an iterative process using prototypes
of interfaces. This prototype-based requirement
elicitation should be done autonomously from
the setting of the requirements for the content
digitization, where the use of an agreed upon
standard is required.
The output of this activity is the requirement
document. For the technical project team, it will
contain the functional specification of the e-service to be developed. For the cultural organization,
it will contain the specification for the preparation
of the art works and the specifications for the
descriptions to be prepared.

507

Development Of E-Government Services For Cultural Heritage

Digitization of Cultural Content

This is the activity of digitization of cultural
content. The reasons for digitization, or more
precisely for the digital conversion of non-digital
cultural source material, are varied and may well
overlap. The decision to digitize has the following objectives:

To increase access: this is the most obvious

and primary reason, where a high demand
from users is estimated and the administration has the desire to improve public access
to a specific collection.
To improve services to an expanding users
group by providing enhanced access to the
institutions resources with respect to education and long life learning.
To reduce the handling and use of fragile or
heavily used original material and create a
backup copy for endangered material such
as brittle books or documents.
To give the administration opportunities for
the development of its technical infrastructure and staff skill capacity.
To develop collaborative resources, sharing
partnerships with other administrations
to create virtual collections and increase
worldwide access of a countrys heritage.
To seek partnerships with other administrations to capitalize on the economic advantages of a shared approach.
To take advantage of financial opportunities, for example the likelihood of securing
funding to implement a program, or of a
particular project being able to generate
significant income.

According to which of these needs are deemed

more important, the requirement document is
taken from the previous activity and transformed
into two outputs: the digital facsimile and a new
systems specification document revised after
the digitization. The technical project team and

508

the cultural organization collaborate in this activity which, as seen in all three cases, is very
demanding.
The control is provided by the MINERVA
guidelines. MINERVA, MInisterial NEtwoRk for
Valorising Activities in digitization (http://www.
minervaeurope.org/), is a resource for guidelines
concerning the digitization of cultural content.
MINERVA was created by a network of European
Union (EU) Member States Ministries to discuss,
correlate, and harmonize activities carried out in
digitization of cultural and scientific content for
creating an agreed European common platform,
recommendations, and guidelines about digitization, metadata, long-term accessibility, and
preservation. Due to the high level of commitment
assured by the involvement of EU governments,
it aims to coordinate national programs, and its
approach is strongly based on the principle of
embeddedness in national digitization activities.
The use of the MINERVA guidelines insures the
contacts with other European countries, international organizations, associations, networks,
international and national projects involved in
this sector.
The output of this activity is the digitized
content. This content can exist independently
from the creation of the e-service that uses it (like
a repository) or will be used in the development
activity. The other output is the revised version of
the specification document in accordance to the
problems and opportunities that the digitization
phase might bring to the surface.

System Interoperability Design

This activity is aimed at assuring the compatibility
of the content generated across e-services. This
activity takes as input the systems specifications
produced during digitization and adjusts them
for interoperability.
For this activity, we propose as control the
adoption of the ISO 21127 standard (http://www.
iso.org/). ISO 21127 is a domain ontology for

Development Of E-Government Services For Cultural Heritage

cultural heritage information. As such, it is designed to be explanatory and extensible, rather

than prescriptive and restrictive. Currently, no
specific formalism for semantic models has been
widely accepted as standard, which is proven by
the existence of the DACO used in the incunabula
case. However, the semantic deviations between
the various available models are minimal. At this
stage, we prefer to recommend the use of ISO
21127 instead of the DACO as control, since the
DACOs are not recognised as a standard yet.
The ISO standard has been formulated as an
object-oriented semantic model, which can easily
be converted into other object-oriented models.
This presentation format will be both natural
and expressive for domain experts, and easily
converted to other machine readable formats, such
as RDF (Resource Description Framework) and
XML. Considerable effort has gone into achieving
these goals; all cross-references and inheritance
of properties, for example, are explicitly resolved.
This has led to a certain degree of redundancy, but
makes the document more readily comprehensible
and facilitates its use as a reference document.
The standard is intended to cover all concepts
relevant to cultural heritage information, but most
particularly, those needed for wide area data exchange between museums, libraries, and archives.
Due to the diversity of the subject matter and the
nature of cultural heritage collections, this goal
can be achieved only by extensions to the standard.
However, thanks to its object-oriented nature, the
ontology offers a backbone of powerful general
concepts, which have a wide application.
The primary purpose of ISO 21127 is to offer a
conceptual basis for the mediation of information
between cultural heritage organizations such as
museums, libraries, and archives. The standard
aims to provide a common reference point against
which divergent and incompatible sources of
information can be compared and, ultimately,
harmonized.
The mechanisms involved are the cultural
organization and the technical project team. The

reason why this cooperation is required is that this

activity is both technicalit will shape the technical development of the e-servicebut is also
highly connected to the conservation practices.
Only the experts in the management of the cultural
content can, in fact, specify how and what parts
of the content would need to be shared.
The output of this activity is the systems specification document, enriched with the information
about interoperability.

Cultural Content Organization

In this activity, the content is organized. In particular, the cultural organization (the mechanism)
takes care of creating the descriptive data about
the digitized material: the metadata. The description is based on the specification document and
takes into consideration the interoperability needs
specified in the previous activities.
The specific activities that we have identified
in our survey of existing e-services are the following:

Primary data registration and description

of cultural objects. This data refers to basic
identification information, classification,
physical shape, condition, geographical
location, construction date, application,
properties, and relation with other objects.
Collection of the administrative information concerning the relevant collection and
monument management. This information
is the prerequisite for the management and
documentation of monuments or objects,
which are part of a museum collection.
Documentation
information

of

digital

products
and

processes

. The

meta-data that
concern the processes of digitization, identification, quality and thematic content of the
digitized material belong to this category.
Preservation information of the cultural
content. This information refers to meta-

509

Development Of E-Government Services For Cultural Heritage

data for the long-term preservation of the

digitized material.
Publishing data of the cultural content. The
digitized material is published on the Internet
and is also stored in optical technology media
(CD-ROMs, DVDs). The data included in
this category aims at the educational use
of the cultural digitized material and the
effective creation of multilingual versions.

Considering these activities, and the specification document we recommend, as control, the
use of the ICOM/CIDOC standards (Abraham
& Means, 2001; Aquarelle, 1999; Baskerville &
Pries-Heje, 2004). The ICOM/CIDOC standards
are the output of a joint effort across museums,
archaeological sites, and other cultural initiatives to create internationally accepted guidelines for museum object information. Since the
ICOM/CIDOC standards have been agreed by
the cultural organizations, their use should be
straightforward.
The output of this activity is the final requirement document containing functional, structural,
and process specifications.

E-Service Development
Once the requirements have been collected, the
material digitized, and the standards set, the time
for the actual e-service development arrives. We
envision the e-service to be, as a starting point,
of the kind presented in the incunabula project:
a complete, widely accessible, shared content of
precious cultural material.
The e-service development can be regarded as
a problem of information system development. In
the particular case of cultural heritage e-services,
there are particular conditions that make them
different in relation to the needs of traditional
Web sites treated in the literature (Baskerville and
Pries-Heje, 2004). The needs for quick adaptability
and flexibility (ibid) do not apply in most cultural
heritage cases where the data is stable and the

510

user interfaces are also quite stable in time. The

presentation of the artworks is also quite stable in
time and it has achieved an agreed upon format.
Furthermore, the digitization might involve very
fragile artworks, besides being very complex and
time consuming, and therefore, it cannot be envisaged that the basic data coming from digitization
can be recreated multiple times.
Therefore, the development should not focus
on rapid and continuous development, but rather,
comply with the more traditional system development lifecycle, focusing on long term usage:
efficiency, reliability, low cost maintenance, and
robust functionalities (Truex, Baskerville, and
Klein, 1999). However, given that most e-services
will be Web-based, some of the characteristics
of Web sites can be used advantageously in
cultural heritage services. The interface must be
particularly cured as visitors are varied in age
and education; therefore, particular attention to
usability must be taken. Furthermore, standards
must be strictly respected as presented previously,
since data might and should be used in multiple
sites or in multiple ways in the same site. Being
prepared to provide multiple views of the same
digitized material will be a critical factor for the
success of cultural heritage e-services over time
because the development of views and interfaces
will demand much less work compared to the
digitization of content. Finally, particular attention
should be posed to scalability of e-services, given
the staggering quantity of material potentially
available for display.
Our case and our survey show, for most eservices, the first user group to be served are
professionals. Serving professionalswith high
requirements on the quality of the digitization
and on the metadataare a very good starting
point because, for the reasons connect to the
aforementioned material, it is easier to simplify
a high-quality digitization, rather than re-digitize
to add information.
Therefore, we envision at least a double
loop development where, in a first instance, the

Development Of E-Government Services For Cultural Heritage

cultural organizations and the professionals are

served. Once professional e-services are created,
then a second loop of development can be carried out where other services are added for the
wider public. The first loop uses as mechanism
the technical team, the cultural organization, and
the professional users. The second loop uses the
technical team and representatives of the public
as mechanisms in the first activity.
The control proposed for this activity is the
ISO 9126 standard, which is concerned primarily
with the definition of quality characteristics to
be used in the evaluation of software products.
Using this standard as control should assure the
quality of the e-service.
The output of this activity is one or multiple
e-services.

still be able to provide indications for what to do,

following which rules, who has to do what, and
what input has to be used.
The order in which the activities are actually carried out, during the project of e-service
development, should be decided contextually by
the actors involved. Our research has pointed out
the importance of user involvement, where users
are not only professionals, but also and foremost,
laymen. Appealing to the wider laymen category
will, in fact, ensure the long term interest in
the e-service. Therefore, the use of a process
that contains iterations based on user testing is
highly recommended and the framework will
guide the choice for inputs, outputs, controls,
and mechanisms for each activity in the multiple
iterations.

Reflection on the framework

and ISD methodologiES

Conclusion

At first sight, the framework resembles a variation

of the waterfall methodology for information system development. This impression is undoubtedly
strengthened by the fact that the e-services for
cultural heritage are indeed information systems.
However, the framework proposed in Figure 15
does not propose a waterfall-like development,
but rather, shows the activities involved with their
inputs, controls, mechanisms, and outputs. The
framework does not specify the sequence of the
activities, even though visually it might seem so.
The fact that the activities are depicted in Figure
15 from the top left to the bottom right follows
simply a tradition in IDEF0 modelling.
In fact, in real cases, the activities might go
on in parallel or even be carried out in different
moments. Typically such an e-service can use
material digitized long before the project is started
or, even more commonly, can use descriptions
(metadata) created before the project is started.
The framework (as it does not specify a sequence)
can be adopted in any ISD methodology and will

The study of the existing initiatives of e-government on cultural heritage shows that the different
member states are pursuing the Lisbon strategy
for eEurope, but in different ways and at different
paces. While some countries are more advanced
than others, even for the leading ones, it is difficult to speak about concerted action to create
a one-stop site for cultural heritage being this at
subject level (archaeology, graphical art, sculptures, and so forth), local level, national level, or
international level. While cooperation obviously
exists among museums, ministries, universities,
and other institutions, this cooperation only very
recently has begun to be addressed seriously.
While cultural organizations, research institutions, and standard organizations have begun to
create usable standards, one of the main issues that
we have observed in an initial survey is the lack
of a methodology that systematically integrates
and adopts these standards.
The in depth examination of three case studies
of e-services has pointed out four dimensions that
are important to address when developing these

511

Development Of E-Government Services For Cultural Heritage

services: digitization, requirements engineering,

standardization, and interoperability.
These dimensions have been linked to the
existing standards to create a development framework, which addresses both practical issues and
future needs.
In our view, the problem of advancement in the
cataloguing and divulgation of electronic material does not reside in technology, but in organizational traditions and cultures that, knowingly
or unknowingly, might be retarding the process,
and in the recognition that electronic cataloguing and presentation of artwork is not locally
limited (e.g., at museum or site). Politicians and
administrators need to understand that to achieve
the goals of eEurope, cultural heritage needs to
move from the physical place to the electronic
space. We believe that this article provides a solid
framework to achieve this goal. The proposed
framework is easy to understand: based on IDEF0,
clear in its content and in the definition of what
is needed, who is needed, what rules apply, and
what are the goals for each activity. It provides,
to the willing and open-minded team, the right
guidance (without being restrictive) towards the
development of e-services for cultural heritage
that are not only effective upon completion for
a specific location, but that will stay effective in
the future and without borders, if the standards
for interoperability are respected.
This framework should be further examined
in an action research environment where not only
the framework should be put into use but the
organizational conditions that characterize this
type of projects could be studied, as well. This
will allow us not only to validate the framework
but also to study the condition under which the
cultural heritage e-service plan (the primary
input of the framework) is actually made.

References
Abramson, M. A., & Means, G. E. (2001). EGovernment 2001. IBM Center for the Business
of Government Series, Lanham, MD: Rowman
and Littlefield.
Aquarelle (1999). Sharing Cultural Heritage
through Multimedia Telematics. DGIII, ESPRIT
Project 01/01/1996 - 31/12/1998.
Baskerville, R., & Pries-Heje, J. (2004).
Short
cycle time systems development. Information
Systems Journal, 14(3), 237-265.
Carroll, J. M. (2001). Community computing
as humancomputer interaction. Behaviour &
Information Technology, 20(5), 307-314.
Coursey, D., & Killingsworth, J. (2000). Managing government web services in Florida: Issues
and lessons. In D. Garson (Ed.), Handbook of
public information systems. New York: Marcel
Dekker.
EU Report (2002). eEurope 2005 executive summary. Retrieved from http://europa.eu.int/information_society/eeurope/2005/index_en.htm
Fountain, J. (2001). Building the virtual state:
Information technology and institutional change.
Washington: Brookings Institution.
Fountain, J. (2003). Electronic government and
electronic civics. In B.Wellman (Ed.), Encyclopedia of community. Berkshire: Great Barrington,
436 441.
Ho, A. T.-K. (2002). Reinventing local governments and the e-government initiative. Public
Administration Review, 62(4), 434445.
Kiernan, K. S. (1981). Beowulf and the Beowulf
manuscript. New Brunswick, NJ: Rutgers University Press.
Klawe, M. and Shneiderman, B. (2005). Crisis and
opportunity in computer science. Communications of the ACM, November, 48(11), 27-28.

512

Development Of E-Government Services For Cultural Heritage

Loebbecke, C., & Thaller, M. (2005, May). Preserving Europes cultural heritage in the digital
world. Proceedings of the European Conference
on Information Systems (ECIS). Regensburg,
Germany.
Malhotra, R., & Jayaraman, S. (1992). An integrated framework for enterprise modeling. Journal of
Manufacturing Systems, 11(6), 426-441.
Marca, D. A. & McGowan, C. L. (1988). SADT:
Structured analysis and design technique. New
York: McGraw-Hill.
Molich, R. (1999). Bruger-venlige edb-systemer.
(in Danish) Copenhagen: Teknisk Forlag.
Moon, M. J. (2002). The evolution of e-government
among municipalities: Reality or rhetoric? Public
Administration Review, 62(4), 424433.
Nielsen, J., & Landauer, T. K. (1993). A mathematical model of the finding of usability problems. In
Proceedings of ACM INTERCHI93 Conference
(pp. 206-213). Amsterdam, The Netherlands.

Truex, D. P., Baskerville, R., & Klein, H. K. (1999).

Growing systems in emergent organizations.
Communications of the ACM, 42(8), 117-123.
Yin, R. K. (2002). Case study research, design and
methods (3rd ed.). Newbury Park, CA: Sage.
West, D. M. (2004). E-government and the transformation of service delivery and citizen attitudes.
Public Administration Review, 64(1), 1527.

Web Sites Referenced (all Web sites

visited on November 14, 2004):
Direzione Generale per i Beni Archeologici Italy:
Virtual visit of the Villa Adriana http://www.
archeologia.beniculturali.it/pages/atlantearcheo/1_VAD/pano/Atla_Pano_VAD_08.html
Uffizi Gallery, Florence, Italy: http://www.uffizi.
firenze.it
The Louvre Museum, Paris, France: http://www.
louvre.fr/

Nielsen, J. (2005). B-to-b users want sites with

b-to-c service, ease. B to B, 90(7), 48-48.

National Museum, Stockholm, Sweden: http://

webart.nationalmuseum.se/

Prescott, A. (1997). The electronic Beowulf and

digital restoration. Literary and Linguistic Computing, 12, 185-195.

Project Piero della Francesca: http://www.pierodellafrancesca.it

Sen, S., Padmanabhan, B., & Tuzhilin, A. (1998).

The identification and satisfaction of consumer
analysis-driven information needs of marketers
on the WWW. European Journal of Marketing,
32 (7/8), 688-702.
Shi, Y., & Scavo, C. (2000). Citizen participation and
direct democracy through computer networking. In
D. Garson (Ed.), Handbook of Public Information
Systems, New York: Marcel Dekker.
Thomas, J. C., & Streib, G. (2003). The

new face of
government: Citizen-initiated contacts in the era of
e-government. Journal of Public Administration
Research and Theory, 13(1), 83101.

Aquarelle Project, http://vcg.isti.cnr.it/projects/

miscellanea/aquarelle/3Dvisual.presentation.
html

Ministries and Departments of

Culture Visited (in alphabetical
order: Web sites visited on
November 14, 2004):
DenmarkKulturministeriet: http://www.
kulturministeriet.dk/sw16739.asp (press release
about the law on conservation of cultural heritage
through digitization in Denmark)

513

Development Of E-Government Services For Cultural Heritage

FranceMinistre de la Culture et de la Communication: http://www.culture.fr/ (specific site

dedicated to cultural activities)
GreeceHellenic Ministry of Culture: http://
www.culture.gr/
IrelandThe Department of Arts, Sport and
Tourism: http://www.arts-sport-tourism.gov.ie/
ItalyMinistero per i Beni e le Attivit Culturali:
http://www.beniculturali.it/
SpainMinisterio de Educacin y Ciencia:
http://wwwn.mec.es/index.html
United KingdomDepartment for Culture,
Media, and Sport: http://www.culture.gov.uk/
default.htm

References for the Cultural Content

Part:
nternational guidelines for museum object information: The CIDOC information categories,
ICOM/CIDOC, Paris, 1995. ISBN 92- 9012124-6.
International Core Data Standard for Ethnology/
Ethnography, ICOM/CIDOC, Paris, 1996, ISBN
960- 214-012-7. http://cidoc.icom.museum/ethst0.
htm
Draft International Core Data Standard for
Archaeological Sites and Monuments, ICOM/
CIDOC, Paris, 1995, ISBN 92-9012-125-4. http://
cidoc.icom.museum/arch0.htm
Metadata Standards for Museum Cataloguing
http://www.chin.gc.ca/English/Standards/metadata_description.html
MIDAS: A manual and Data Standard for Monument Inventories. http://www.english-heritage.
org.uk/Filestore/nmr/standards/Midas3rdReprint.pdf

514

OpenGIS Reference Model (Kurt Buehler, ed.),

Open GIS Consortium Inc., Ref. No. OGC 03040 Ver. 0.1.2 http://www.opengeospatial.org/
specs/?page=orm
SPECTRUM: The UK Museum Documentation
Standard, 2nd edition, Museum Documentation
Association, Cambridge, United Kingdom http://
www.mda.org.uk/spectrum.htm
MPEG7 multimedia content description, ISO/
IEC TR 15938-8:2002 http://archive.dstc.edu.
au/mpeg7-ddl /
DIG35 Metadata Specification Version 1.1, Digital Imaging Group http://www.i3a.org/i_dig35.
html
Text Encoding Initiative (TEI) http://www.teic.org/
Encoded Archival Description (EAD) Version
2002, Society of American Archivists and the
Library of Congress. http://www.loc.gov/ead/
Digital Preservation Coalition Handbook http://
www.dpconline.org/graphics/handbook/
A Metadata Framework to Support the Preservation of Digital Objects, OCLC/RLG Working
Group on Preservation metadata. http://www.
oclc.org/research/projects/pmwg/pm_framework.
pdf
IEEE Final 1484.12.1-2002 LOM Draft Standard
for Learning Object Metadata http://ltsc.ieee.
org/wg12/20020612-Final-LOM-Draft.html
Sharable Content Object Reference Model (SCORM), http://www.adlnet.org/index.
cfm?fuseaction=SCORMDown
Translation memory exchange format (TMX),
OSCAR recommendation, Localisation Industry
Standards Association, 2004. http://www.lisa.
org/tmx/tmx.htm

Development Of E-Government Services For Cultural Heritage

Endnotes

English, Italian, Greek, French, Danish,

Swedish, German, Spanish
Prof. Kiernan, University of Kentucky,
is considered the highest authority in the
Beowulf manuscript and he was the main
figure in the Electronic Beowulf Project. His

websitehttp://www.uky.edu/~kiernan/eBeowulf/guide.htmprovides very extensive

material about the project.
Information coming from http://www.pierodellafrancesca.it
In the Piero della Francesca project, we
highlighted the existence of the games.
This is an example of customization of the
content to a young audience.

This work was previously published in the International Journal of Technology and Human Interaction, Vol. 3, Issue 2, edited
by B. C. Stahl, pp. 45-70, copyright 2007 by IGI Publishing, formerly known as Idea Group Publishing (an imprint of IGI
Global).

515

516

Chapter 2.10

An Activity-Oriented Approach
to Designing a User Interface
for Digital Television
Shang Hwa Hsu
National Chiao Tung University, Taiwan
Ming-Hui Weng
National Chiao Tung University, Taiwan
Cha-Hoang Lee
National Chiao Tung University, Taiwan

Abstract
This chapter proposes an activity-oriented approach to digital television (DTV) user interface
design. Our approach addresses DTV usefulness
and usability issues and entails two phases. A
user activity analysis is conducted in phase one,
and activities and their social/cultural context are
identified. DTV service functions are then conceived to support user activities and their context.
DTV service usefulness can be ensured as a result.
The user interface design considers both activity
requirements and user requirements such as users
related product experience, mental model, and
preferences in phase two. Consequently, DTV us-

ability is achieved. A DTV user interface concept

is thus proposed. The interface design concept
contains the following design features: activityoriented user interface flow, remote control for
universal access, shallow menu hierarchy, display
management, adaptive information presentation,
and context sensitive functions. Usability evaluation results indicate that the user interface is easy
to use to all participants.

Introduction
DTV has several advantages over conventional
analogue television: better picture and sound quality, more channels, interactivity, and accessibility.

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

An Activity-Oriented Approach to Designing a User Interface for Digital Television

Many countries plan to switch to DTV within the

next 5 to 10 years. To facilitate DTV user adoption,
Hsu (2005) conducted a survey exploring factors
driving DTV diffusion. Identified factors were:
government support, reliable technology, price,
DTV service usefulness, and easy-to-use user
interface. Service usefulness and easy-to-use user
interface were among the most important driving
factors determining user adoption.
Identifying useful services and designing an
easy-to-use user interface is challenging. First,
DTV can be complicated for most users. After
switching from analogue to digital, television is
no longer just a standalone audio-video device.
DTV is becoming central to the digital home
by providing information services; controlling
home appliances and security devices; supporting
communication activities; performing business
transactions; storing video images; and so on.
Therefore, users manipulation of numerous service applications and easily accessing what they
want is critical. Secondly, nearly everyone may
use DTV. However, different users may use DTV
services for their own purposes and have different
usage patterns. Moreover, users interacting with
the system have different experiences and abilities. Therefore, accommodating diversified user
needs and skill levels is also important.
The user-centered design (Norman & Draper,
1986) has been a dominant approach to user interface design in meeting usability requirements. The
user-centered design approach essentially focuses
on and includes the user in the design process,
beginning with user identification and user tasks.
Users target-market characteristics are identified
according to user analysis. These characteristics
include: demographic characteristics, knowledge
and skills, limitations, attitude, and preferences.
Based on these characteristics, product requirements can be determined. Task analysis identifies
user needs in task performance. These needs include functions and information as well as control,
enabling users to perform tasks. Task analysis
results are used to specify functional and use

requirements. The user interface considers compatibility between users information processing
limitations and task demands, to ensure usability.
Since the user-centered approach recognizes
that the design will not be right the first time, it
suggests that an iterative design be incorporated
into the product development process. By using
a product prototype, the design can be refined
according to user feedback. The user-centered
design approach has been successfully applied
to many computer products.
However, the user-centered approach has
some drawbacks and limitations. Norman (2005)
warned against using it in designing everyday
products. He argued that the user-centered approach provides a limited design view. It is suitable
for products targeted for a particular market and for
specific user-task support, but everyday products
are designed for everyone and support a variety
of tasks. These tasks are typically coordinated
and integrated into higher -level activity units.
For everyday product design, Norman proposed
an activity-oriented design approach. He asserted
that a higher-level activity focus enables designers
to take a broader view, yielding an activity supportive design. In a nutshell, the activity-oriented
approach focuses on user activity understanding,
and its design fits activity requirements.
In line with Normans (2005) notion of activity-centered design, Kuutti (1996) proposed
an activity theory application framework to the
human-computer interaction design (HCI). According to activity theory (Engestrm, 1987), an
activity is the way a subject acts toward an object.
An activity may vary as the object changes, and the
relationship between the subject and object is tool
mediated. Tools enable the subject to transform an
object into an outcome. Furthermore, an activity
is conducted in an environment that has social and
cultural context. Two new relationships (subjectcommunity and community-object) were added
to the subject-object model. The community is a
shareholder group in a particular activity or those
who share the same activity objective. Rules and

517

An Activity-Oriented Approach to Designing a User Interface for Digital Television

regulations govern how an activity is carried out

in a community and impose environmental influences and conditions on an activity. Furthermore,
community members assume different activity
roles and responsibilities. These components
provide a social contextual framework for HCI
design. Therefore, activity theory not only focuses
on user activity analysis, but also considers cooperative work and social relationship interactions
(Kuutti, 1996). That is, activity theory treats
HCI as a complete interaction perspective, not
limited to interaction between an individual and
a computer.
Although several researchers (e.g., Bdker,
1996; Nardi, 1996) have attempted to apply activity theory to HCI design, there is no formal
method for putting activity theory into practice.
To resolve this problem, Mwanza (2002) proposed
a systematic design framework called activityoriented design method (AODM), based on the
activity triangle model (Engestrm, 1987). The
method contains eight steps aimed at successively
collecting information regarding eight activity
components: Subjects, objects, tool, rules, community, division of labor, transformation process
and outcomes. Mwanza successfully applied
activity theory to two projects by employing this
eight-step model.
DTV is intended for individual use in daily
life activities. A life activity can include many
others (family members, friends, or others) besides the individual, and may include many social
processes. Therefore, DTV can be a social device
shared by family and friends alike. DTV design
should consider social and cultural contexts.
Furthermore, life activities are dynamic. They are
composed of task sets that change as conditions
change. Thus, the design should reflect possible
actions and conditions in which people function
(Norman, 2005).
This chapter proposes an activity-oriented
approach to DTV user interface design. This
approach begins with an understanding of life
activities and their requirements in use context.

518

Activity theory is used as a framework for life

activity understanding. Life activities are treated
as analysis units. Analysis on use context includes
interaction among community members underlying the activity, rules governing the activity,
and physical, as well as sociocultural aspects
of the use environment. From activity analysis,
activity requirements are determined and, in
turn, functional requirements can be specified.
After activity requirements are determined, the
approach moves to identifying user requirements
such as users experiences, skills, knowledge,
and preferences. User interface design tries to
optimize compatibility between an activity and
its contextual elements so that it not only meets
activity requirements, but also satisfies individual
as well as social needs. In so doing, DTV usefulness and usability can be ensured.
The proposed activity-oriented approach
consists of four steps:

Analyze DTV activity requirements.

People engage in an activity in order to
attain an objective. The objective is then
transformed into an outcome that motivates
people to engage in the activity. Accordingly,
the activity can be identified by the motive
for using DTV. Since the identified activities
may be too numerous, related activities are
grouped into activity groups for simplicity.
The intertwined relationships among activity components in each activity group
are analyzed. All members and their roles
in the underlying activity are identified.
Moreover, rules governing DTV use and the
social processes underlying the activity are
analyzed. Social and cultural issues related
to DTV use are identified. Activity models
are then constructed to depict relationships
between the eight activity components.
Envision services to support each activity
group. By analyzing activity requirements
and its social context, appropriate applica-

An Activity-Oriented Approach to Designing a User Interface for Digital Television

tion services and contents are envisioned.

The envisioning process is based on activity
models constructed in the previous stage.
Conceptual scenarios developed describe
how the conceived services support user
activity. A DTV system conceptual model
is then developed describing system services, their attributes, and the relationships
between services. The DTV system can then
smoothly meet activity requirements and
prove useful to users as a result.
Establish a use model. A use model delineates how users will use DTV services to
carry out their activity. Use model development begins with detailed activity analysis.
An activity has a hierarchical structure containing a coordinated, integrated set of actions. Actions are targeted toward a specific
goal and are achieved through operations.
Conditions will control what operations will
be taken, and conditions and goals together
will determine tool use. Based on the activity-actions-operations hierarchy and activity
flow pattern, a use model is then developed
to describe the mapping between DTV
services and activity flow. This mapping is
used as a basis to determine user interface
structure and components.
Understand user requirements. To achieve
usability goal, user interface design must
consider user characteristics. User characteristic exploration includes two parts: (1)
understanding users previous experiences,
and (2) exploring users mental models and
preferences for interaction methods. Users
accumulate previous experience from earlier
product interaction. Previous experience
contains users knowledge, skills, mental
models, and affective responses to the related
products. Leveraging users previous related
product experiences can ease users new
product learning and prevent user frustration. Furthermore, users interact with the
system based on their system knowledge

that is, their mental models. Designing

a user interface compatible with a users
mental model enhances product usability.
Interaction method preference affects users
product attitude. User interface design must
meet these user requirements to enhance
usability and increase user acceptance.

Case Study: A DTV User

Interface Design
Several qualitative and quantitative methods
employed in this study applied the activity-oriented approach. Qualitative methods include
ethnographic research, in-depth interviews, and
a focus group. These qualitative methods analyze
activities and their social/cultural context and conceive DTV services. Quantitative methods were
also conducted, assessing activity importance
and collecting user characteristic information.
A survey was conducted to meet the purpose.
The survey comprised of questions regarding
user activities; services and contents needs; user
experiences with related products; user expectations of user interface style; and user characteristics such as demographic variables and attitudes
toward technology adoption. The one thousand
questionnaire-sampling schemedistributed
across Taiwanproportionately represented the
Taiwanese population in terms of age, sex, occupation, and geographical location. Of the returned
questionnaires, 960 were valid.

Understanding User Activities and

Their Context
Ethnographic research conducted observed current life activity practices. Twelve lead users
(Rogers, 1995) participated in the research. Observations focused on activities engaged in; when
and where they were engaged in; how the activity
was carried out; who took part in the activity and
their roles; and what rules governed this activity.

519

An Activity-Oriented Approach to Designing a User Interface for Digital Television

Observers tried to piece information together

using the activity triangle model (Engestrm,
1987). In some cases, observation results were
insufficient to provide a whole activity system
picture. Thus, in-depth interviews conducted
collected supplementary information. Through
ethnographic research and in-depth interviews,
34 life activities were identified and their activity
models obtained.
These activities then served as a candidate
activity list in the survey questionnaire. Survey
respondents were asked to rate the importance and
frequency of activity performance. Exploratory
factor analysis conducted explored the interrelationships of 37 activities. A Kaiser-Meyer-Olkin
(KMO) test of sampling adequacy obtained a good
value of 0.911, the Bartletts test of Sphericity was
significant (p <0.001), and the matrix determinant
was in acceptable range. Factor analysis results
determined five activity groups (refer to appendix
A): (1) entertainment (9 activities), (2) information (6 activities), (3) education (9 activities), (4)
transactions (5 activities), and (5) daily living (5
activities).

Identifying DTV Service Needs to

Support each Activity Group
A focus group meeting met to envision potential
DTV services and contents to support each activity group. The focus group consisted of two DTV
R&D engineers and six lead users with DTV

experience in other countries. The meeting began

with a briefing on the latest DTV technology and
service developments by the two R&D engineers.
Next, lead users were given the activity model obtained in the earlier stage and asked to brainstorm
potential DTV services that would be useful in
supporting each activity group. Thirty-seven candidate DTV services and contents were gathered
as a result. After identifying potential services,
conceptual scenarios were developed, describing
service use in carrying out activities.
These 37 candidate DTV services were then
included in the survey questionnaire. Respondents in the survey rated the services/contents
in terms of their usefulness in supporting life
activities. Useful services for each activity group
are shown in Table 1. Some services/contents
were considered useful in supporting more than
one activity group.

Developing a DTV use Model

The second ethnographic research conducted
collected detailed activity analysis information to determine identified service and content
functional requirements in each activity group.
Five families were recruited for the ethnographic
research. These families were lead users of 3C
products. The research took place on one weekday and one weekend. Observations were made
to learn how life activities are performed in the
family context and how each family member

Table 1. Useful services for each activity group

Entertainment Information Education

Video-on-demand
Music-on-demand
Games
Travel
Horoscope
TV shopping
Ticket Purchase
News-on-demand
Voting
Transportation

520

Meteorology
Transportation
Travel
Financial
Employment
Horoscope
TV banking
Movie
News-on-demand
Games

Employment
e-Library
Games
e-Learning
e-Newspaper

Transactions

Financial services
TV banking
TV shopping
Reserving
Ticket Purchase
Lottery
e-Health management
Commodity

Daily living

Appliance control
Voting
e-Health care
Video phones
Lottery
Home security
e-Health management
Commodity

An Activity-Oriented Approach to Designing a User Interface for Digital Television

took part in the activities. Observation results

gathered decomposed activities into action and
operation levels. Moreover, in-depth interviews
requested interviewees to envision how they would
use DTV services and contents to support their
activities. Use models were then developed based
on interview results. These use models served as
a basis for determining user interface components
and its structure.
Ethnographic research results indicated that
each activity group had its own specific use
model. For example, entertainment activities are
performed in a sequential manner. After users
find the right program, they will watch the program for a while before they move onto the next
activity. Therefore, a sequential menu structure
is appropriate to support this type of activity. On
the contrary, an information-searching activity
is an iterative process. Users navigate through
information space in order to seek for related
information. Users may easily get lost and find
it difficult to return to their original location.
Therefore, a network structure is suitable for this
type of activity.

Analyzing User Requirements

Understanding User Experience
Survey respondents rated past experience with
related products in terms of ease of use and
frequency of use. The ease-of-use score reflects
product usability and the frequency-of-use score
indicates activity support usefulness. Combining
the two scores yielded a good experience index
for each product. Results showed that nine multimedia products, four information products, and
two communication products were rated as good
experience products. Among these products,
respondents with high interaction skill rated television sets, TV games, and personal computer as
the top three good experience products, whereas

television sets and TV games were regarded as

good experience products by respondents with
low interaction skill.

Identifying UI Design Features

In-depth interviews conducted identified design
features that contribute to good experience. Eight
persons participated in the in-depth interview sessions, including two businessmen, two children,
two housewives, and two older adults. They were
frequent users of good experience products.
In the interviews, participants were asked to describe their interaction experience with TV sets,
TV games, and personal computers and then list
design features they considered good.
Good design features for TV sets were: (1)
control function visibilityall control functions
have dedicated physical keys easily located on the
remote control; (2) shallow menu hierarchymost
TV menu options are organized in one or two
levels. Too many menu layers confuse and disorient users so that they have difficulty reaching
the target option; (3) feedback information visibility TV provides immediate selection and
system response feedback. Selection feedback
shows immediate user selection. System response
feedback provides the function whether the user
activates it or not; and (4) automatic setting and
display management provisionTV users do not
need to memorize and learn how to go back to the
initial display management setting. When the user
turns off the TV set, it will automatically return
to the initial setting when it is turned on again.
TV games are popular entertainment for children. Good design features indicated by the interviewees include: (1) comprehensible iconsusers
can easily identify menu option icons in games;
(2) unambiguous function groupinggames
always use a meaningful group function rule so
users can easily locate the function; (3) legible text
displayTV games fit information presentation
into the display space so children can read the

521

An Activity-Oriented Approach to Designing a User Interface for Digital Television

instruction; (4) easy-to-use function keysall

function keys are unambiguously mapped to the
players task; (5) context-sensitive helpwhen
playing a TV game, the player does not need to
refer to the manual. Help information for each
possible problem is always provided to the player
throughout the game play.
PC users pointed out several good design
features of a PC: (1) interactivityPC provides
interactive functions and dynamic feedback allowing users to use appropriate commands to achieve
their task goal; (2) graphical user interfacePC
contains easy-to-understand icons and intuitive
metaphor that makes users feel they are directly
manipulating the device; (3) direct linkage to related informationthe hyperlink mechanism helps
users easily and efficiently locate information
they are interested in; (4) multi-taskingusers
can perform more then one task simultaneously.
PC users wished that DTV would also allow users
multiple viewing of TV programs and services
at the same time.
These good design features provide insight
into DTV design. They can also be used as design
references to ensure interoperability of networked
products.

Exploring Users Mental Models and

Preferences
Respondents with low interaction skills, such
as housewives and older adults, stated their
preference for TV interaction style. In fact, even
respondents with high interaction skill expected
DTV to retain a TV context for entertainment
activities, while good PC design features provide
support for information-related activities. TV design features all respondents would like to retain
on DTV were: (1) TV screen as a default screen,
(2) remote control as the main interaction device,
(3) shallow menu structure, (4) automatic display
settings, and (5) direct access function keys.

522

The User Interface Design

Concept
This study developed a user interface concept
based on both activity and user requirements. The
concept and the features are described next.

Activity-Oriented User Interface Flow

The use model obtained from ethnographic
research served as the design basis for the user
interface structure. In this design, services are
provided in accordance with activity needs. In
addition, services are embedded in the activity so
that users need only focus on the activity without
contemplating how to apply appropriate services
to support the activity. For example, when the user
chooses the activity he/she wants to engage in,
he/she may get related services in the service menu.
Also, when the user uses a service to perform an
activity, the system will automatically display
related functions. In so doing, users cognitive
load can be reduced.

Remote Control for Universal Access

Survey data showed that most respondents preferred to use a remote control as the primary
interaction device. The remote control design is
intended to support users with different interaction
skills. Thus, in our design, the remote control is divided into two areas: the primary control area and
the advanced control area (Figure 1). The primary
control area is similar to the traditional analog
TV remote control. It provides basic function
keys (e.g., number key, volume up/down, channel
up/down, mute, and so on) sufficient for operating
multimedia devices. The primary control area
also provides eight function keys dedicated to
frequently performed activities identified from
the activity analysis. The function keys allow users direct access to critical services or programs
for those activities. The primary control area is
designed mainly for supporting entertainment and

An Activity-Oriented Approach to Designing a User Interface for Digital Television

daily life activities. The advanced control area is

a touch pad serving several functions: (1) points
and selects menu items, (2) inputs text, (3) displays
personalized Electronic Program Guide, and (4)
manages displays. It is suitable for information,
transaction, and education activities.

Shallow Menu Hierarchy

To provide a simple use environment, the menu
function items were arranged based on importance and frequency-of-use for each service. The
sequential menus were limited to three layers,
helping users avoid unnecessary menu traversal.
For simpler operation, menu items can be selected
using arrow keys to move the cursor and then
pressing the enter (ok) key. Moreover, eight hot
keys on the remote control provide direct access
to critical activity group services (Figure 2) so
that menu traversal is minimized.

Figure 1. The remote control

Multiple Viewing and Display

Management
To accommodate single and multiple viewing
behaviors, two display modes (single viewport
and multiple viewport) are provided. The singleview mode is designated as a default mode. When
the DTV is turned on, it displays single-view
mode (Figure 3a). In the single-view mode, if
another service or program is selected, the DTV
will automatically switch to multiple-view mode
(Figure 3b). In the design, the system supports up
to four active viewports at the same time. With
multiple-view mode display, the user can resize
and relocate the viewports at will, or go back to
the initial single-view mode by pressing the cancel
key on the remote control.

Adaptive Information Presentation

In the multiple-view mode, display information
legibility may suffer. To solve this problem,
our study developed an adaptive information
presentation method. The method contains two
mechanisms: menu legibility adjustment and
content legibility. Menu legibility affects how
many menu options can be displayed. The adaptive
information presentation method automatically
adjusts the numbers of legible options for viewport
size to ensure menu option legibility. Content
legibility affects how much information can be
displayed. The adaptive information presentation

Primary control area

Figure 2. Eight hot keys for activity support

Advanced control area

523

An Activity-Oriented Approach to Designing a User Interface for Digital Television

method decides which information is displayed

according to its importance. When a display
space is reduced, only important information
will be displayed, and font size is automatically
adjusted to keep information legible (Figure 3b
and 4). In addition to the adaptive information
presentation method, the system also provides a
virtual magnifying glass that enlarges focused
content while it shrinks the neighborhood area
using a fisheye display method.

Context-Sensitive Menu Functions

context and provide suitable activity support options. The functions are displayed in a menu bar
and numbered in terms of their appropriateness
(Figure 4).

Usability Evaluation
The purpose of usability evaluation is twofold:
(1) assessing user interface design usability, and
(2) evaluating usefulness of each user interface
design feature from the users point of view.

To avoid unnecessary menu traversals, functions

appropriate to the current use context are provided
to users. The context-sensitive menu functions
can detect a users location in the activity menu

Participants

Figure 3a. Single-View mode

Figure 3b. Multiple-view mode

28

Figure 4. Context-sensitive menu function

524

Twenty subjects participated in the usability

evaluation. In order to represent user population,

28

An Activity-Oriented Approach to Designing a User Interface for Digital Television

the recruited participants included 11 high skill

level subjects (i.e., three high school students,
four college students, four middle-aged technical
professionals) and nine low skill level subjects
(i.e., three children in elementary school, three
middle-aged housewives, and three older adults).
They all regarded themselves as early adopters of
DTV when they answered the survey question.

Material, User Tasks and Procedures

A user interface concept prototype was developed
using Microsoft Visual C++. The prototype simulated the user interface to support three kinds of
activities: entertainment, transaction, and information browsing. A liquid-crystal display (LCD)
TV was used to display the prototype. The experimental environment is shown in Figure 5.
Before the usability evaluation session began, the experimenter introduced the evaluation
procedure and tasks to the participants. In the
evaluation stage, subjects were asked to perform
tasks by following a test scenario. The test scenario represented activities frequently engaged in
by family members in a morning. It consisted of
three scenarios: (1) searching for news/movie program, (2) browsing information, and (3) shopping
on TV. Each activity scenario in turn contained
several tasks (Table 2). Specifically, the searching for news/movie program scenario involved
three tasks: (1) selecting a TV program from the

program guide, (2) browsing a movie from a videoon-demand (VoD) service, and (3) scheduling a
program recording. The information-browsing
scenario illustrated daily information-related
activities such as checking weather, stock information, and traffic information. Finally, the TV
shopping task simulated TV-shopping behavior
including searching for products, browsing product information, comparing different products,
and engaging in online transactions.
When performing these tasks, subjects were
asked to think aloud through every task step. After
performing the tasks, performance data (task error rate and verbal protocol) were recorded. Upon
each task completion, subjects rated the perceived
usefulness (five-point Likert scale: 1-non-useful,
5-very useful) and perceived usability (five-point
Likert scale: 1-unusable, 5-easy to use) of each
user interface feature.

Results
Task Performance
Scenario 1: Searching for News/Movie
Programs
In the searching for news/movie programs
scenario, almost all operations were error free,
except that a few subjects made an error at the

Figure 5. The experimental environment

525

An Activity-Oriented Approach to Designing a User Interface for Digital Television

step press function key (error rate = 0.4) in the

scheduling a program recording task (Table 3). In
order to understand why subjects made such an
error, we had a debriefing interview with subjects
who made the error. We found that the low skill
group subjects expected to find the record key
on the remote control, while the high skill group
subjects anticipated that the record function was
in the main menu on the screen. These expectations were not consistent with our design. In our
design, the record function is only embedded
in the context sensitive function bar. The result
caused us to reconsider adding an array of the

hard function keys on the remote control for those

frequently used functions (e.g., record, play, stop,
forward, back).

Scenario 2: Browsing Information

There is only one operational error found in this
scenario (Table 4). Subjects made errors in the
select the information item step of the information-browsing task (error rate = 0.15). After
completing the task, we interviewed the subjects
and found that three low-skilled subjects did not
understand what submenu means, because they

Table 2. Users activities, tasks, and operations

Tasks

Sub Tasks
selecting a TV
program

Searching for a
News/Movie
program

Browsing
Information

Shopping
on TV

526

Operations & [DTV design features in the task]

(1) power on; (2) key in target program number; (3) press enter

browsing a movie
from a video-ondemand

(1) power on; (2) main menu [shallow menu structure]; (3) select
VOD;
(4) select program type; (5) select program; (6) start movie

scheduling a
program recording

(1) power on; (2) key in target program number; (3) press enter;
(4) press function key [context sensitive functions]; (5) select
function;
(6) start recording

checking weather

(1) power on; (2) main menu [shallow menu structure]; (3) select
information services; (4) select the item weathers
(In TV viewing state, use multiple display):
(1) press short-cut bar; (2) select items weather information
adaptive information presentation], multiple viewing & display
management]

stock information

(1) power on; (2) main menu [shallow menu structure], (3) select
information services, (4) select the item stock; (5) select a
target stock.
(In TV viewing state, use multiple display):
(1) press short cut bar; (2) select items my stock [adaptive
information presentation], multiple viewing & display
management]

traffic information

(1) Power on, main menu [shallow menu structure]; (2) select
information services, (3) select the item traffic
(in TV viewing state, use multiple display):
(1) press short cut bar; (2) select the item traffic information;
adaptive information presentation], multiple viewing & display
management]

TV - shopping

(1) power on; (2) main menu; (3) select purchase; (4) TVshopping;
(5) select shopping store; (6) select target products; (7) compare
products;
(8) buy the product. [shallow menu structure], [activity oriented
UI flow]

An Activity-Oriented Approach to Designing a User Interface for Digital Television

Table 3. Error rate of select news/movie programs tasks

Tasks

Sub Tasks

Operations
(1) power on
selecting a TV
(2) key in target program number
program
(3) press enter
(1) main menu
browsing a
(2) select Video On Demand
Searching movie from
(3) select program type
for News/
video-on(4) select program
Movie
demand
(6) start movie
programs
(1) key in target program number
scheduling a (2) press enter
program
(3) press function key
recording (4) select function
(5) start recording

Error rate
0.05
0.00
0.00
0.00
0.00
0.05
0.00
0.00
0.00
0.00
0.40
0.00
0.00

Table 4. Error rate of browsing information task

Tasks

browsing
information

Sub Tasks

Access services

Operations

Error rate

(1) power on
(2) main menu [shallow menu structure]
(3) select information services
(4) select the information item

0.00
0.00
0.00
0.15

(In TV viewing state, use multiple display):

(1) press short-cut bar
(2) select services items

0.00
0.00

did not have any computer experience. In revising

the menu design, we considered providing a clue to
guide the low-skill user group when they interact
with the TV menu. Or, we would also provide a
number key input for the low skill group to allow
them direct function access.

Scenario 3: Shopping on TV
All subjects not only understood how to perform the TV shopping task but were also able to
complete it within a short time. In the debriefing interview, we asked subjects why they were
so skilled. Subjects indicated that they felt the
TV shopping functions were similar to real life
shopping situations. In addition, the function
arrangement in the menu structure was easy to
understand and navigate.

Evaluation of Usefulness for Each User

Interface Design Feature ActivityOriented User Interface Flow
All subjects were able to follow the user interface
structure to perform TV viewing, information
retrieving, and transaction activities with minimal
errors (mean error rate = 0.03%). Subjects felt
that the activity-oriented user interface flow was
useful (mean of perceived usefulness = 4.50, Std.
= 0.83) and usable (perceived ease of use = 4.30,
Std. = 0.98) for the TV-shopping task. With this
design feature, they did not have to search for the
service they needed at every task step. That is, the
system automatically provided relative information and functions to support the TV-shopping
task without traversing through multiple service
menus. Therefore, they were not afraid that they
might get lost while performing the task.

527

An Activity-Oriented Approach to Designing a User Interface for Digital Television

Table 5. Error rate of TV shopping task

Tasks

Shopping on TV

Sub Tasks

TV shopping

Operations
(1) power on
(2) main menu
(3) select purchase
(4) TV-shopping
(5) select shopping store
(6) select target products
(7) buy the product

Shallow Menu Structure

Twenty users gave a high rating of perceived usefulness (mean = 3.60, Std. = 1.23) and perceived
usability (mean = 3.80, Std. = 1.11) to this design
feature when they completed three information
retrieving, VoD, and TV-shopping activities.
Subjects were able to complete tasks without
making any errors (mean error rate = 0.00). Time
recordings of menu navigation indicated that the
subjects did not need much time thinking about
where to go next. Furthermore, the verbal protocol did not show that subjects experienced any
frustration or confusion.

Multiple Viewing and Viewport

Management
The multiple-viewing feature was evaluated by
the information browsing activity. Technical
professionals and students exhibited higher acceptance (i.e., mean of perceived usefulness =
4.07 Std. = 0.37; mean of perceived ease of use =
4.36, Std. = 0.74) of the multiple-viewing mode,
while housewives (mean of perceived usefulness = 2.33, Std. = 0.58) and older adults (mean
of perceived usefulness = 1.67, Std. = 0.58) had
problems with this feature. The former expressed
that the multiple-view mode was convenient for
them because they could use the information
service while they were watching a TV program.
However, the latter just wanted to use single-view
mode. If multiple-view mode were the only choice,

528

Error rate
0.00
0.05
0.05
0.00
0.00
0.00
0.00

housewives and older adults hoped that the system

manufacturer could provide an automatic display
management function. Although this design
feature may not be well received by housewives
and older adults, their task performance was still
above the baseline (mean of perceived ease of use
= 3.50, Std. = 0.58).

Adaptive Information Presentation

This feature was well received by all subjects
(mean of perceived usefulness = 3.85, Std. = 1.27)
in information-browsing tasks, especially for older
adults. It can enhance displayed information legibility. However, a concern was raised that adaptive
information presentation might cause a side effect,
that is, increasing font size may decrease the number of menu options displayed and consequently
hamper menu interaction performance. However,
usability evaluation results indicated that this was
not the case. Results revealed that it did not have
a negative effect on menu selection or searching
for menu options (mean of perceived easy of use
= 3.85 Std. = 1.27). In fact, it enhanced menu
interaction performance for older adults.

Context-Sensitive Functions
All subjects gave high marks to this design feature. Students and housewives expressed that this
design feature was helpful (mean of perceived
usefulness = 3.80, Std. = 1.01) and usable (mean
of perceived easy to use = 4.00, Std. = 1.08) be-

An Activity-Oriented Approach to Designing a User Interface for Digital Television

cause it could alleviate efforts to traverse menus

and reduce operation time.

Conclusion
This study attempts to apply an activity-oriented
design approach to digital TV. This approach
analyzes activities and identifies requirements
for each type of activity. Activity-oriented user
interface flow can help users complete their
activities and reduce their workload. The evaluation results support this notion, and services are
considered to be accessible to users and useful in
supporting activities.
This study also explores good design features
of related products used by users for the same
activities. Incorporating good design features
of related products into the new design not only
eases the learning process for the new product,
but also helps establish a familiar and comfortable
feeling for first-time users. This helps users gain
out-of-the-box experience (Ketola, 2005).
We also found that different user groups have
different user interface expectations. For example,
technicians and students prefer multiple viewing
and multiple tasks. On the contrary, older adults
prefer a simple TV environment. Therefore, the
proposed design concept is flexible enough to
accommodate these two types of interaction
methods in order to meet two separate user requirements.

Acknowledgment
This research is supported by MediaTek Research Center, National Chiao Tung University,
Taiwan.

References
Bdker, S. (1996). Applying activity theory to
video analysis: How to make sense of video data

in human-computer interaction. In B. A. Nardi

(Ed.), Context and consciousness: Activity theory
and human-computer interaction (pp. 147-174).
Cambridge, MA: The MIT Press.
Engestrm, Y. (1987). Learning by expanding: An
activity-theoretical approach to developmental
research. Helsinki, Finland: Orienta-Konsultit.
Hsu, S. H. (2005). Identify user requirements for
user interfaces in digital television environment.
Unpublished technical report, Human Factors
Laboratory, National Chiao Tung university,
Taiwan.
Kaptelinin, V. (1996). Computer-mediated activity: Functional organs in social and developmental
con-texts. In B. A. Nardi (Ed.), Context and consciousness: Activity theory and human-computer
interaction (pp. 103-116). Cambridge, MA: The
MIT Press.
Kaufman, L., & Rousueeuw, P. J. (1990). Finding groups in data: An introduction to cluster
analysis. John Wiley & Sons.
Ketola, P. (2005). Special issue on out-of-the-box
experience and consumer devices. Personal and
Ubiquitous Computing, 9, 187-190.
Kuutti, K. (1996). Activity theory as a potential
framework for human-computer interaction
research. In B. A. Nardi (Ed.), Context and consciousness: Activity theory and human-computer
interaction (pp. 17-44). Cambridge, MA: The
MIT Press.
Mwanza, D. (2002). Towards an activity-oriented
design method for HCI research and practice.
Unpublished doctoral dissertation, The Open
University, UK.
Nardi, B. A. (Ed.). (1996a). Context and consciousness: Activity theory and human-computer
interaction. Cambridge, MA: The MIT Press.
Nardi, B. A. (1996b). Some reflections on the
application of activity theory. In B. A. Nardi

529

An Activity-Oriented Approach to Designing a User Interface for Digital Television

(Ed.), Context and consciousness: Activity theory

and human-computer interaction (pp. 235-246).
Cambridge, MA: The MIT Press.

Norman, D. (2005). Human-centered design

considered harmful. ACM Interactions, 12(4),
14-19.

Norman, D. A. (1988). The psychology of everyday

things. New York: Basic Books.

Rogers, E. (1995). Diffusions of innovations (4th

ed.). New York: The Free Press.

This work was previously published in Interactive Digital Television: Technologies and Applications, edited by G. Lekakos, K.
Chorianopoulos, and G. Doukidis, pp. 148-168, copyright 2007 by IGI Publishing, formerly known as Idea Group Publishing
(an imprint of IGI Global).

530

An Activity-Oriented Approach to Designing a User Interface for Digital Television

Appendix: Factor Loading of Each Activity Group

Information

Education

Entertainment

Daily living

Transaction

Activity Group (5 factors)

Internet

.788

.191

.173

.243

.413

Ball gaming

.002

.450

.535

.011

.272

Watching movie

.704

.454

.385

.064

.488

Body-building

.297

.357

.503

.082

.476

Newspaper

.673

.200

.270

.309

.335

Dinner party

.259

.241

.492

.035

.173

Bookstore shopping

.616

.257

.203

.193

.268

Camping

.355

.437

.481

.446

.364

Shopping

.502

.175

.289

.142

.029

Ball game watching

.136

.250

.478

.308

.283

Magazine

.440

.228

.301

-.147

.368

Singing (Karaoke)

.236

.338

.476

.145

.185

Information showing

.372

.665

.500

.015

.475

Religious activity

.009

.375

.433

.008

.127

Social service

.127

.659

.451

.062

.310

Housekeeping

.305

.170

.104

.661

.243

Art exhibition

.470

.611

.490

-.007

.483

Computer gaming

.362

.145

.082

.615

.285

Accomplishment course

.294

.594

.242

.262

.398

Pet breeding

-.035

.356

.377

.527

.312

School

.239

.587

.370

.267

.252

Gardening

.021

.370

.262

.441

.400

Refresher course

.312

.565

.294

.153

.378

Cooking

.217

.328

.279

.412

.129

Photography

.395

.547

.489

.210

.414

Commodity buying

.346

.341

.158

.360

.670

Tai-Chi chuan

-.108

.471

.250

.178

.187

TV / e-shopping

.300

.301

.338

.384

.638

Handicraft making

.214

.445

.361

.263

.106

Investment

.300

.527

.343

.303

.538

Traveling

.422

.462

.611

.063

.397

Finical news

.444

.360

.407

.227

.498

Joy riding

.413

.311

.559

.325

.341

Lottery

.194

.387

.375

.001

.460

531

532

Chapter 2.11

Designing and Evaluating

In-Car User-Interfaces
Gary Burnett
University of Nottingham, UK

Abstract

INTRODUCTION

The introduction of computing and communications technologies within cars raises a range of
novel human-computer interaction (HCI) issues.
In particular, it is critical to understand how userinterfaces within cars can best be designed to
account for the severe physical, perceptual and
cognitive constraints placed on users by the driving context. This chapter introduces the driving
situation and explains the range of computing
systems being introduced within cars and their
associated user-interfaces. The overall humanfocused factors that designers must consider
for this technology are raised. Furthermore, the
range of methods (e.g., use of simulators, instrumented vehicles) available to designers of in-car
user-interfaces are compared and contrasted.
Specific guidance for one key system, vehicle
navigation, is provided in a case study discussion.
To conclude, overall trends in the development
of in-car user-interfaces are discussed and the
research challenges are raised.

The motor car is an integral part of modern

society. These self-propelled driver-guided vehicles transport millions of people every day for
a multitude of different purposes, for example
as part of work, for visiting friends and family,
or for leisure activities. Likewise, computers are
essential to many peoples regular lives. It is only
relatively recently that these two products have
begun to merge, as computing-related technology
is increasingly implemented within road-going
vehicles. The functions of an in-car computing
system can be broad, supporting tasks as diverse
as navigation, lane keeping, collision avoidance,
and parking. Ultimately, by implementing such
systems car manufacturers aim to improve the
safety, efficiency, and comfort and entertainment
of the driving experience (Bishop, 2005)
Designing the user-interface for in-car computing systems raises many novel challenges, quite
unlike those traditionally associated with interface
design. For instance, in many situations, the use
of an in-car system is secondary to the complex
and already demanding primary task of safely

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Designing and Evaluating In-Car User-Interfaces

controlling a vehicle in 2D space, whilst simultaneously maintaining an awareness of hazards,

largely using the visual sense. Consequently, the
level of workload (physical, visual, and mental)
when using displays and controls becomes a critical safety-related factor. As a further example,
in-car computing systems have to be used by
a driver (and possible also, a passenger) who is
sat in a constrained posture and is unlikely to
be able to undertake a two handed operation.
Therefore, the design (location, type, size, etc.)
of input devices has to be carefully considered,
accounting in particular for comfort, as well as
safety, requirements.
This chapter aims primarily to provide the
reader with an overall awareness of novel in-car
computing systems and the key HCI design and
evaluation issues. The focus is on the user-interface, that is, the means by which the system
reveals itself to the users and behaves in relation
to the users needs (Hackos & Redish, 1998,
p.5). Topics of relevance to both researchers and
practitioners are raised throughout. Given the
complexity of the driving task and the wide range
of computing systems of relevance, the chapter
principally provides breadth in its consideration of
the subject. Nevertheless, some depth is explored
in a case study investigation on the design and
evaluation of user-interfaces for vehicle navigation systems.

TYPES OF IN-CAR COMPUTING

SYSTEMs
Technology is increasingly being seen to have
a critical role to play in alleviating the negative
aspects of road transport, such as congestion, pollution and road traffic accidents (Bishop, 2005).
Many technological initiatives are considered
under the umbrella term, intelligent transport
systems (ITS), where ITS provides the intelligent link between travelers, vehicles, and infrastructure (www.itsa.org, September, 2006).

In this respect, in-vehicle computing systems are

an important facet of ITS. Specifically, there are
two core types of computing and communications
systems which are either being implemented or
developed for use in vehicles:

Information-based systems: These systems

provide information relevant to components
of the driving environment, the vehicle or
the driver. Examples of systems include
navigation (facilitating route planning and
following), travel and traffic information
(traffic conditions, car parking availability,
etc.), vision enhancement (providing an
enhanced view of the road ahead, when
driving at night, in fog or in heavy rain),
driver alertness monitoring (informing the
incapacitated driver if they are unfit to drive)
and collision warnings (presenting warnings
or advice regarding hazards).
Control-based systems: These systems affect the routine, operational elements of the
driving task. Examples of systems include
adaptive cruise control (where the car is
kept at a set time gap from a lead vehicle),
speed limiting (the car speed cannot exceed
the current limit), lane keeping (the drivers
vehicle is kept within a given lane), self
parking (vehicle automatically steers in low
speed operation to position itself within a
selected parking space) and collision avoidance (the vehicle automatically responds
to an emergency situation). Clearly, such
systems fundamentally change the nature
of what we consider to be driving.

It is important to note that there is a third category

of in-car computing system, those which do not
provide any functionality to support the driving
task. These systems are an important consideration
though, as they can negatively influence safety,
particularly through the potential for distraction
(Young, Regan & Hammer, 2003). Such systems
may aim to enhance work-oriented productivity

533

Designing and Evaluating In-Car User-Interfaces

whilst driving (e.g., mobile phones, e-mail and

Internet access) or be primarily conceived for
entertainment and comfort purposes (e.g., music
and DVD players, games). Moreover, they may
be designed for dedicated use in a vehicle or for
operation in a range of different contexts (often
termed nomadic devices).

OVERALL HUMAN FACTORS

ISSUES
Driving is a complex task involving a large number
of subtasks that can be conceptualised as existing within three levels of an overall hierarchical
structure (Michon, 1985):

Strategic tasks (highest level global travel

decisionse.g., which car to take, which
route to take);
Tactical tasks (making concrete maneuvers requiring interaction with other road
userse.g., changing lane, turning at a
roundabout);
Operational tasks (motor execution of tasks
planned at higher levelse.g., turning steering wheel, pressing brake).

Inevitably, the introduction of new technologies

into the driving context will have a considerable
impact across all three levels. As a result, there
are many human-focused issues that must be considered in the design and evaluation process for
in-car computing systems. To provide structure to
a discussion of these issues, two overall scenarios
are envisaged which may arise from poor design
and/or implementation of the technology.

534

Overload: Many of these systems (particularly those providing novel types of information and/or interactions) lead to situations in
which a driver must divide their attention
between core driving tasks (e.g., watching
out for hazards) and secondary system tasks

(e.g., inputting information). Furthermore,

systems may provide excessive information
in an inappropriate way leading to high levels
of mental workload, stress and frustration.
Such issues often manifest themselves as
distraction to the driver (biomechanical,
visual, auditory and/or cognitive).
Underload: Control-based systems clearly
automate certain aspects of driving, transferring certain responsibilities from operator
to computer (e.g., staying in lane), whilst
potentially providing new tasks for the driver
(e.g., monitoring system performance). Automation is a fundamental human factors
topic with a considerable research literature
(see Wickens et al., 2004). Key concerns
in this context relate to the potential for a
driver exhibiting reduced situational awareness (e.g., for other road users), negative
behavioral adaptation (e.g., by taking greater
risks) and de-skilling (e.g., driver not able
to resume control in the event of system
failure).

THE HUMAN-CENTRED DESIGN

PROCESS
The fundamental components of a human-focused
approach hold true for in-car computing, as much
as for any interactive product or system, that is,
early focus on users and tasks, empirical measurement and iterative design (Gould & Lewis,
1985). A comprehensive understanding of the
context in which in-car computing devices will
be used is especially important early in the design
process. Context of use refers to the users, tasks
and equipment (hardware, software, and materials), and the physical and social environments in
which a product is used (Maguire, 2001, p.457).
A context of use analysis assists in developing the
initial requirements for a design and also provides
an early basis for testing scenarios. Moreover,
context of use analysis provides a focused ap-

Designing and Evaluating In-Car User-Interfaces

proach that helps to ensure a shared view among

a design team. In the driving situation, there are
several context of use issues which will have a
significant effect on how an in-car computing
system is subsequently designed. Accounting for
these raises many unique challenges for in-car
user-interface designers.

Users
As with many other consumer products, there will
be a large variability in user characteristics (e.g.,
in perceptual and cognitive abilities, computer
experience, anthropometry) to consider when
designing in-car computing systems. Car manufacturers may have particular socio-economic
groups in mind when designing a vehicle, but the
user base may still be extremely large.
One fundamental individual difference factor
often addressed in research is driver agedrivers
can be as young as 16 (in certain countries) and
as old as 90. In this respect, younger drivers may
be particularly skilled in the use of computing
technology, in comparison with the population
at large, but are especially prone to risk taking
(Green, 2003). Moreover, studies have shown a
limited ability to divide attention and prioritize
sources of information, largely due to lack of
driving experience (Wickman, Nieminem &
Summala, 1998). Subsequently, system block outs,
which prevent the use of complex functions in
inappropriate driving situations, are likely to be
of particular benefit for these individuals.
In contrast, older drivers often suffer from
a range of visual impairments that can lead to a
range of problems with in-vehicle displays. For
instance, presbyopia (loss of elasticity in the lens
of the eye) is extremely common amongst older
people, as is reduced contrast sensitivity. Studies
consistently show that older drivers can take 1.5
to 2 times longer to read information from an
in-vehicle display compared to younger drivers
(Green, 2003). Given that drivers have a limited
ability to change the distance between themselves

and an in-vehicle display, the size, luminance and

contrast of presented information are obviously
critical design factors.

Tasks
A key task-related issue is that the use of an in-car
computing system is likely to be discretionary.
Drivers do not necessarily have to use the system to achieve their goals and alternatives will
be available (e.g., a paper map, using the brake
themselves). As a result, the perceived utility
of the device is critical. Furthermore, drivers
affective requirements may be particularly important. In certain cases, this requirement may
conflict with safety-related needs, for instance,
for a simple, rather than flashy or overly engaging
user-interface.
The factor that most differentiates the driving
context from traditional user-interface design is
the multiple-task nature of system use, and in this
respect, there are two critical issues that designers
must take into consideration. The first concerns
the relationship between primary driving tasks and
secondary system tasks, as drivers seek to divide
their attention between competing sources of information. Driving is largely a performance and
time-critical visual-manual task with significant
spatial components (e.g., estimating distances).
Consequently, secondary tasks must not be overly
time-consuming to achieve or require attentional
resources that are largely visual, manual, and
spatial in nature, if they are to avoid having a
significant impact on primary driving.
A second fundamental issue is the amount
of information processing or decision making
required for successful task performance, known
as mental workload (Wickens et al., 2004). Novel
in-car computing systems may provide functionality of utility to a driver or passengers, but
interaction with the technology will inevitably
increase (or in some cases decrease) overall workload. Context is very important here, as driving
is a task in which workload varies considerably

535

Designing and Evaluating In-Car User-Interfaces

from one situation to another (compare driving

in city traffic versus on the motorway). In this
respect, certain authors (e.g., Green, 2004; Jones,
2002; Markkula, Kutila, & Engstrm, 2005) have
taken the view that workload managers must be
developed which make real-time predictions of
the workload a driver is under and only present
information or enable interactions to occur when
overall workload is considered to be at an acceptable level. As an example, an incoming phone
call may be sent straight to voice mail when the
driver is considered to be particularly loaded (e.g.,
when driving in an unfamiliar city), but may be
permitted in a lower workload scenario (e.g.,
driving along a dual carriageway and following a
lead vehicle). Simple workload managers already
exist in some vehicles (e.g., http://driving.timesonline.co.uk/article/0,,12929-2319048,00.html,
September 2006), nevertheless, there are several
complex research issues which must be addressed
to fully realize the benefits of adaptive software
in this context. For instance, workload managers
need a comprehensive and accurate model of the
driver, driving tasks and the driving environment.
Given the vast range of variables of relevance to
these categories, many of which do not lend to
accurate and reliable measurement, extensive
workload managers are likely to remain in the
research domain for several years.

Figure 1. Example of a head-up display (HUD)

536

Equipment
The driving situation necessitates the use of input and output devices which are familiar to the
majority of user-interface designers (pushbuttons,
rockers, rotaries, LCDs, touchscreens, digitized
or synthesized speech), together with equipment
which is perhaps less known. For instance, there
is a considerable research literature regarding the
use of Head-Up Displays (HUDs) within vehicles.
A HUD uses projection technology to provide
virtual images which can be seen in the drivers
line of sight through the front windscreen (see
Figure 1). They are widely used within the aviation
and military fields, and are now beginning to be
implemented on a large-scale within road-based
vehicles. HUDs will potentially allow drivers to
continue attending to the road ahead whilst taking
in secondary information more quickly (Ward &
Parkes, 1994). As a consequence, they may be
most applicable to situations in which the visual
modality is highly loaded (e.g., urban driving),
and for older drivers who experience difficulties
in rapidly changing accommodation between near
and far objects (Burns, 1999).
From a human-focused perspective, there are
clear dangers in simply translating a technology
from one context to another, given that vehiclebased HUDs will be used by people of varying
perceptual and cognitive capabilities within an

Designing and Evaluating In-Car User-Interfaces

environment where there is a complex, continually

changing visual scene. Specifically, researchers
have established that poorly designed HUDs can
mask critical road information, disrupt distance
perception and visual scanning patterns, and negatively affect the ability of drivers to detect hazards
in their peripheral vision (known as perceptual
tunneling)summarized by Tufano (1997) and
Ward and Parkes (1994). Critical design factors
that emerge from these findings include: display
complexity; contrast and luminance; color choice;
size of image; spatial location; and virtual image
distance. Perhaps the most important design-related requirements are to consider carefully what
and how much information is most appropriate
to present on a HUD. There are temptations for
designers to present ever-increasing amounts of
information on HUDs. However, in contrast with
traditional in-vehicle displays, a HUD image, by
its very presence in the drivers line of sight, will
demand focused attention (Burnett, 2003).

Environments
The physical environment is also a specific area
that designers need to be aware of. In particular,
the light, sound, thermal and vibration environment within a car can be highly variable. A range
of design requirements will emerge from a consideration of these factors, for instance, potential
for glare, problems with speech interfaces, use
with gloves, and so on.
From anthropometric and biomechanical perspectives, the vehicle cabin environment provides
many challenges for designers. This is an area in
which designers make considerable use of CAD
modeling to analyze different locations for displays and controls, ultimately aiming to ensure
good fit for the design population. However, drivers sit in a constrained posture, often for several
hours and have limited physical mobility (e.g., to
comfortably view displays or reach controls). Consequently, there is limited space within a vehicle
for the placement of a physical user-interface, a

key problem for designers hoping to implement

additional functionality within the vehicle.
To a large extent, this factor has fueled the
development of multi-modal user-interfaces,
where a small number of controls, together with
menu-driven screens, provide access to many
functions within the vehicle. Clearly, such visually-oriented user-interfaces are likely to promote
a considerable amount of eyes-off-road time,
and empirical studies have confirmed this prediction (Dewar, 2002). Moreover, users mistaking
the current mode is a well-established problem
in user-interface design, and clear feedback is an
important design requirement (Preece, Rogers &
Sharp, 2002). In many respects, there is a trade-off
in design between the number of discrete controls
that a user must scan within a vehicle and the
number of levels within a menu-based system
that must be explored and understood. This is a
very similar problem to that considered by HCI
researchers in the 1980s and 1990s interested in
the breadth versus depth of menus in graphical
user-interfaces (Shneiderman, 1998). An overall
recommendation from such HCI research is that
breadth should generally be favored over depth,
as excessive depth can cause considerably more
problems for the user than an equivalent breadth,
largely due to the cognitive problems of navigation (Shneiderman, 1998). Whilst such guidance
is considered to be of relevance to the design of
in-car computing, research is still required which
considers the trade-off existing in the multipletask driving environment.

METHODS FOR USE IN DESIGN

AND EVALUATION
In considering the range of methods that a designer can utilize when designing and evaluating
in-car computing systems, the first difficulty is
in establishing what is meant by a method. In
this respect, a human factors method for testing

537

Designing and Evaluating In-Car User-Interfaces

in-car systems can be seen to be a combination

of three factors:
1.

2.

3.

Which environment is the method used

in (road, test track, simulator, laboratory,
etc.). As can be seen in Figure 2 (redrawn
and adapted from Parkes, 1991), there is a
fundamental trade off in choosing a method
environment between the need for control
and the validity of results. Choosing an
environment will also be largely influenced
by practical considerations, the knowledge/
skills of the design and evaluation team and
resource limitations.
Which task manipulations occur (multiple
task, single task loading, no tasks given,
etc.)? In certain methods, there is an attempt to replicate or simulate the multiple
task nature of driving. For other methods,
performance and/or behavior on a single task
may be assessed and the potential impact on
other tasks inferred from this. Most removed
from actual driving, some methods do not
involve users, but instead aim to predict impacts or issues, for instance through the use
of expert ratings or modeling techniques.
Which dependent variables (operationalized as metrics) are of interest. In assessing an in-car computing user-interface, a

large range of possible metrics could be

implemented. Some will relate to drivers
performance with primary driving tasks
(e.g., lane position, hazard detection) or their
use of primary vehicle controls (e.g., use of
brake, steering wheel). Other metrics focus
on driver performance and/or the demand
of secondary tasks (e.g., task times, errors,
display glances). As noted by Parkes (1991),
usability evaluations of in-car computing
devices should incorporate a wide range of
measures relevant to the different levels of
the driving task. For instance, at the strategic
level, observation techniques and surveys
are of relevance, whereas verbal protocols,
interviews and questionnaires can capture
the behavior of drivers at the tactical level. As
noted by Parkes, such an approach provides
complete, rather than partial, pictures of
product usability (p.1445).
There is presently considerable research investigating specific methods for use in the design
and evaluation of in-car user-interfaces. As noted
by Preece et al. (2002), in deciding on any HCI
method, the design team must consider the overall
goals of the work, specific questions to be addressed, the practical and ethical issues and how
data will need to be analyzed and reported. For

Figure 2. Environments for evaluation of in-car computing devices and the relationship between validity and control
Real road field trials
Increasing
confidence
that data
correspond to
real phenomena

Real road test trials

Test track studies
Dynamic vehicle simulations
Static vehicle simulations
Part task/laboratory studies

538

Increasing
control
of variables
and replication

Designing and Evaluating In-Car User-Interfaces

in-car computing, these principles still hold, and

many of the same global techniques used in the
HCI area (for example, questionnaires, interviews,
guidelines/checklists) will be used. However, by
necessity, bespoke methods (or at least specific
versions of generic methods) are required that
account for the particular complex, safety-critical
characteristics of the driving context. The following section summarizes key methods currently
used and highlights some of the important research
issues under investigation. Moreover, primary
advantages and disadvantages are given. Table
1 summarizes some of the key issues.

Field Trials
Participants are given a car fitted with an operational system for several months for use in
everyday activities. This method tends to look
at broad issues relating to the long-term use of a
system, for example, drivers acceptance of the
technology, and whether any behavioral adaptation effects arise. Objective data can be measured
using on-board instrumentation (e.g., cameras,
speed sensors) whereas subjective data is often
captured using survey or interview-based approaches. Clearly, such a method provides an ecologically valid test of a system, and is particularly

Table 1. Overview of methods used to evaluate the user-interface for in-car computing systems
Method

Environment

Task manipulations

Overall
Measures

Primary
Advantages

Primary
Disadvantages

Field trials

Real road (in

everyday driving)

Multi-task (according
to driver motivation)

Primary/ secondary
task performance/
behavior, user
opinions, etc.

Ecological validity,
can assess behavioral
adaptation

Resource intensive,
ethical/liability
issues to consider

Road trials

Real road (in predefined settings)

Multi-task (commonly,
evaluator-manipulated)

Primary/ secondary
task performance/
behavior, user
opinions, etc.

Balance of
ecological validity
with control

Resource intensive,
ethical/liability
issues to consider

Simulator
trials

Virtual driving
environment
(varying in fidelity)

Multi-task (commonly,
evaluator-manipulated)

Primary/ secondary
task performance/
behavior, user
opinions, etc.

Control over
variables, safe
environment,
cost-effective

Validity of driver
behavior, simulator
sickness

Occlusion

Laboratory/
statically in car

Secondary task
achieved in controlled
visual experience

Visual demand of
user-interface

Standardized
approach, control
over variables

Limited scope, concern over validity of

approach and metrics

Peripheral
detection

Road/virtual
driving
environment

Multi-task (although
commonly, evaluatormanipulated)

Visual/ cognitive
workload

Assesses cognitive,
as well as visual
demand

Can be resource
intensive, range of
approaches

Lane
change task

Specific lo-fidelity virtual driving

environment

Multi-task motorway
driving scenario

Primary lateral control

of vehicle

Standardized
approach, control
over variables

Difficult to relate
results to interface
characteristics

15 second
rule

Laboratory/ statically in car

Secondary task
achieved without presence of driving task

Secondary task time

(whilst stationary)

Simple approach

Only relates to
certain aspects of
visual demand

KeystrokeLevel
Model
(KLM)

Modeling exercise

No user trials take

place - models expert
performance

Secondary task time

(whilst stationary)

Quick/cheap,
analysis explains
results

Only relates to
certain aspects of
visual demand

Extended
KLM

Modeling exercise

As for KLM, but with

additional assumptions

Visual demand of
user-interface

Quick/cheap, analysis explains results

Requires reliability
assessments

539

Designing and Evaluating In-Car User-Interfaces

appropriate to the late stages of the design process

where a robust prototype is available. Nevertheless, field trials can be extremely expensive and
various ethical and liability considerations must
be accounted for. An example of a field trial that
was carried out in Sweden concerned drivers use
of intelligent speed adaptation systems (whereby
a vehicles speed is automatically kept within the
speed limit for the current area. Wallen, Warner
& Aberg, 2005).

Road Trials
Drivers take part in a short-term (normally less
than one day) focused study using a system in
an instrumented car on public roads (occasionally on test tracks). For such trials, a wide range
of variables may be measured and analyzed
(e.g., visual behavior, workload, vehicle control,
subjective preference) depending on the aims of
the study. Road trials enable more experimental
control than field trials, but are still potentially
affected by a wide range of confounding variables
(e.g., traffic conditions, weather). Furthermore,
such a method remains costly to implement and
requires robust protocols to ensure the safety of
all concerned. Many road trials are reported in
the literature, particularly concerning information
and entertainment/productivity oriented systems.
For instance, Burnett and Joyner (1997) describe

a study which evaluated two different user-interfaces for vehicle navigation systems.

Simulator Trials
Drivers take part in a short-term (normally less
than one day) focused study using a system fitted
or mocked up within a driving simulator. The
faithfulness that a simulator represents the driving
task (known as its fidelity) can vary considerably,
and configurations range from those with single
computer screens and game controller configurations, through to real car cabins with multiple
projections and motion systems. An example of
a medium fidelity driving simulator is shown in
Figure 3.
Driving simulators have become increasingly
popular in recent years as a result of reduced
hardware and software costs, and potentially offer
an extremely cost-effective way of investigating
many different design and evaluation issues in a
safe and controlled environment (Reed & Green,
1999). Nevertheless, there are two key research
issues concerning the use of driving simulators.
Firstly, it is well known that individuals can experience symptoms of sickness in driving simulators,
manifested as feelings of nausea, dizziness, and
headaches. There has been considerable research
regarding such sickness in virtual environments,
and whilst there is still debate regarding the

Figure 3. Example of a medium fidelity driving simulator

540

Designing and Evaluating In-Car User-Interfaces

theoretical basis for the phenomenon (see for

instance Nichols & Patel, 2002), there is practical
guidance for those using driving simulators. For
instance, screening questionnaires can be used
to eliminate individuals who are most likely to
experience sickness during a trial (Kennedy et
al., 2001). Furthermore, various countermeasures
can be used in the development of the simulator
and its environment to reduce the prevalence of
sickness (e.g., high, consistent frame rate, air-conditioning, natural background lighting. Nichols
& Patel, 2002).
A second and more complex issue concerns
validity, particularly behavioral (or construct)
validity, that is, the extent to which drivers behave
in the simulator as they would in the real world
(Blaauw, 1982; Reed & Green, 1999). Driving
simulator validity is problematic to study for
several reasons. Running both road and simulator trials which are comparable (in terms of
participants, tasks, measures, procedures, etc.)
can be extremely difficult to achieve, and ultimately will be resource intensive. Furthermore,
validity in this area is widely recognized to be a
function of a large number of variables, including
those relating to how the vehicle is represented
(e.g., primary and secondary control design, the
sense of enclosure, viewing angles, engine noise,
vibration, motion, etc.) and those concerning the
driving environment (e.g., visual field of view,
screen resolution, graphical complexity, traffic
representation, wind/road noise, etc. Kaptein et
al.,1996; Peters & Peters, 2002). Most importantly,
our understanding of validity must consider the
driving task itself. Driving is a complex task, involving a substantial number of discrete physical,
perceptual and cognitive behaviors, and a specific
simulator configuration will only enable a subset
of these to be investigated (e.g., speed control,
headway maintenance).
As a consequence, despite the importance of
the topic, there are few driving simulator validity
studies in the open literature. Moreover, various

limitations can be expressed for previous research

in this area:

It is difficult to generalize from existing

validity studies, as they tend to be very
specific to (a) the simulator configuration
under investigation, and (b) the technology (hardware and software) available at
that time (see for instance, Tornros (1998)
compared with Blaauw (1982)).
Studies inevitably only concern a small
number of variables, for instance the effect
of screen resolution and forward field of
view on speed and headway choice (Jamson, 2001); or the effect of characteristics
of torque feedback for steering on curve
negotiation (Toffin et al., 2003).
Studies often do not report critical data
regarding the simulator configuration (e.g.,
steering sensitivity, max/min acceleration,
driver eye height) which, for given types of
study will be critical in results interpretation
and cross-study comparison.

Occlusion
This is a laboratory-based method which focuses
on the visual demand of in-vehicle systems. Participants carry out tasks with an in-vehicle system
(stationary within a vehicle or vehicle mock up)
whilst wearing computer-controlled goggles with
LCDs as lenses which can open and shut in a
precise manner (see Figure 4). Consequently, by
stipulating a cycle of vision for a short period of
time (e.g., 1.5 seconds), followed by an occlusion
interval (e.g., 1.5 seconds), glancing behaviour
is mimicked in a controlled fashion. Occlusion
offers a relatively simple method of predicting
visual demand, but is has been pointed out that
its emphasis on user trials and performance data
means that it requires a robust prototype and is
therefore of limited use early in the design process
(Pettitt et al., 2006).

541

Designing and Evaluating In-Car User-Interfaces

Figure 4. The occlusion method with participant wearing occlusion goggles, with shutters open (left)
and closed (right)

Following considerable research, the occlusion method has recently been formalized as an
international standard (ISO, 2005). In particular,
guidance is given on how many participants are
required, how much training to give, how many
task variations to set, data analysis procedures, and
so on. Moreover, two key metrics are stipulated:
total shutter open time (the total time required
to carry out tasks when vision is available); and
resumability (the ratio of total shutter open time
to task time when full vision is provided). For
resumability, there is considerable debate regarding the merit of the measure. Advocates believe
the metric provides an indication of the ease by
which a task can be resumed following a period
without vision (Baumann et al., 2004). Critics
point out that the metric is also influenced by the
degree to which participants are able to achieve
tasks during occluded (non-vision) periods (Pettitt et al., 2006). Consequently, it can be difficult
for a design team to interpret the results of an
occlusion trial.

Peripheral Detection Task

This method requires drivers to carry out tasks
with an in-car system (either on road or in a
simulator) and to respond to the presence of lights
within their periphery. The speed and accuracy of
responses are considered to relate to the mental

542

workload and distraction associated with secondary tasks (Young et al., 2003). The advantage of
this method over occlusion is that it offers an assessment of cognitive, as well as visual demand (of
relevance to the assessment of speech interfaces,
for instance). The primary disadvantage is that the
method still requires some form of driving task.
Moreover, in contrast with occlusion, the method
has not been fully standardized, and the ability to
make cross study comparisons is severely limited
by the specific choice of driving task scenarios
(affecting task load and the conspicuity of the
peripheral stimuli). It has also been noted that it
is very difficult to discern between the level of
cognitive demand and the visual demand for a
given user-interface (Young et al., 2003).
An interesting recent development addresses
some of these limitations. Engstrom, Aberg and
Johansson (2005) considered the potential for
the use of a haptic peripheral detection task,
where drivers respond to vibro-tactile stimulation through the wrist whilst interacting with an
in-vehicle system. Clearly, such a variation of
peripheral detection is not affected by variations
in lighting conditions. Furthermore, the authors
argue on the basis of their validation work that
this method provides a pure measure of cognitive load not mixed up with the effect of simply
looking away (p.233).

Designing and Evaluating In-Car User-Interfaces

Lane Change Task

This method occurs in a basic PC simulated environment in which drivers are requested to make
various lane change maneuvers whilst engaging
with an in-vehicle system. The extent to which the
profile of maneuver made by a driver varies from
the optimum maneuver (the normative model) is
considered to be a measure of the quality of their
driving. Specifically, the method has the ability to
assess the impact of an in-car computing system
on a drivers awareness of the driving environment
(perception, reaction), and, their ability to safely
control the vehicle (maneuvering, lane keeping)
Mattes (2003). Considerable research is ongoing
with the lane change task in an attempt to develop
an international standard (Transport Canada,
2006). Key research issues concern participant
choice, training requirements and developing
acceptable limits for performance.

15 Second Rule
Participants carry out tasks with an in-car computing system whilst stationary within a vehicle
or mock up (i.e., with no driving task) and with
full vision. The mean time to undertake a task is
considered to be a basic measure of how demanding visually it is likely to be when driving (Green,
1999). A cut-off of 15 seconds has been set by
the Society for Automotive Engineers (SAE). If
the task on average takes longer than 15 seconds
to achieve when stationary, it should not be allowed in a moving vehicle. The method is simple
to implement and has the key advantage that it
has been formalized in an SAE statement of best
practice (SAE, 2000).
Research by Green (1999) and other research
teams (e.g., Pettitt et al., 2006) has shown strong
correlations between static task times and the
total amount of time spent looking away from the
road at displays/controls, both in simulator and
road studies. However, the correlation between
static task times and the duration of single glances

towards an in-vehicle display is generally poor.

This is important because a user-interface may
promote a small number of very long glances (e.g.,
as a result of dynamically changing visual information) which can have a considerable negative
effect on driving performance (Burnett & Joyner,
1997). It is for this primary reason that many
authors advocate the use of the occlusion method
as a better low-cost method for investigating the
visual demand of an in-car user-interface (Pettitt
et al., 2006; Stevens et al., 2004).

Keystroke Level Model (KLM)

The KLM method from the GOMs family of techniques is well known to HCI researchers and (to
a lesser extent) practitioners (Preece et al., 2002;
Shneiderman, 1998). It is a form of task analysis
in which system tasks with a given user-interface
are broken down into their underlying physical and
mental operators, e.g., pressing buttons, moving
hand between controls, scanning for information. This is a method that is extremely cheap to
implement, as there is no need for participants,
and the method can be used with very basic prototypes early in the design process. Time values
are associated with each operator and summed
to give a prediction of task times. Researchers
have developed new operator values relevant to
the in-car situation (e.g., time to search a visual
display, locate a control, move hand back to steering wheel) and have reported strong correlations
between predicted task times and times based on
user trials (Green, 2003; Pettitt et al., 2005). Task
times can be related to certain measures of visual
demand for in-car user-interfaces.
In an extension of the KLM method, Pettitt,
Burnett and Stevens (2007) recently developed
new rules that enable designers to develop predictions for a broader range of visual demand
measures. In particular, the extended KLM
considers a time-line view of an interaction in
which a cycle of vision/non-vision occurs with a
user-interface (similar to the occlusion protocol).

543

Designing and Evaluating In-Car User-Interfaces

The authors have found that their version of KLM

can differentiate between tasks as effectively as
does the occlusion technique, but recommend that
further development is carried out to ensure that
practitioners can utilize the method reliably.

CASE STUDY: VEHICLE

NAVIGATION SYSTEMS
To ground many of the issues previously mentioned, a specific system type has been chosen for
further discussion (vehicle navigation systems).
Many of the individual points made for this system
can be generalized and are applicable to other
in-car computing technologies.
Vehicle navigation systems aim to support
the strategic (e.g., route planning) and tactical
(e.g., route following) components of the overall
driving task. They have the greatest potential to
assist drivers who undertake many unfamiliar
journeys, for instance as part of work, or during
leisure trips (e.g., when on holiday) and those
who experience extreme difficulties with existing methods of navigation (particularly paper
maps). When linked with reliable, real-time traffic
information (thus providing dynamic guidance),
the perceived utility of navigation systems to
the everyday motorist is significantly enhanced
(Bishop, 2005).
The market potential for vehicle navigation
systems has already been demonstrated in Japan,
where the technology has been available since
the early 1990s. Approximately 40 percent of all
vehicles on Japans roads now have a navigation
system installed (http://www.jetro.go.jp/en/market/trend/topic/2004_12_carnavi.html, September 2006). In many other countries, the popularity
of navigation systems is currently reduced in
relation to Japan, but is predicted to rise rapidly
over the next few years (Bishop, 2005).
The majority of human factors issues relevant
to this form of technology relate to overload, although as shall be seen, underload is increasingly

544

being researched. With respect to overload, clearly,

a key concern is the potential for driver distraction
and there has been considerable research on this
topic since the mid 1980s (see Young et al., 2003
and Srinivisan, 1999, for reviews). In using a vehicle navigation system, drivers must interact with
controls (e.g., to enter a destination, chance map
scale) and view/understand displays (e.g., to decide
which turn to make, to examine options within a
menu). In many cases, these interactions will arise
when the vehicle is in motion. Consequently, to
provide guidance for designers, researchers have
aimed to understand how the user-interface design
for a vehicle navigation system impacts on both
navigating and primary driving performance.
Specifically, research has aimed to answer the
following three design-oriented questions:

What Information Should a

Navigation System Provide?
To support route following, there are a wide range
of different information types that a system could
present, either referring to something real in the
road environment (junction representations,
street/road signs, landmarks, etc.) or indirectly
referring to or pointing at aspects of the environment (distance to turn, directions, etc.). In this
respect, researchers have established through a
range of methodologies that the use of distinctive
features of the environment (landmarks) within
navigation instructions (e.g., turn right at the
church) offer considerable advantages over the
use of distance to turn information (e.g., turn
right in 300 meters Burnett, 2000; Ross, May &
Grimsley, 2004). Moreover, research has identified the fundamental characteristics of landmarks
which designers of vehicle navigation systems
and providers of underlying map databases must
consider in choosing appropriate landmarks for
presentation by a navigation system (Burnett,
Smith & May, 2001).

Designing and Evaluating In-Car User-Interfaces

How Should Information be

Presented?
Navigation and related information has to be
presented to the driver in some way, and there has
been considerable research on a range of topics.
One key concern has been the impact of system
modality (voice and/or visual) on driving and
navigating performance. The general consensus
here is the primary modality for presentation of
navigation instructions should be auditory to
reduce the conflict with the predominately visual
driving task. However, information should also
be presented visually, in particular, to support
drivers understanding of more spatially complex
maneuvers which cannot be represented easily in
voice directions (Ross et al., 1995). Recently, Van
Erp (2005) investigated empirically the potential
for the use of passive touch as a novel modality
for presentation of navigation instructions (specifically, vibro-tactile direction and distance to
turn presented through the drivers seat). They
concluded that haptic navigation displays offer
various advantages over visual displays, for
example, they provide a private display to the
driver appropriate for very simple maneuvers.
Nevertheless, it must be noted that the authors did
not make comparisons with the prevailing visual
and auditory interfaces. Other research related to
information presentation has considered a wide
range of issues, such as the format of information
(map-based vs. turn-by-turn based), the scheduling of information (when to present instructions),
and the location of information (positioning of
displays). On these topics, the reader is directed
to Ross et al. (1995) and Srinivisan (1999).

How Should Drivers Interact with a

Navigation System?
For drivers (or passengers) to interact with a vehicle navigation system, there must be a means
by which they can enter data (e.g., postcode for an
address), select from continuous/discrete options

(e.g., voice volume levels, stored destinations),

request/repeat information (e.g., voice directions),
and move through the system (e.g., within and
between menu screens). There is understandably a natural tendency for designers to utilise
the familiar desktop computing paradigms, thus
utilizing specific hardware devices (e.g., joysticks,
touchscreens, buttons) and associated software
approaches (e.g., use of menus, lists, scrolling).
Historically, such paradigms were conceived as a
means of overcoming the significant limitations
of command-line user-interfaces and provided
a what-you-see-is-what-you-get (WYSIWYG)
experience for the user (Shneiderman, 1998). In
the driving context, several studies have shown
that such highly visual-manual user-interfaces
can have a considerable impact on safety (Nowakowski, Utsui & Green, 2000; Tijerina, Palmer
& Goodman, 1998).
As an alternative to such user-interfaces,
speech shows promise as a largely non-visual/
manual input method for navigation systems
(Tsimhoni, Smith, & Green, 2002). Nevertheless,
research has also shown that there is considerable
potential for cognitive distraction with speech
interfaces (Grtner, Knig, & Wittig, 2001), and
it is critical that recognition accuracy is very high.
Moreover, designers must provide clear dialogue
structures, familiar vocabulary, strong feedback
and error recovery strategies. These issues are
of particular importance given the potentially
large number of terms (e.g., towns, street names)
that might be uttered and the difficulties that a
speech recognition system can experience with
alphabet spelling (specifically, the e-setb, c,
d, e, g etc.).
Recent research has also shown the potential for
handwriting recognition in a driving context for
inputting alphanumeric data (Burnett et al., 2005;
Kamp et al., 2001). Whilst handwriting requires
manual input, there is a reduced cognitive component and it is a more familiar method for users
in contrast with speech interfaces. Nevertheless,
issues relating to recognition accuracy remain

545

Designing and Evaluating In-Car User-Interfaces

and it is critical to place a handwriting touchpad

in a location that facilitates the use of a drivers
preferred hand (Burnett et al., 2005).
The difficulties for complex interactions with
vehicle navigation systems are considered to be so
significant that many authors believe that systems
should disable overly demanding functionality
when the vehicle is in motion (e.g., by greying
out options when the vehicle is moving. Burnett,
Summerskill & Porter, 2004; Green, 2003). This
is currently a rich area for research, requiring an
understanding of (a) what is meant by overly
demanding, (b) establishing valid/reliable metrics for the assessment of demand and finally,
c) deciding where to put limits on acceptability
(Burnett et al., 2004).

Underload for Vehicle Navigation

Systems
In contrast with the overload perspective, over
the last five years some researchers have viewed
navigation systems as a form of automation,
where underload issues become central. Vehicle
navigation systems calculate a route for a driver
according to pre-defined algorithms and then
present filtered information, often via visual and
auditory instructions. Two related concerns are
emerging as important research questions, of
particular relevance to user-interfaces which place
a reliance on turn-by-turn guidance.
Firstly, it has been noted that there may be a
poor calibration in the perceived versus objective
reliability of in-car computing systems (Lee &
See, 2004). This is of relevance as a navigation
system (particularly the underlying digital map)
that is unlikely ever to be 100 percent reliable.
Nevertheless, drivers, largely based on their
accumulated experience, may believe this to be
the case. In certain situations, such overtrust in
a system (commonly referred to as complacency)
may lead to drivers following inappropriate routes
and potentially making dangerous decisions, for
instance, turning the wrong way down a one-way

546

street. There is plenty of anecdotal evidence for

such behavior in the popular press (e.g., http://
www.timesonline.co.uk/article/0,,2-2142179,00.
html, September 2006). Recently, research has
replicated the effect in a simulated environment
and indicated that there are considerable individual
differences in the likelihood of a driver showing
a complacency effect (Forbes & Burnett, 2007).
Further research is considering what role the
extended user-interface (training procedures,
manuals, marketing information) can have in
reducing complacency effects.
Secondly, drivers who use vehicle navigation
systems may not develop a strong mental representation of the environments in which they
travel, commonly referred to as a cognitive map.
It has been stressed that traditional methods (e.g.,
using a paper map) require drivers to be active
in the navigation task (route planning and following. Jackson, 1998; Burnett and Lee, 2005).
Whilst the demands (particularly the cognitive
demands) can initially be high, drivers who are
engaged are able to develop landmark, then route
knowledge, ultimately progressing to a map-like
mental understanding (survey knowledge). Such a
well-developed cognitive map means that drivers
are able to navigate independent of any external
source of information. Empirical research in this
area has shown that drivers using current forms
of user-interface for vehicle navigation system
do indeed experience reduced environmental
knowledge in relation to drivers using traditional
methods (Burnett & Lee, 2005; Jackson, 1998). A
key research question here is how user-interfaces
can be developed which balance the need for low
demands (workload) whilst simultaneously aiding
drivers in developing a well formed cognitive map
(Burnett & Lee, 2005).

FUTURE TRENDS AND

CONCLUSION
The incessant growth in the use of cars and worries about road safety have led car manufacturers

Designing and Evaluating In-Car User-Interfaces

to offer more intelligent cars providing a range

of novel functions to drivers. Moreover, existing
mobile technologies such as PDAs, MP3 players,
mobile phones, and so on, are increasingly being used within cars, as drivers seek to be more
productive and to enjoy the time spent in their
vehicles.
All of these computing-based systems offer potential benefits to drivers. This chapter has focused
on some key design issues for user-interfaces from
the perspective of the individual driver. However,
as systems become commonplace within vehicles,
there are fundamental conflicts to resolve between
the requirements of an individual versus the overall
traffic system. In this respect, the design of an
in-car computing user-interface will be a critical
consideration. As an example scenario, one can
envisage many drivers using information systems
providing the same information at the same time.
Such a situation may lead to a range of problems,
for instance the use of roads not designed for high
volumes of traffic. Clearly, there is a need for
overall management and an understanding of the
impact that specific styles of user-interface will
have on driver behavior.
A second broad issue for research concerns
the interaction between multiple systems. This
chapter has introduced the overload and underload
concepts and discussed them in turn relating them
to different individual systems. It is highly likely
that in the short to medium term, overload will be
given a prominent position in research and development work, whereas underload will emerge as
an increasingly important topic in the medium to
long term. However, this singular view neglects
the fact that information and control-based systems are likely to be used together in a vehicle.
Clearly, there will be various interaction effects
for researchers to investigate. Moreover, there is a
fundamental need to find the right balance between
the two extremes of overload and underload. As
noted by Dewar (2002, p. 330), humans operate
best at an optimal level of arousal, and either too

much or too little workload can be detrimental

to performance.
The development of suitable methods for
designing and evaluating in-car computing
user-interfaces will continue to be an important
research topic. Reliable and valid methods are
required which are accepted within industry. A
key motivation will be to establish quick and
dirty methods (and associated metrics) enabling
designers to understand the likely demands of
their user-interfaces early in the design process
when very rudimentary prototypes are available.
A further critical requirement is for benchmarking, that is, establishing a point of reference
from which user-interfaces can be compared or
assessed. Such benchmarks will be of particular
benefit when identifying user-interface designs
that are considered acceptable or unacceptable,
particularly from a safety perspective.

REFERENCES
Baumann, M., Keinath, A., Krems, J.F., & Bengler, K (2004). Evaluation of in-vehicle HMI
using occlusion techniques: Experimental results
and practical implications. Applied Ergonomics,
35(3), 197-205.
Bishop, R. (2005). Intelligent vehicle technology
and trends. London: Artech House Publishers.
Blaauw, G.J. (1982). Driving experience and
task demands in simulator and instrumented
car: A validation study. Human Factors, 24(4),
473-486.
Burnett, G.E. (2000). Turn right at the traffic
lights. The requirement for landmarks in vehicle
navigation systems. The Journal of Navigation,
53(3), 499-510.
Burnett, G.E. (2003). A road-based evaluation
of a head-up display for presenting navigation
information. In Proceedings of HCI International

547

Designing and Evaluating In-Car User-Interfaces

conference (Human-Centred Computing), Vol.

3, (pp. 180-184).
Burnett, G.E., & Joyner, S.M. (1997). An assessment of moving map and symbol-based route
guidance systems. In Y.I. Noy (Ed.), Ergonomics
and safety of intelligent driver interfaces (pp.
115-137). Lawrence Erlbaum.
Burnett, G.E., & Lee, K. (2005). The effect of
vehicle navigation systems on the formation of
cognitive maps, In G. Underwood (Ed.), Traffic
and transport psychology: Theory and application (pp. 407-418). Elsevier.
Burnett, G.E., Lomas, S., Mason, B., Porter, J.M.,
& Summerskill, S.J. (2005), Writing and driving:
An assessment of handwriting recognition as a
means of alphanumeric data entry in a driving
context. Advances in Transportation Studies, 2005
Special Issue, 59-72.
Burnett, G.E., Smith, D., & May, A.J. (2001).
Supporting the navigation task: Characteristics
of good landmarks. In Hanson, M.A. (Ed.),
Proceedings of the Annual Conference of the
Ergonomics Society (pp. 441-446). Taylor &
Francis.
Burnett, G.E., Summerskill, S.J., & Porter, J.M.
(2004). On-the-move destination entry for vehicle navigation systems: Unsafe by any means?
Behaviour and Information Technology, 23(4),
265-272.
Burns, P. (1999). Navigation and the mobility
of older drivers. Journal of Gerontology. Series
B, Psychological sciences and social sciences,
54(1), 49-55.
Dewar, R.E. (2002). Vehicle design. In R.E. Dewar
& P.L.Olson (Eds.), Human factors in traffic
safety, (pp. 303-339). USA: Lawyers & Judges
Publishing Company, Inc.
Engstrom, J., Aberg, N., & Johansson, E. (2005).
Comparison between visual and tactile signal
detection tasks applied to the safety assessment

548

of in-vehicle information systems. In Proceedings

of the third international driving symposium on
Human Factors in driver assessment, training
and vehicle design. (University of Iowa public
policy center, Iowa).
Forbes, N., & Burnett, G.E. (2007). Can we
trust in-vehicle navigation systems too much? In
preparation, School of Computer Science and IT,
University of Nottingham, UK.
Grtner, U., Knig, W., & Wittig, T. (2001).
Evaluation of manual vs. speech input when using a driver information system in real traffic. In
Proceedings of International driving symposium
on human factors in driver assessment, training
and vehicle design. (University of Iowa public
policy center, Iowa), Available at: http://ppc.uiowa.
edu/driving-assessment//2001/index.html
Gould, J. D., & Lewis, C. H. (1985). Designing for
usability: key principles and what designers think.
Communications of the ACM, 28(3), 300-311.
Green, P. (1999). The 15 second rule for driver
information systems. In Proceedings of the ITS
America 9th annual meeting, Washington, D.C.
Green, P. (2003). Motor vehicle driver interfaces.
In J. A. Jacko & A. Sears (Eds.), The human
computer interactions handbook (pp. 844-860).
UK: Lawrence-Earlbaum Associates.
Green, P. (2004). Driver Distraction, telematics
design, and workload managers: Safety issues
and solutions. (SAE paper 2004-21-0022). In Proceedings of the 2004 International Congress on
Transportation Electronics (Convergence 2004,
SAE publication P-387), (pp. 165-180). Warrendale, PA: Society of Automotive Engineers.
Hackos, J. T., & Redish, J. C. (1998). User and
task analysis for interface design. NY: John Wiley
and Sons.
ISO. (2005). Road vehiclesErgonomic aspects of
transport information and control systemsOcclusion method to assess visual distraction due

Designing and Evaluating In-Car User-Interfaces

to the use of in-vehicle information and communication systems. Draft International Standard
ISO/DIS 16673. ISO/TC 22/SC 13.

shop on Driving Simulator Scenarios, October

2003 - Dearborn, Michigan. Retrieved from www.
nrd.nhtsa.dot.gov/IHRA/ITS/MATTES.pdf

Jackson, P. G. (1998). In search of better route

guidance instructions. Ergonomics, 41(7), 10001013.

Michon, J.A. (1985). A critical view of driver

behaviour models. In L. Evans & R.S. Schwing
(Eds.), Human behaviour and traffic safety. NY:
Plenum Press.

Jamson, H. (2001). Image characteristics and their

effect on driving simulator validity. In Proceedings of International driving symposium on Human Factors in driver assessment, training and
vehicle design. (University of Iowa public policy
center, Iowa).
Jones, W.D. (2002). Building safer cars. IEEE
Spectrum, 39(1), 82-85.
Kamp, J.F., Marin-Lamellet, C., Forzy, J.F., &
Causeur, D. (2001). HMI aspects of the usability of
internet services with an in-car terminal on a driving simulator. IATSS Research, 25(2), 29-39.
Kaptein, N. A., Theeuwes, J., & Van der Horst, R.
(1996). Driving simulator validity: Some considerations. Transp. Res. Rec., 1550, 3036.
Kennedy, R. S., Lane, N. E., Stanney, K. M.,
Lanham, S., & Kingdon, K. (2001, May 8-10).
Use of a motion experience questionnaire to
predict simulator sickness. In Proceedings of HCI
International, New Orelans, LA.
Lee, J.D., & See, K. (2004). Trust in automation:
Designing for appropriate reliance. Human Factors, 46(1), 50-80.
Maguire, M. (2001). Context of use within usability activities. International Journal of Human-Computer Studies, 55, 453-483.
Markkula, G., Kutila, M., & Engstrm, J. (2005,
November,14 - 16). Online Detection of Driver
Distraction - Preliminary Results from the AIDE
Project. Proceedings of International Truck & Bus
Safety & Security Symposium Alexandria, VA.
Mattes, S. (2003). The lane change task as a tool
for driver distraction evaluation. IHRA-ITS Work-

Nichols, S., & Patel, H. (2002). Health and safety

implications of virtual reality: A review of empirical evidence. Applied Ergonomics, 33(3),
251-271.
Nowakowski, C., Utsui, Y., & Green, P. (2000).
Navigation system evaluation: The effects of
driver workload and input devices on destination
entry time and driving performance and their
implications to the SAE recommended practice.
(Tech. Rep. UMTRI-2000-20). USA: The University of Michigan Transportation Research
Institute.
Parkes, A.M. (1991). Data capture techniques for
RTI usability evaluation. In Advanced telematics
in road transportThe DRIVE conference: Vol.
2. (pp. 1440-1456). Amsterdam: Elsevier Science
Publishers B.V.
Peters, G.A., & Peters, B.J. (2002). Automotive
vehicle safety. London: Taylor and Francis.
Pettitt, M.A., & Burnett, G.E. (2005, November).
Defining driver distraction. In Proceedings of
World Congress on Intelligent Transport Systems.
San Francisco.
Pettitt, M.A., Burnett, G.E., Bayer, S., & Stevens,
A. (2006). Assessment of the occlusion technique
as a means for evaluating the distraction potential
of driver support systems. IEE Proceedings in
Intelligent Transport Systems, 4(1), 259-266.
Pettitt, M.A., Burnett, G.E., & Karbassioun, D.
(2006). Applying the Keystroke Level Model in a
driving context. In Bust, P.D. (Ed.), Proceedings
of the Ergonomics Society Annual conference,

549

Designing and Evaluating In-Car User-Interfaces

Taylor & Francis, Contemporary Ergonomics

2006 (pp. 219-223).
Pettitt, M.A., Burnett, G.E., & Stevens, A. (2007).
An extended Keystroke Level Model (KLM)
for predicting the visual demand of in-vehicle
information systems. In Proceedings of the ACM
Computer-Human Interaction (CHI) conference.
May 2007, San Jose, CA.
Preece, J., Rogers, Y., & Sharp. H. (2002). Interaction design: Beyond human-computer interaction.
NY: John Wiley and Sons.
Reed, M.P., & Green, P.A. (1999). Comparison of
driving performance on-road and in a low-cost
simulator using a concurrent telephone dialling
task. Ergonomics, 42(8). 1015-1037.
Ross, T., May, A.J., & Grimsley, P.J. (2004). Using traffic light information as navigation cues:
implications for navigation system design. Transportation research Part F, 7(2), 119-134.
Ross, T., Vaughan, G., Engert, A., Peters, H.,
Burnett, G.E., & May, A.J. (1995). Human factors guidelines for information presentation by
route guidance and navigation systems (DRIVE II
V2008 HARDIE, Deliverable 19). Loughborough,
UK: HUSAT Research Institute.
Shneiderman, B. (1998). Designing the user-interface (3rd edition). Addison Wesley.
Srinivisan, R. (1999). Overview of some human
factors design issues for in-vehicle navigation and
route guidance systems. Transportation Research
Record 1694 (paper no. 99-0884). Washington,
DC: National Academy Press.
Society of Automotive Engineers (SAE) (2000).
Navigation and Route Guidance Function Accessibility While Driving (SAE Recommended
Practice J2364), version of January 20th, 2000.
Warrendale, PA: Society of Automotive Engineers.

550

Stevens, A., Bygrave, S., Brook-Carter, N., &

Luke, T. (2004). Occlusion as a technique for
measuring in-vehicle information system (IVIS)
visual distraction: a research literature review.
Transport Research Laboratory (TRL) report no.
TRL609. Crowthorne, Berkshire: TRL.
Tijerina, L., Palmer, E., & Goodman, M.J. (1998).
Driver workload assessment of route guidance
system destination entry while driving. (Tech.
Rep. UMTRI-96-30). The University of Michigan
Transportation Research Institute.
Toffin, D., Reymond, G., Kemeny, A., & Droulez,
J. (2003, October 8-10). Influence of steering wheel
torque feedback in a dynamic driving simulator.
In Proceedings of the Driving Simulation Conference North America. Dearborn, USA.
Tornros, J. (1998). Driving behaviour in a real and
a simulated road tunnelA validation study. Accident Analysis and Prevention, 30(4), 497-503.
Transport Canada. (2006, June). R Occlusion
Research. Presentation given to the AAM driver
focus task force. Detroit.
Tsimhoni, O., Smith, D., & Green, P. (2002).
Destination entry while driving: Speech recognition versus a touch-screen keyboard. (Tech. Rep.
UMTRI-2001-24). The University of Michigan
Transportation Research Institute.
Tufano, D.R. (1997). Automotive HUDs: The
overlooked safety issues. Human Factors, 39(2),
303-311.
Van Erp, J.B.F. (2005). Presenting directions
with a vibrotactile torso display. Ergonomics,
48(3), 302-313.
Wallen Warner, H.M., & Aberg, L. (2005). Why
do drivers speed? In Underwood, G. (Ed.), Traffic
and transport psychology: theory and application
(pp. 505-511). Oxford, UK: Elsevier.
Ward, N. J., & Parkes, A.M.. (1994). Head-up
displays and their automotive application: An

Designing and Evaluating In-Car User-Interfaces

overview of human factors issues affecting

safety. Accident Analysis and Prevention, 26(6),
703-717.
Wickens, C.D., Lee, J.D., Liu, Y., & Becker,
S.E.G., (2004). An introduction to human factors engineering (2nd Edition). New Jersey, USA:
Pearson.
Wikman, A., Nieminen, T., & Summala, H. (1998).
Driving experience and time-sharing during incar tasks on roads of different width. Ergonomics,
41(3), 358-372
Young, K., Regan, M., & Hammer, M. (2003).
Driver distraction: a review of the literature.
Monash University Accident Research Centre,
Report No. 206, November 2003.

KEY TERMS
Driver Distraction: Occurs when there is a
delay by the driver in the recognition of information necessary to safely maintain the lateral and
longitudinal control of the vehicle. Distraction may
arise due to some event, activity, object or person,
within or outside the vehicle that compels or tends
to induce the drivers shifting attention away
from fundamental driving tasks. Distraction may
compromise the drivers auditory, biomechanical,
cognitive or visual faculties, or combinations
thereof (Pettitt & Burnett, 2005).
Driving Simulators: Provide a safe, controlled
and cost-effective virtual environment in which
research and training issues related to driving can
be considered. Simulators vary considerably in

their fidelity (i.e., the extent to which they replicate

aspects of real driving).
In-Car Computing Systems: Provide information to support the driving task or control some
aspect/s of the driving task. In-car computing
systems may also provide information and/or
services that are unrelated to driving.
Keystroke Level Model: Is an established
HCI method used to predict experts task times
with a user-interface. It can be used with in-car
user-interfaces to predict static task time, that
is, the time taken to achieve tasks in a stationary
vehicle. Recently, the KLM has been extended
to predict visual demand measures related to the
occlusion protocol.
Overload: (Due to in-car computing systems)
occurs when a drivers information processing
resources are overwhelmed and performance on
primary driving tasks inevitably suffers.
Underload: (Due to in-car computing systems) occurs when automation of core driving
tasks (such as steering, braking, etc.) has led to
a situation in which driving performance has
deteriorated. This may have arisen because the
driver has reduced awareness of other road users,
has changed their behavior in negative ways or
has inferior skills/knowledge in driving.
The Occlusion Protocol: Is a user trial
method used in the design and evaluation of incar user-interfaces. Participants typically wear
LCD glasses which restrict the visual experience
by only enabling short (e.g., 1.5 seconds) chunks
of visual attention with an in-car user interface.
Measures related to the visual demand of an
interface can be established.

This work was previously published in Handbook of Research on User Interface Design and Evaluation for Mobile Technology,
edited byJ. Lumsden, pp. 218-236, copyright 2008 by Information Science Reference, formerly known as Idea Group Reference
(an imprint of IGI Global).

551

552

Chapter 2.12

The Semiotics of Smart

Appliances and Pervasive
Computing
Peter Bgh Andersen
University of Aarhus, Denmark
Martin Brynskov
Center for Interactive Spaces, University of Aarhus, Denmark

Abstract
This chapter presents digital habitats, a conceptual
and methodological framework for analyzing
and designing smart appliances in the context
of pervasive computing. The concrete topic is
a project in pervasive gaming for children. The
framework consists of a set of theoretical concepts supplemented by diagrams for representing
semiformal models. We give a short overview
of selected theories of play and gaming and apply the framework to an implemented simple
pervasive game. Finally, we use the framework
in a constructive manner to produce a concrete
design of a new game. The result is discussed
and compared to other approaches. The main
points are the following: (a) it can describe communicative as well as material acts plus the way
they hang together; (b) it provides an explicit
link between human activities and their spatial

context; (c) it has an explicit dynamic model that

precisely describes the conditions for executing
actions; and (d) it offers a typology of participant
roles based on linguistic theory, which supports
design processes.

Introduction
In this chapter, we will present an approach to
analysis and design of computing systems that
transcends the boundaries of traditional office
computer systems such as PCs and laptops. These
transcending systems are called ambient, ubiquitous, or pervasive computing systems, and they
pose new challenges to the way we understand,
analyze, and design information technology.
With such systems, computing power spreads
from dedicated computing hardware into other
artifacts and places, both at the workplace and

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

The Semiotics of Smart Appliances and Pervasive Computing

in everyday life. Microcontrollers, sensors, and

actuators have been embedded in machines for
decades, but the functionality was tied closely
to the artefact in which it was embedded (e.g., a
washing machine or the side mirror of a car), and
therefore, the computational smartness was not
foregrounded in itself. Two factors have changed
this: (a) the increasing flexibility and computing power of smaller-scale devices, and (b) the
wireless networking capabilities and structured
exchange of information. In this world of smart
phones, GPS (location tracking), and software
agents, we need concepts to communicate about
human needs and activities as well as technical
infrastructures. Digital habitats is a suggestion for
such a framework. In our presentation, we focus
on fun and games, but the framework originally
was conceived in a workplace setting.
On the Concept of Intelligence discusses
the concept of intelligence and concludes that the
everyday use of the concept applies to a disposition to act in a certain way in a network of other
actors and artefacts. To be intelligent is to fill out
Agent roles appropriately in activities conducted
by such networks.
The habitat concept defines the framework we
use. On the one hand, it defines activities in terms
of roles, participants, actions, and glue binding
participants to roles. The well-known automatic
door opener is used as an example. On the other
hand, it offers a maplike representation that ties
activities to physical and informational spaces
and describes the various types of interplay between physical and informational space. This is
particularly relevant to pervasive computing, since
pervasive computing is characterized by being
distributed in physical space and by overlaying
physical space with digital information.
In Play and Games, we give a short overview
of theories of play and gaming. In Designing
Pervasive Games, we adapt the framework to the
domain of games and describe an implemented
simple pervasive game called StarCatcher. The
Bogeyman puts all the ends together in a concrete

design of the game Bogeyman that elaborates on

StarCatcher by drawing on the presented theory.
Related Work compares this chapter to related
fields, and the conclusion summarizes the advantages of the present approach, as we see it.

On the Concept of
Intelligence
The first issue that must be discussed is what
smart and intelligent mean.

Intelligence
Since there is no universally accepted definition
of intelligence (Roth & Dicke, 2005), we accept
Gilbert Ryles (1970) claim that these words denote
(a) the manner in which an action is performed
and (b) a prediction about the way other actions
are performed. In this case, intelligent does not
denote a special mental process that is the cause
of the action but rather a disposition generally to
act in a certain manner. What is intelligent depends upon the circumstances but often involves
features such as: the action achieves its goal, it
does not contain superfluous steps, it economizes
with resources, it does not destroy or hurt participants, it is an innovative way of solving a difficult
problem, and so forth.
This definition is similar to a prevalent view
in contemporary cognitive science, that mental or behavioral flexibility is a good measure
of intelligence, resulting in the appearance of
novel solutions that are not part of the animals
normal repertoire (Roth & Dick, 2005, 250).
We choose to focus on the behavioral aspect and,
thus, preclude ourselves from making inferences
about neuroanatomy or mental mechanisms. On
the other hand, this choice to focus strictly on
behavior allows us to use the word about humans
as well as artifacts without committing ourselves
to philosophical doctrines about the nature of the
mind (see Dennett, 1991).

553

The Semiotics of Smart Appliances and Pervasive Computing

If intelligence denotes a disposition to act in

a certain way, then it follows that its unsophisticated and immediate reference is to networks
of actors, instruments and objects. The reason
is quite simply that human action normally is
mediated by an instrument and directed toward
an object. These elements form a network, and
it is such networks that we can immediately call
intelligent or smart. There will often be a next
analytical step in which we ascribe the main honor
to one or more of the participants in the network:
one of the actors, tools, or objects may be seen as
the main contributor.

Networks of Stupidity
This analysis is easier to verify with negative
predicates such as negligence, stupidity, and inability. For example, accident reports must point
to a participant that is guilty of the accident in
order to suggest future remedies against the type
of accident and because of insurance issues. Although accidents are mostly caused by a particular
configuration of participants in the network, the
report must point out the weak link in the chain.
However, this is often difficult and a matter of
interpretation.
Here is an example, due to PhD student Thomas
Koester: a ferry was fitted with a system that
compensated the heeling of the boat by moving
water in the ballast tanks. The system had a manual
and an automatic mode, but sometimes it would
unexpectedly go from automatic to manual mode.
The mode change was displayed on the bridge
and in a closed locker on the deck. The accident
occurred when the deck crew was emptying the
deck for its cargo of trucks and cars. When cargo
is removed from one side of the ship, it will heel,
and the system is supposed to compensate; in
this case, it had switched to manual. The result
was that the heeling was more than six degrees,
and the ramp was damaged. Who was to blame?
The crew? It knew about the fault; should they

554

have opened the locker and checked the mode

regularly? But then the disembarkation would
have gone more slowly. The manufacturer? The
system obviously should not be allowed to switch
to manual by itself. The ship owner? He could
have counteracted the fault by mounting a warning system on the deck, informing the deck crew
of the mode shift.
The fault is clearly due to the whole network:
a faulty system + missing indications of the error
+ lack of attention in the crew. It is a practical
and political matter, not a philosophical issue, to
decide on a remedy: correcting the mode error,
mounting a warning light on the deck, or changing the operation procedures.
The importance of networks vis vis individual actors has been emphasized in the actor
network theory (Latour, 1994, 1999; Law, 1987).
For example, power is a predicate that pertains
to an actors position in a network, not to the
individual actor.
The example illustrates the point made previously: predicates such as intelligent, smart, sloppy,
inefficient, and so forth are in the first place
predicates of networks, not of their parts. Only
analytically can one of its participants be singled
out as the main factor, and this choice depends
upon the countermeasures that are considered
feasible and desirable.
The same line of reasoning can be applied to
intelligence. The IBM chess program Deep Blue
beat grandmaster Kasparov on May 4, 1997.
Therefore, there is an intelligent network of actors
comprising at least the following participants:
the developers, the chess literature they used, the
system, and the operator. But which participant
should be picked as the winner of the match? It is
as difficult to decide as in the accident case, but
it was the development team behind Deep Blue
that took home the $700,000 first prize. Thus, in
practice, the development team was singled out
as the intelligent part of the network.

The Semiotics of Smart Appliances and Pervasive Computing

Intelligent Technology
If we are to single out one participant of an intelligently conducted activity, it must be because its
contribution is particularly conspicuous. How do
we decide this? One way is a simple substitution:
if we keep the chain constant and replace one
participant, does the performance of the chain
become more or less intelligent? If the chain
performs less intelligently, we may tentatively
attribute intelligence to this part of the chain.
From these arguments follows a definition of
smart/intelligent technology:
1.

Intelligent technology is a kind of technology that is able to contribute positively to

activities whose manner of performance we
will intuitively call intelligent.

In the following, we incorporate this definition

in a broader framework, which we call a habitat.
The habitat concept is a general attempt to link
networks of activities and actors to physical space
on the one hand and signs and sign usage on the
other hand.

The Habitat Concept

The habitat concept was elaborated in the research
group Frameworks for Understanding Software
Brynskov
Systems (Andersen & Nowack, 2004;
& Andersen, 2004;

May & Kristensen, 2003

).
The purpose of the research was to devise a
methodology that is better suited for handling
context-sensitive pervasive computing than are
traditional ones. Traditional object-oriented
modeling methods (Mathiassen, Munk-Madsen,
Nielsen, & Stage, 2001) are characterized by
two features: (a) they model conceptual structures while the physical system represented by
these models is absent in the method, and (b) the
physical location of the software components
is postponed to the last implementation phase

(Andersen & Nowack, 2002). This is a problem

in context-sensitive pervasive computing, because
one of its possibilities is the use sensors to collect
information about physical objects located in the
spatial or temporal vicinity of the physical device
where the software resides. Therefore, we need
to coin concepts for the information available
in the vicinity of the physical device and for the
physical objects referred to by its software model.
The former concept is called the access area; it
denotes the locations from where sensors can
access information; the latter is called the reference area and denotes the location of the objects
referred to by the software model. Furthermore,
a description of information is incomplete if we
do not specify its use (i.e., what activities is the
information used for?). Therefore, we need to
enter the notion of activities into our framework.
Finally, we are interested in systems that are sensitive to their physical surroundings, which means
that our third component must be space. We are
interested in physical spaces that are designed or
have evolved to support a delimited set of human
activities. This is true for an overwhelming number of the spaces in which we live daily: private
houses, airports, hospitals, railroad stations, road
networks, and so forth. We have chosen the term
habitat to denote a physical space that is designed
to support some set of activities and that provides
access to information about objects relevant to the
activities (Brynskov & Andersen, 2004).
2.

A habitat is a chunk of space-time that is

designed or has evolved to support a delimited set of activities by offering physical
artifacts and information sources useful for
conducting the activities.

From this definition, it follows that a habitat

should be characterized along three dimensions:
3.

The physical habitat: The physical layout

and boundaries of the habitat plus the available physical artifacts.
555

The Semiotics of Smart Appliances and Pervasive Computing

The informational habitat: The signs

available (access and reference area) to
participants in the activities (digital and
nondigital signs).
The pragmatic habitat: The action affordances offered by the habitat, the macroroles, and the role-requirements of the
participants.

4.

5.

The contribution of the present chapter is threefold: (a) it explores the usefulness of the habitat
concept in the domain of pervasive games; (b) it
presents a framework for describing communicative and material activities and the way they
interact; and (c) it presents a notion of intelligent
or smart technology that is consistent with the
framework. The chapter mostly draws on semiotic
and linguistic theory, and our basic understanding
of the habitat concept is well-captured in Peirces
semiotic triangle (Figure 1). The space itself and
its manufactured representations (e.g., signposts
and electronic displays) are representamens; the
interpretant of these signs is the activities associated to the habitat, and the object is the phenomena
inside the reference area (i.e., things or events that
are relevant to the activities).

Activities
In order to use definition (1) for design purposes,
we need a general functional definition of activi-

Figure 1. The habitat as a sign

Interpretant:
activities

representamen:
Space and representations in
access area

556

Object:
Phenomena in reference
area

ties, and a set of context-dependent definitions of

what criteria intelligent activities must meet in
various domains. The general definition is given
in this section. The game-specific adaptation is
presented in the section Applying the Theory
to Games.

Goal, Roles, Participants, and Actions

As in the case of intelligence, we adapt a functional view of activities (Andersen, 2004a, 2004b,
2005; Bdker & Andersen, 2005; Brynskov &
Andersen, 2004). The conceptual framework
consists of roles, participants, actions, and activities. Activities consist of actions subsumed
under a shared goal, and participants play roles
in relation to actions and activities.
At the action level, we use traditional linguistic semantic roles like agent, theme, instrument,
beneficiary, source, destination, time, place,
direction, and so forth (see Fillmore, 1968, 1977;
Blake, 2001). At the activity level, we use macro
roles like librarian/borrower, customer/clerk,
lawyer/client, judge/defendant, and so forth. The
macro roles are defined by the semantic roles that
participants can play in actions: thus, a Judge is
the Agent of actions like planning the sessions
of the court, considering the judgment, giving
a verdict, sentencing, and so forth. This is the
method used in the classical work on narratology
(Propp, 1968).
In addition to specifying actions, roles, and
activities, we need to specify the goal of the activity. We distinguish between four types: creating,
destroying, preventing, and maintaining a state
(Lind, 1994).
Not all participants are equally qualified to
play these roles; a role requires its filler to possess
the requisite incentives and qualifications in order
for it to contribute in an intelligent and competent
manner. Incentives include intentions, desires,
and obligations; qualifications cover the relevant
abilities, rights, and knowledge. Sometimes we

The Semiotics of Smart Appliances and Pervasive Computing

will refer to these qualifications as a lump of glue

that binds a participant to a role.
Some of these requirements are realized differently in humans and nonhumans. For example,
intentions are implemented as negative feedback
loops in automatic systems like thermostats that
aim at keeping a certain process variable (e.g.,
the heat) locked at a certain set-point. Its goal is
to maintain the process variable at the set-point.
In software agents (Russell & Norvig, 2003),
intentions often are implemented as some kind
of means/end structure along with a planning
and execution algorithm. The goal often will be
to create a certain state, but in order to do so,
the algorithm will have to maintain other states
(protected states) and prevent others (forbidden
states).
According to the approach outlined previously, we should not necessarily ask whether
this implementation resembles the way humans
make and execute plans. Instead, we should ask
for qualifications: to which degree is the behavior
of the automatic system suited for playing its assigned roles in the activity? The same goes for
knowledge: there are many types of knowledge
representations (predicate calculus, semantic
networks, frame systems, conceptual graphs, and
neural networks), but for the present purpose, it
is not interesting whether humans use something
comparable but only to which degree the system
component using these methods is able to play
its allotted role. The viewpoint, thus, is strictly
behavioral.
In this chapter, we will concentrate on cooperative activities involving humans, intelligent
systems, and dumb artefacts. Since the activity
is cooperative, the following requirements of
intelligent agents are important:
6.

7.

The agent should have intentions to act that

are understandable to the others (comprehensibility);
Agents should display their intentions and
actions in a way that is accessible to the
other participants (accessibility);

8.

9.

Agents should be able to perceive and

manipulate the attention-focus of their
colleagues and adapt their own actions accordingly ( joint attention);
Agents should be able to enter and leave the
activity (intermittent participation).

Applied to mechanical agents, comprehensibility (6) means that they only should form plans that
are understandable to their human and nonhuman
colleagues. Understandable algorithms have a
higher priority than cunning ones, but there is a
tradeoff between intelligibility and efficiency.
The intentions of the agent should be signaled
to the other participants in a way that allows them
to reliably infer intentions from behavior (accessibility, 7). In other words, there is a mutual
commitment to signal the truth, the whole truth,
and nothing but the truth in the least complicated
manner (equivalent to the so-called Gricean
maxims [Grice, 1975] from the pragmatics of
natural language). On the other hand, once the
deliberate representation of intentions has been
introduced, it is possible to misuse it to lie about
intentions as well.
Attention is intentionally directed perception
(Gibson & Rader, 1979; Tomasello, 1995). Joint or
shared attention (8) is necessary for cooperation:
If I cannot see what my colleague is concentrating
on, I cannot collaborate with him or her. However,
shared attention is more than just looking at the
same object or detecting gaze direction; it is the
mutual understanding between two intentional
agents that they are, indeed, intentional agents
(Tomasello, 1995) (see Dennetts, 1987, The
Intentional Stance). In order for me to share a
goal-directed activity with somebody, I not only
must be sure that the other participants have their
minds focused on the objects or topics I think of
myself, but all participants also must share the
fundamental assumption that activities can be
goal-directed and that they can be coordinated
by detecting and manipulating attention. In fact,
shared attention is a constituent factor in cultural

557

The Semiotics of Smart Appliances and Pervasive Computing

learning (Tomasello, Kruger, & Ratner, 1993).

Humans share attention all the time, beginning
near the end of their first year of life as a result
of a constellation of emerging social behaviors
(e.g., communicative gestures, imitation, social
referencing) (see Tomasello, 1995). We can read
or infer and, therefore, continuously follow and
manipulate each others focus of attention (and
thereby intention) throughout a joint activity.
This interpreted focus is a common anchor point
without which two persons would drift away from
each othermentally and physicallyinstead of
staying focused on the same subject matter. They
would be solipsist nomads.
Intermittent participation (9) means that
participants must be able to leave and enter the
network without its breaking down. Much pervasive technology, including games, must allow
participants to enter and leave without notice. For
example, a door opener would be unusable if the
pedestrian were to type in complicated commands
preparing the door for the event before walking
through the door and signing off after the passage.
The ferry system in Networks of Stupidity was not
sufficiently robust to allow the mechanical agent
to leave the network without notice; the network
broke down and damaged the ramp.
Joint attention and intermittent participation
are also important in our concrete topic: childrens
games. When children play, it is important for
them to keep track of the others focus. Children
gradually develop this sense of playing or doing
things together as opposed to just doing things
in parallel. At the age of two, they have all the
prerequisites for joint attention (Kaplan & Hafner, 2004), and they continue to develop still
more complex social interaction based on this
fundamental competence. Also, children enter
and leave activities very often (i.e., their participation is intermittent and highly dependent on
the environment). They have to go and eat, must
go outside, or one participant is picked up by his
parents.

558

An example of a game that supports intermittent participation is LEGO Star Wars (www.
lego.com/starwars), which allows players to join
and leave a game in progress so that a parent can
step in and help a child but leave again without
having to play the entire game. When a player
leaves, the character continues as an independent
agent controlled by the software (at least one
player must be controlled by a human, however;
otherwise, the game ends). Massively Multiplayer
Online Games (MMOGs) (e.g., World of Warcraft
[www.worldofwarcraft.com]) are another example
of games built around intermittent participation. Mobile MMOGs exist, too, and are called
3MOGs (e.g., Undercover 2: Merc Wars [www.
undercover2.com]). Often, children can be seen
not only as mobile but also as nomadic, because
they are dependent on the resources and partners
offered by the environment (see the following
section). In this case, the proximate environment
has a marked influence on their patterns of play
(Brynskov, Christensen, Ludvigsen, Collins, &
Grnbk, 2005).
The problem of sharing attention between humans and machines is that machines do not have
the ability to read attention, let alone intentional
stance (see, however, Kaplan & Hafner, 2004). In
the ferry incident, one of the problems was also
the lack of joint attention. The automatic pump
system switched to manual operation without
monitoring whether this important change was
brought to the attention of the crew. The crew, on
the other hand, could not monitor the attention of
the pump system and, therefore, failed to realize
that they were supposed to operate the pumps
manually. Thus, intermittent participation seems
to require joint attention.
People use language to a large extent to signal
their focus of attention. With machines in the
loop, this is not possible, not because computers
cannot produce or parse speech at all (they can
to some extent), but because their underlying
representation of attention and states do not map
easily onto their human partners. This is because

The Semiotics of Smart Appliances and Pervasive Computing

the semantics of a language is not context-free

but grounded in experience. Humans seem to
take the others perspective by mapping observations onto their own nervous system in a kind of
simulation. Therefore, joint attention presupposes
understandable algorithms.
If we still want humans and machines to
share attention during activities, we have two
options:

We can try to develop epigenetic artificial

intelligence that can learn to share attention
by means of contextual grounding, imitating
human cognition (although it is still an open
question how exactly to achieve this) (see
Steels, 2003; Kaplan & Hafner, 2004), or
We can try to create a fixed, hardcoded, and
more simplistic common ground where both
humans and machines can read and interpret
behaviors.

We choose the latter, and one of the methods

is to design comprehensible algorithms so that
there is something for the human participants to
share. As a consequence, one might say that the
burden of flexibility, to a large extent, is put on
the shoulders of the human participants.
The challenge of handling attention focus
and intermittent participation is not at all new in
human-computer interaction (HCI), especially in
the field of computer-supported cooperative work
(CSCW) (Bardram & Hansen, 2004). There is a
difference, though, in signaling an applications
states and functions as a tool or medium, as is
the traditional case in HCI, or a persons availability in CSCW, to signaling an agents states.
The difference is whether we attempt to represent
intentions (Tomasello, 1995). Attentive agents
and calm systems that work in a sort of symbiosis
with their owner and sense or guess his or her
intentions is an old idea (Licklider, 1960; Weiser
& Brown, 1995), and it is still being pursued
(Maes, 1994; Maglio & Campbell, 2003). As already mentioned, participants in our framework

are glued to their roles not only by intentions

but also by desires, abilities, rights, obligations,
and knowledge, which also, to a higher or lesser
degree, indicate intentions. Therefore, a minimal
requirement is to signal these binders not only
in a form that humans can interpret but also the
other way around. We see play as an interesting
case, since this is an important part of the situations in which children learn the intricacies of
the semiotics of behavior.

Door Openers
In this section, we illustrate the concepts used
to model activities by a widespread technology;
namely door openers. In addition, we offer some
simple diagramming techniques. Door openers
are good examples because they involve networks
of humans and nonhumans, and nonhumans play
an Agent role in the sense that they initiate and
monitor activities.
The activity concept is the one described
previously. We use a diagram consisting of actions, and we highlight the relations between the
actions by graphical means. Two actions can be
connected by arrows signifying dependencies

Figure 2. Automatic door opener in shopping

mall

559

The Semiotics of Smart Appliances and Pervasive Computing

between participating in the two actions. The

rules are as follows:
10. An activity is executed if all participants are
able to fulfill their roles to some degree and
if the Agent filler is strongly obligated and/or
strongly desires to fulfill his or her role. In
the game described in The Bogeyman
section, desires are created by application of
dissonance theory (Eskola, 1973), whereas
ability is created by the participants location. Whenever the necessary participants
are assembled in the same place, the activity
is enabled.
11. If something successfully participates as role
A in executing action X, then its intentions,
desire, abilities, rights, obligations, and/or
knowledge (its glue) to participate as role B
in action Y are changed.
We shall use various ways to diagram these
notions (a discussion of diagramming techniques
can be found in Andersen, 2004b). Figure 3 shows
an ordinary door opener from a shopping mall. The
dependency between actions is shown by arrows

annotated by the glue dimension affected. The

participants involved are shown in boldface.
The activity contains five actions that are connected as follows: when a person walks toward
the door in order to get in, the door opener is
required to open; when it does, it becomes able
to function as a passage for persons passing the
doorway but also for heat disappearing. When
it closes, it loses these abilities. The goal of the
activity is letting people in through the door while
preventing heat from disappearing. Since the two
goals are contradictory, we must dissolve the
contradiction by unfolding it in time. Therefore,
when the door opens, it becomes obligated to
close after, say, 30 seconds.
In the game-specifications, we use a simplified
version of the diagram in Figure 3. The simplified
diagram is on a high level of abstraction, displaying only the essence of the game.
The abstraction is made in the following
way:
12. The participants, roles, and glue changes
responsible for the dependency between the
actions are stripped away.

Figure 3. The activity of door-opening focusing on the door participant. Obl = obligation, abil = ability.
+ means increases, - means decreases
Participant: Door

Prevent:

Achieve:

Heat evaporates
through door

Persons pass
through door

[+abil ]

Persons walks
towards door

[-abil]

[+abil]

[+obl
+ abil]

[+obl ]

Door opens

[-abil ]

Door closes after

30 seconds

[+abil]
[-abil]

560

[-abil]

The Semiotics of Smart Appliances and Pervasive Computing

13. Abilities and rights are merged under the

heading possibility and represented by a
single-headed arrow
14. Desire and obligation are merged under
the heading necessity and represented by a
double headed arrow
15. Inability or prohibitions are merged as impossibility and represented by
The behavior of the door opener now looks as
shown in Figure 4.
The reason for these formal abridgements is
that diagram design is a tradeoff between precision and overview. The reader should be able to
grasp immediately the possible paths of the story
by reading the diagram. However, there is still an
underlying detailed semantics in which participants, roles, glue, and glue changes are specified.
They are just not shown in the diagram.
The previous notations highlight the interdependences between actions but do not provide
a clear picture of the relation between activities
and space. Such relations are important when we
design context-sensitive, pervasive technology,
and therefore, we introduce a maplike notation
in the next section. It associates activities to a

physical space, if the space plays the role of location in the activity.

Habitats
In pervasive computing, the physical space and
its relations to the informational space becomes
important. Therefore, we need a diagram that
highlights the spatial properties of activities and
networks, in opposition to the previous section
in which we highlighted the relations between
actions.
Figure 5 shows a diagram that codes the spatial
participants graphically and tones down relations
between actions. We have selected the spatial
participants (i.e., spaces that participate in the
Location role of the activities, spatial participants
filling the location roles are called habitats. We
have shown two habitats: the shopping mall and the
entrance. To each habitat is associated the actions
that can be performed there: selling and paying
in the shopping mall, and walking through doors
at the entrance. In addition, we have represented
the signs involved in terms of (a) the area from
where they can be accessed (the access area) and
(b) the area containing the object denoted (the
reference area). Both areas are shown by means

Figure 4. Modal diagram over a door opener

Persons walks
towards door

Prevent:

Achieve:

Heat evaporates
through door

Persons pass
through door

Door opens

Door closes after

30 seconds

561

The Semiotics of Smart Appliances and Pervasive Computing

of dashed polygons, and an arrow points from

access to reference area. Thus, the door opener
has access to the areas inside and outside the
door. The arrows are decorated by the main signal
path. Sensors often transmit and receive signals
in two directions. For example, radar sends out
radio waves and records their echo. The outgoing
waves are only a means for receiving the ingoing
echo. Therefore, the main signal path in radar is
from reference to access area, which is also true
in the door-opener case.
We distinguish between relative and absolute
references. The difference is that if the access
area of relative references moves in time or
space, the reference area moves, too. This is not
the case with absolute references. The reference
of the door opener is relative, since if we move
the door opener, it will refer to the new environment in which it is placed. Transponders, RFIDtags, and radars create relative references. The
distinction also shows up in language, where the
so-called deictic words (here, now, me, you) have
references relative to the speaker. Compare the
two sentences: The Second World War ended in
1945 and The Second World War ends now. The
former does not change meaning if it is uttered
in year 2000, whereas the latter becomes false if
uttered in that year.

To each habitat is associated half-baked actions

that can be performed therethe affordances
(Gibson, 1986) of the habitat. In the following,
# means that the role is instantiated, whereas a
type indication such as pedestrian means that the
role can be filled by a specific type of participant.
Pedestrian walks through #door thus means that
the Location role is instantiated by the particular
door and fixed, while the Agent role can be filled
by any pedestrian. Fully instantiated actions are
created by unifying the action possibilities of the
participants with the affordances of the habitat.
For example, #smith can walk through doors +
pedestrians can walk through #door unifies to
#smith can walk through #door, because #smith
is a pedestrian and #door is a door. If #smith was
not a pedestrian but a car, or if #door was not a
door but a wall, unification would be impossible,
and no action could be instantiated. There are
many ways to implement instantiations. In (10),
the Activities section, we suggested a particular
way of doing it: shared location gives the participants the ability to instantiate roles in the actions,
while desire gives a participant the incentive to
instantiate the agent role.
This way of combining abilities from agent
and environment is useful in agent design, as
argued by Cabri, Ferrari, and Zambonelli (2004),

Figure 5. Shopping mall and entrance. #door means an instance of the class of
shopping mall
Roles: customer, clerk
Clerk sells commodity to customer
Customer pays price to clerk

562

Entrance
Roles: #door, pedestrian
#door opens
#door closes
Pedestrian walks through #door

The Semiotics of Smart Appliances and Pervasive Computing

who also uses a role-based architecture. We shall

term it contextual execution. The term means
that an execution of an action consists of pieces,
some of which originate from the participant,
while others come from the habitat. This enables
participants to act flexibly in different contexts
without having to remember the exact details of
each particular context.
The notion of half-baked actions (Andersen,
2004c) reflects the idea that we normally never
start from scratch when we act. We know that
specific participants normally are useful for
the purpose we want to accomplish, and we try
them out first. To take the door opener example,
if we want to enter a room, we do not start with
the isolated verb enter and spend time reflecting which participants it should have. Our point
of departure is enter through #door, where the
direction case is instantiated by a specific door
we can see. Only if the door is locked may we
begin considering the windows and the chimney.
Another argument is that isolated verbs do not allow predictions: drinking milk, beer, and cyanide
has quite different consequences.
In this method, all actions are bound to a certain
physical habitat, and, indeed, many actions work
in this way. We do need to be inside the airport to
embark on an airplane. But what about the planning of the travel? Do we have to be in Portugal
in order to plan the travel from the airport to the
hotel in Lisbon? No, of course not. Some activities,
such as planning, should not be and are not bound
to specific locations. Instead, we imagine how
things are in Lisbon and simulate the remaining
part of the travel in our minds (or on a computer,
or let the travel agency take care of the problem).
This means that we need the concept of fictive
habitats: habitats we imagine and in which we
can test solutions without being there in reality.
Fiction in general enables us to enjoy living in
fictive habitats for a delimited period of time, and
therefore, we shall need the concept in Designing
Pervasive Games and The Bogeyman, where
we address the question of designing pervasive

games for children (on the logic of fictive worlds

(see Ryan, 1991). Pervasive games thrive on the
interplay between actual and fictive habitats, as
we shall see presently.

Play and Games

The preceding conceptual framework has mostly
been used for analytical purposes or for instrumental design topics like brain surgery; evacuation
procedures on ships; and pervasive applications in
shopping malls, airports, and hospitals (Andersen,
2005; Andersen & Nowack, 2002, 2004; Brynskov
& Andersen, 2004; Haase, Musaeus, & Boisen,
2004; Kristensen, 2002, 2003; May & Kristensen,
2003; May, Kristensen, & Nowack, 2001).
Designing technology for playful activities
is very different from designing tools for instrumental activities. It is not necessarily the most
efficient solution that is the best solution. The
user should have funand fun is hard to predict
and design. Thus, playful activities often are
not goal-directed, at least in the sense that the
goal may be hard to define or quantify, and the
activity may be enjoyable for its own sake (see
Csikszentmihalyis, 1990, concept of flow).
In the rest of the chapter, we shall test our
framework in a design context. In particular, we
shall explore whether it yields creative ideas for
designing pervasive technology for childrens
play. The project called Nomadic Play in Mixed
Environments is a part of the Center for Interactive Spaces at the University of Aarhus.
The notion of nomadic play is different from
mobile play. Mobile indicates independence of
physical location (i.e., activities are enabled, and
resources are available regardless of the users
location). For example, a mobile phone may allow
a child riding a bus to play a game, check e-mail,
or use instant messaging. Nomadic, on the other
hand, indicates dependence of physical location.
The analogy is nomads traveling through the
desert, being dependent upon oases and family

563

The Semiotics of Smart Appliances and Pervasive Computing

networks. In a similar fashion, nomadic users

are dependent upon, or take advantage of, resources and services that are made available by
the environment, including other peoples devices
(peer-to-peer). Thus, nomadic play indicates a
playful use of smart appliances and pervasive
computing systems that could not occur if the
child was not at a certain location (e.g., at a club
or at home) that offers certain services (e.g., a
home game server or media center) to a certain
group of people (e.g., friends of the house). One
could argue that certain features of mobile systems
are also nomadic in nature (e.g., the coverage of
mobile phone networks, but we restrict the term
nomadic to systems that are more local in the
physical sense and, thus, are designed to support
a more focused set of human activities.

Theories of Play and Gaming

Childrens play has mostly been treated in developmental psychology (Piaget, 1962; Sutton-Smith,
1979; Vygotsky, 1976) and in the context of game
design (Salen & Zimmerman, 2004). Here, we
adopt the view that play is the search for fun. To be
a bit more specific, play is a focused but internally
motivated activity without serious consequences.
It may not have a goal outside itself other than
it keeps a child occupied in a way that allows it
to gain experience in the world, both physically
and socially. From this perspective, play could be
seen (by adults) as training and preparation for
adulthood, whether structured or not. The search
for fun is the fuel that drives the activity. Thus,
learning may be considered play, as long as the
internal motivation is high, because it is fun.
Playing a game can be seen as a special case of
ludic activities, which, in turn, are a part of being
playful (Salen & Zimmerman, 2004). A game can
be defined in many ways, and there are plenty of
suggestions from which to choose. Koster (2005)
lists the following from academia: Roger Callois
an activity which is voluntary ... uncertain,
unproductive, governed by rules, make-believe;

564

Johan Huizingas free activity ... outside ordinary life ; and Jesper Juuls (2005):
A game is (1) a rule-based formal system with (2)
a variable and quantifiable outcome, where (3) the
different outcomes are assigned different values,
(4) the player exerts effort in order to influence
the outcome, (5) the player feels attached to the
outcome, and 6) the consequences of the activity
are optional and negotiable. (Juul, 2005, p. 3)
We distinguish between two types of playful
activities: playing and gaming. A game has fixed
rules and a quantifiable outcome (a clear goal),
whereas play has ad hoc negotiable rules and a
fuzzy outcome. A game can be embedded in play
(see Juuls point about the optional and negotiable
consequences of a game). Both types of activities
should be fun.
An important distinction is that between
emergent games and games of progression (Juul,
2005). In the former, the game is defined by a
small number of simple rules governing a possibly
large number of human or nonhuman participants.
An emergent game can be played many times,
since each round is different. Emergent games
invite planning and strategic thinking. Games of
progression are defined by a more or less fixed
sequence of events and often tell an elaborate
story. Like novels and films, they seldom can be
played more than once.
Our two games are games of emergence, but
in Bogeyman, we have tried to add narrative
traits.

Applying the Theory to Games

This section adapts the concepts presented previously to the special case of games. In this section,
we consider games in general, and in the next
section, we focus on pervasive games.
We first show that the glue concept (that binds
participants to activities) is useful to characterize games.

The Semiotics of Smart Appliances and Pervasive Computing

The childrens desire motivates them to enter

activities (come and join us; this is fun) and a
lack of desire to leave them (I am going home
because this is boring). Playthings must create
desires in children to assume the Agent role.
Ability, rights, knowledge and obligations
may regulate participation for human as well
as nonhuman participants. An able goal keeper
is sure to be enrolled in a soccer game. In the
Danish game called The Earth Is Poisonous, it is
allowed to tread on furniture but forbidden to step
on the floor, because it is poisonous. These tokens
of glue may be earned from previous activities,
and if they are given a physical shape, they can be
exchanged by the children. For example, a child
only can participate in secret club activities if he
or she has been allowed access to the club and
can show his or her membership card.
A child may only get entrance to the club den
if he or she knows the secret password. Or abilities
may be missing: You are not allowed to take part
because you are stupid. (One might ask, Who is
lacking abilities here?)
Glue like rights and obligations can be distributed: Im the president and you must do what I
tell you. No, that is silly. Well, then you can be
the vice-president and tell Billy what to do.
Actions can be associated to spaces via the
habitat concept. This is a pirate ship, and the
grass is water. You cannot walk on the water. If
you do, you drown. This chair is a car that can
go 200 miles per hour. The flagstone is a hole
in the Earth, and if you tread on it, you will fall
down.
Games can have different purposes:

Creation: The objective of soccer is to

accumulate goals. The goal of exchanging
Pokmon cards is to increase ones collection.
Destroying: The objective of a Star Wars
game is to destroy the Death Star. The objective of teasing and mobbing is to destroy
the self-confidence of the victim.

Preventing: The objective of a cowboys and

Indians play is to prevent the Indians from
conquering the fort.
Maintaining: The goal of skipping is to
keep the skipping rope moving.

If we focus on pervasive games, the following

possibilities present themselves. According to
(10), activities can be executed if its role fillers
are qualified for the roles and if the Agent filler
desires or intends to fill the Agent role. The roles,
therefore, are filled by entities that meet the requirements (glue). In the nomadic play scenario,
roles are filled by children, artefacts, hardware,
software, and locations. Joining and leaving the
activity is done on an ad hoc basis, like when
children play. At the conceptual level, we do
not discriminate between children and software
agents. It is the definition of the role at hand that
matters. As a consequence, if a software agent
meets the requirements to fill a role, it can play
it. This also means that a child could fill a role
that usually might be played by an artefact or
piece of software.
Children and software may compete for filling
a role: I do it vs. My mobile phone does it.
More interesting, however, is in-between situations in which child and software together fill
a role. Part of the role is automated; part of it is
manual. Sometimes a child may be represented
by a less capable but sufficient software agent,
and agency could be transferred back and forth.
Such shifts between manual and automatic control
are very common in process control; sometimes
the officer of the watch maintains the course, and
sometimes the autopilot does it.
A collection of roles that typically go together
can be collected in macro roles. These macro
roles either could be statically defined, like in
the rules of a game, or they could be assigned
dynamically, based on the actual history of
actions taken by or experienced by the entity
(bottom-up role filling). This goes for human as
well as nonhuman participants. The latter gives

565

The Semiotics of Smart Appliances and Pervasive Computing

us interesting opportunities, since the playthings

can have a memory that remembers the glue they
have earned in previous plays. This is well-known
in role-playing games (RPGs). You must earn
experience points to become a level-32 wizard.
In the extended case, the character can develop
behavioral traits and reputation in an emergent
fashion, based on previous actions (e.g., by choosing to practice certain skills).
Sharing things is a central part of many activities. We want to be able to share passive objects
as well as active or intelligent ones. This also
means that pieces of runnable code should be able
to migrate live on a variety of different hardware
(contextual execution).
The effect is that we get a network, as described
previously in actor networks, in which the possibilities are not tied to the entities themselves
but rather to the possible roles they can fill in a
given set of activities.

Designing Pervasive Games

In this and the next section, we use the contextualized concepts previously defined for designing
two pervasive games.

The Technology
In the Nomadic Play project, we try to invent new
ways of having fun or being cool using digital
media, specifically pervasive and ubiquitous computing devices. Those devices are characterized by
being small enough to be carried around as other
small objects that surround children (e.g., toys,
phones, clothes, pencils, books, stones), or they
may be embedded in the environment like other
resources we knowin playgrounds, computers,
blackboards, and furniture. In essence, these new
digital artefacts and environments should be
designed in a way that allows them to become a
natural part of playful activities.

566

Compared to traditional artefacts, these devices present new opportunities. They can process
information, sense the environment via sensors,
interact with it physically through actuators, and
communicate with other devices through wireless networks (for early thoughts on ubiquitous
gaming in the context of construction kits, see
Sargent, Resnick, Martin, and Silvermans, 1996,
list of Twenty Things to Do With a Programmable Brick). Apart from actually making the
technology work, one of the biggest challenges
is to find out how activities involving children,
artefacts, hardware, and software can be brought
to play in a way that feels natural and fun. The
digital media should not be a separate distraction
from the play, not in the sense that it should be
invisible, but rather that it should be an integrated
part of the activities. In order to test our ideas,
we have begun designing playful activities for
children involving pervasive computing.

StarCatcher: A Simple Example

In this section, we describe an existing pervasive
game, StarCatcher (Brynskov et al., 2005). In the
next section, we use our framework to elaborate
on the game by introducing features that are more
complex and, hopefully, more entertaining.
StarCatcher is a simple version of Capture
the Flag (Figure 6). It can be played by two or
three teams in any urban area of approximately
500x500 meters, and the objective is simply to
catch a virtual star (Figure 6b). The first team to
reach its location and run to its home base wins
and become Star Winners, and the game is over.
The losers get the Loser Star (Figure 6c). There
is also a slightly more complex version: If the
team with the star is intercepted by the other
team before it reaches the base, it drops the star,
and both teams must visit the home base before
it can be picked up again.
Technically, each team is equipped with a mobile phone and a GPS unit (Figure 6a). The game
software runs as a client on each phone and on a

The Semiotics of Smart Appliances and Pervasive Computing

Figure 6. (a) GPS unit and mobile phone; (b) phone interface with team (red dot) and star; (c) Loser
Star screen presented to losing team

(a)

(b)

(c)

Figure 7. Diagram of the activity Playing StarCatcher (simple version) with each of the actions and
their relations seen from both teams
B is
home

A is
home

B walks

A walks

B
catches
S

A
catches
S

central server. The clients send their positions to

the server, and the server returns the game state.
The phone interface is a map of the close proximity of the team (50-m radius) in which each team
is represented by a pulsating dot.
The framework described in section Activities offers two ways to describe the game: either
focusing on (a) relationships between actions or (b)
on the habitats. Figures 7 and 8 use the formalism
introduced in Activities to diagram the activity
Playing StarCatcher. Figure 7 shows the simple
version, Figure 8 the elaborated one. The + and
- marks represent the desires of the participants:
both team A and B strongly desire to win.
As appears from the diagrams, our game
specification uses half-baked actions, as argued
in The Habitat Concept. For example, in Figure
8, there are three actions with the same verb: B
walks, B walks home, and B walks with S.
In the game, these variants have quite different

B:
B walks
with S

A walks
with S

B is home
with S

B wins

A is home
with S

A wins
A:

consequences and, therefore, are treated as three

different actions. If we magnify these actions, they
will look as shown in Table 1, but a description
at this level of detail will clutter the diagram and
make it useless.
Figure 9 shows StarCatcher described as a
habitat. The dashed fat rectangle represents a fictive habitat (see section Habitats and Andersen,
2004a). The child C can see its surroundings in
the actual (the dashed circle) as well as the fictive
world (the dashed rectangle). The reference from
the actual to the fictive world is relative (section
Habitats), since the reference area moves as the
access area (the child and the mobile phone) moves.
Figure 9 is rather abstract, but in the implementation phase, it will be replaced by a map that shows
the exact shape of the actual and fictive habitat
and the exact shape of the access and reference
area. In the figure, we have indicated that the
reference area is different in the actual and the

567

The Semiotics of Smart Appliances and Pervasive Computing

Figure 8. Diagram of the activity Playing StarCatcher (extended version)

B is
home

B
catches
S

B walks

B:
B walks
with S

B is home
with S

B wins

B meets
A
B walks
home

B drops
S
A drops
S

A walks
home

A meets
B
A is
home

A
catches
S

A walks

A walks
with S

A is home
with S

A wins
A:

Table 1. The action walks with necessary (agent and action) and possible roles filled (goal and sociative); the action has no glue, since it is not a role filled by a participant
Role

Agent

Action

Goal

Sociative

Glue

Desire (to win)

Ability (to be reached)

Ability (to be picked up/carried)

Filler

walks

[home]

[with star]

Figure 9. The habitats of StarCatcher; C = child. A = Agent role, O = Object

The actual world

The fictive world

A

O
Catch star
A = the child
O = the star

Walk home
Walk with star
Drop star

home
Win
Meet the other team
Walk

568

The Semiotics of Smart Appliances and Pervasive Computing

fictive world. In the actual world, the child can

se things within a circle centered in the child; in
the fictive world, the access area is rectangular
due to the shape of the display.
The fictive habitat contains the avatars of the
children and the star. The children, therefore, lead
a double existence in the real and the fictive world,
while the star and the winning position only exist
in the fictive world. They are not marked in the
actual world. The same difference is found in the
actions: catching the star, walking with the star,
and winning are fictive, while meeting the other
team and walking around in the city are done in
both worlds.
The fictive habitat contains two subhabitats:
the area around the star where catching is possible
and the home base where winning is possible. The
action of catching requires two roles: the child
(Agent) and the star (Object). In Figure 9, the agent
role and catching the star is empty (shown by the
white color), so the action cannot be executed.
Only when the child moves (the fat arrow) into
the habitat will it be able to participate. This is a
graphical way of representing the process of unification described in section Habitats. Habitats,
thus, are loaded with activities that are triggered
when qualified participants move into them. This,
in fact, is one of the reasons for mobility work
(Bardram & Bossen, 2005; Brynskov, 2004); in
hospitals, for example, many activities are bound
to specific locations, and nurses, doctors, and
patients spend much time just moving physically
from one habitat to another in order to fill empty
roles in urgent activities that are executed as soon
as all necessary participants are assembled.
The example shows that our framework is
indeed able to conceptualize essential aspects of
the game; the dependencies between the actions
and, in particular, the relation between actions and
physical environment and the interplay between
the actual habitat and the fictive one.
In the following, we will describe a new game,
Bogeyman, which is an extension of StarCatcher.

The purpose is to find out whether the framework

is a good support for creative design and whether
it can provide new and entertaining ideas while
still modeling relevant elements of complex pervasive gaming.

The Bogeyman
If all we wanted to do was to model locative
games as simple as StarCatcher, there would not
be much need for a framework. But with the expanding opportunities of pervasive gaming (due
to technological evolution and commoditization
of pervasive computing systems). a whole range
of complex issues arise that make formalized support for development of such systems more and
more difficult. We can design more sophisticated
games using the dynamic interplay of aspects at
different levels (e.g., physical, technological and
social) but at the cost of complexity. Addressing
this complexity at appropriate levels is an important motivation for developing our framework.
StarCatcher actually was implemented and
tested; Bogeyman is an extension of StarCatcher
and has not been implemented yet, although we
do present a core algorithm of the play. The purpose is to test our framework as a design support
tool; does it help produce interesting ideas for an
extended game?

Activities and Habitats

Let us start with the first two desiderata: we want
the following:
16. An emergent game in which simple interactions between more participants can create
surprising and entertaining situations (section Play and Games).
17. A better exploitation of the interplay between the real and fictive habitats (section
Starcatcher).

569

The Semiotics of Smart Appliances and Pervasive Computing

Desiderata (16) concerns the activities of the

game, whereas (17) relates to its real and fictive
habitats.
In StarCatcher, there was only one thing to
collect: the star. In Bogeyman, the purpose is for
the children to collect as much candy as possible
and avoid being eaten by the bogeyman. While
the star was purely fictive, candy may be fictive
or real or both. The Bogeyman game is also set
in the city (i.e., an area not especially built for the
game). However, to a certain extent it does incorporate physical aspects of the surroundings (e.g.,
the real sewers of the city, which the fictive mice
can enter and leave). When the mice are in the
sewers, they are invisible in the fictive world.
There can be from two to 10 kids (single players or teams). The technical setup is basically as
in the simple StarCatcher game (mobile phone +
GPS), with the addition of Bluetooth interaction
with places and artefacts.
The objective of the game is to collect as
much candy as possible. The purpose, however, is
broader, defined as (a) having fun together and (b)
developing relationships between the players.
We want some kind of recognizable narrative
macro structure (e.g., built upon Greimas schema)
(Greimas, 1966). A Subject desires an Object, is
helped by Helpers, and opposed by Antagonists.
In the game, the kids (Subjects) run around and
try to catch candy (Object of desire) without being
caught by the bogeyman (Antagonist). Whenever
they see a piece of candy, virtual or not, they approach it cautiously. If the bogeyman is hiding
close by, he will catch them, strip them of all their
candy, and eat them. When this happens, they
must return to the home base and be brought back
to life again (spawned) before they can enter
the game again. The kids cannot see whether the
bogeyman is near. So, in order to avoid him, the
kids can get help from the dogs (Helpers). We
want these macro roles to emerge from the action
dependencies constituting the game.
The dogs, however, are not necessarily interested in helping the kids. But if they receive

570

a sausage, they acquire a positive attitude to the

childrens goals and thus desire to act in a way
that supports the children.
The execution of the rules controlling the fictive
participants conforms to (10) above: the Agent of
the action must desire or be obligated to act, and
all participants must be able to do so. This means
that (a) the software must create desires in the
software Agents that are understandable to the
children, (b) that actions are only enabled when
all participants are in the same location. (b) will
involve a lot of physical and virtual movements,
which is desirable for a pervasive game.
For example, if the dog is given a sausage, it
desires to chase the bogeyman away. But if there
is no bogeyman present, the role of the chasee is
empty, and the chase-action cannot be executed.
Only when the bogeyman appears and fills the
chasee role can the dogs execute the action they
desire. Processes involving obligations and desires
are clearly fictive, whereas processes involving
ability is both real and fictive, since the fictive
locations of the software agents must coincide
with the real locations of the children.
In the section Activities, (6)-(9), we listed
four requirements to cooperative software agents:
comprehensibility, accessibility, joint attention,
and intermittent participation. Accessibility
means that the action potentials and desires of one
participant should be accessible to the others. We
can do this by letting the interface show which actions are possible in the current habitat (as already
illustrated in Figure 5 and Figure 9) and what the
fictive effect of the action is. For example, when a
child is near a fictive or real sausage, the display
shows that the action of taking the sausage is possible and that it may lead to enrolling dogs in the
childs team. Similarly, when a child approaches
a fictive dog, the display suggests that it is now
possible to give the sausage to the dog.
If a player wants to know more about a nonplayer characters (NPCs) attitude before interacting with it, he or she may take a closer look first
(by using the scan mode described next).

The Semiotics of Smart Appliances and Pervasive Computing

The game is designed in levels of increasing

complexity. In the beginning, you only have to
find candy. No bogeyman. Then come the mice
(an Antagonist) and begin eating your candy.
Then cats, which are useful because they eat mice.
Then, at last, the dogs and the bogeyman. Dogs
protect you from him, but if there is a cat nearby,
they may run after the cat instead. If you forget
to feed them, they may get angry and bite you.
The mice will come out of the sewers, which are
connected underground.
When players meet, they can fight each other,
but they dont have to do it. Strength is measured
by the amount of candy they have. They can freely

negotiate offline, if they prefer (e.g., that the weaker

player gives some candy to the stronger team)
(security payment). They also can have a virtual
fight throwing different objects and power-ups
at each other. If they fight and one of them loses,
the winner takes all the losers belongings. If
you find other players home bases, you can strip
them of all their belongings. A base is defined by
a Bluetooth tag. You probably have to look around
a bit in order to see it. When you reveal their base,
their icon explodes, and they must come to you
and collect it to set up a new base. They do this
by placing their tag somewhere else.
According to (11) in section Activities,
execution of actions changes the glue binding

Figure 10. Dependencies between actions in Bogeyman

+
-

Cats:
Mice:
Cats eat
mice

Mice enter
sewer

Mice:

Dogs:

Dogs catch
cats

Cats catch
mice

Mice eat
candy

Cats:
Children:

Cats chase
mice

Children
eat candy

Cats eats
fish

Bogeyman:

Dogs catch
bogeyman

Mice:
Mice
multiply

Mice leave
sewer
Bogeyman
runs away

Bogeyman
catches
children
Bogeyman:
Children:

Dogs chase
bogeyman

Dogs:

Children
give fish to
cats

Children
take candy
Children
run away

Cats run
away

Children
take fish
Dogs chase
cats

Dogs eat
sausage

Children give
sausage to
dogs

Children
take
sausage

571

The Semiotics of Smart Appliances and Pervasive Computing

participants to other actions, and so changes of

these dependencies must be an integrated and
visible part of the game. Figure 10 shows the
dependencies between the actions. As in section
Activities, we abstract the participant-glue-role
relations into three main relations: possibility,
necessity, and impossibility in order to maintain
overview. To give an example, the capture of the
children by the bogeyman is prevented by him
running away, which is triggered by dogs chasing
him or by the children running away. The capture also can be prevented by the dogs catching
the bogeyman. A capture will prevent children
from taking more candy; but, and this is the snag,
collecting candy is the one action that makes it
possible for the bogeyman to catch the children.
Collecting candy is risky.
As appears from the +/- marks, dogs desire to
catch cats and eat sausage, cats to eat mice and
fish, children to eat candy, mice to eat candy and
to multiply, and bogeymen to catch children. On
the other hand, mice hate being eaten by cats,
bogeymen to be caught by dogs, and children to be
caught by bogeymen. Thus, the diagram contains
conflicts that can be used in a narrative: mice oppose being eaten but cats think it is a good idea;
bogeymen like catching children, but children are
dead against it. There are also implicit conflicts:
mice eating candy prevents children from doing
it, and vice versa.
A cursory inspection of the diagram shows
that it also contains explicit and implicit contradictions: the childrens taking candy enables the
bogeyman to catch them, which again prevents
their taking it; the childrens giving sausage to
dogs prevents dogs from chasing cats (which they

desire) but enables them to eat sausage (which they

also desire). Mice entering the sewers prevents
them from eating candy (which they love to do)
but saves them from being eaten by cats (which
they dont like). The contradictions and conflicts
are built into the rules in order for them to generate an interesting narrative, but a consequence is
that we cannot use standard logic to implement
the rules.
Two rules used by the system are not shown in
the diagram, because they would clutter it completely: cats eat mice obstructs all actions in which
mice participate: mice multiply, mice enter sewer,
mice leave sewer, and mice eat candy. Similarly,
Bogeyman catches children obstructs all actions
involving children: children give sausage to dogs,
children give fish to cats, children take candy,
children take sausages, children take fish, children
run away, and children eat candy.
Note that we have only distributed desires to
a few actions: as it stands, dogs only know that
they like to catch cats and eat sausage; they have
no particular desire to chase and catch bogeymen
or to chase cats. In addition, they have no attitude
toward bogeymen catching or not catching children. The idea is that such derivative attitudes
should be generated as an emergent phenomenon
as the game progresses. Enrolling virtual agents
as helpers and avoiding defections to the enemy
should be part of the gameplay. Thus, Figure 10
only works as a seed for the game.
The second desideratum was a better interplay
between the real and fictive habitats. We shall
deal with that now.
The phone is the primary artefact in the informational habitat. It has three modes that are

Table 2. Modes of the client running on mobile phone

572

Mode

Scale

Use

Interaction

Technology

Proximity

1:1

Navigate

Walking

GPS

Global

1:10

Get overview of entire arena

Cursor

GPS

Scan

10:1

Inspect place, thing, or person

Menu

Bluetooth

The Semiotics of Smart Appliances and Pervasive Computing

Figure 11. Informational habitats

Global

NPc

Proximity

related to the physical habitat: proximity, global,

and scan (Table 2).
In the proximity mode, the player can see all
objects of the real and fictive habitat within the
field of vision. In the global view, the player can
only see other players and the map of the arena,
including the players own home base. The scan
view offers a closer look at something (e.g., the
attitude of an animal or the presence of a fictive
sausage), but it also obstructs the other two views.
Whenever a player is in the scan view, it is reflected
by the players representation in the global and
proximity view. This is useful information to other
players because they now know that there may be
something interesting to check out at that location.
Players can also allow each other actually to share
their proximity or scan view, thus enhancing each
others access to information.
Figure 11 shows the three informational modes
of a habitat. The scan and proximity modes are
relative references that change as the player
moves. The global mode is an absolute reference
(dashed arrows), since the whole playing area is
accessible all the time. The scan mode differs
from the two other in that the access area is defined by the location of the reference and not by
the receiver. The reason is that the referent (the
NPC in Figure 11) carries a Bluetooth tag that
creates an information cloud (access area) around

Scan

it with information about it. If the mobile phone

in Figure 11 is Bluetooth-enabled, it can receive
this information.
The mobile phone in Figure 11 always can
see the global layout of the game area; it can see
more details in the local area around it; and it can
scan the NPC since the phone is inside the access
area of the NPC.
Figure 11, however, does not capture the interplay between the actual and the fictive in the
game. In order to do that, we need the concept of
fictive habitats introduced in section Habitats.
Parts of the game exist in the real physical habitat;
this is true of the sewers and the children and
possibly of the candy and sausages, depending on
how you set up the gameand possibly a child
may wish to play the bogeyman. Some of the actions also exist in the real habitat; for example,
the action of walking and running. However, the
game rules only exist in the fictive habitat, and all
the participants in the game must have a fictive
representation as well. Otherwise their actions
could not influence the game state.
Of course, we can depict this dichotomy as
in Figure 9 by drawing the actual and the fictive
habitat side by side. But here, the correspondence
between the fictive and the actual positions of the
participants does not stand out. In the following,
we collapse the two diagrams into one. The fic-

573

The Semiotics of Smart Appliances and Pervasive Computing

Figure 12. Basic setup with two players, the bogeyman, some candy, and the home bases

Figure 13. Space-time plot of discrepancies between

the physical and fictive habitat
Space

Real motion
of player

s
s
b

M
M
s

Delayed avatar
Avatar on
autopilot

Real motion
of player
Time

tive diagram is the main one, since it defines the

game state. If a fictive participant has a real-world
counterpart, we show this by adding a shadow to
the fictive entity. Thus, the real world is shown
as a platonic projection of the fictive world. We
distinguish between fictive (only in the fictive
habitat), real (only in the real habitat), and mixed
entities (both in the fictive and real habitat).
Figure 12 shows the basic setup with two mixed
players (C), a mixed bogeyman (B), three pieces
on fictive candy (S), one piece of mixed candy (S),
two fictive mice (M), and one fictive dog (D).
In addition to candy and sausages, there are
a number of special items, power-ups, the players can encounter: a dog shit, a shadow, and a
torch.
The shit is left by dogs, and if you run over it,
you slip, and your avatar is slowed down. You have
to slow down yourself, because generally, if you
are not close to your avatar, it shifts to autopilot
and begins moving on its own. Holiday!
If you walk carefully past the shit and do not
tread in it, you can pick it up and drop it later at
your convenience (e.g., in front of another player).
The stickiness is, of course, entirely virtual and
only causes the virtual representation of the player
to move slowly.
The shadow, inspired by H. C. Andersens
fairy tale, The Shadow, is a way to disguise the
574

actual location of a player. When it is activated,

the players avatar will use dead reckoning to
move in the same direction, regardless of the
actual movement of the player. In this way, a
player can sneak up on another without being
noticed. The problem is that if your avatar meets
any other avatar, it gets scared because its master
is not present, it flees in an erratic manner, and
the child must catch it.
The space-time plot in Figure 13 illustrates the
effect of delaying dog shit (topmost) and the effect
of a shadow on autopilot (bottommost).
The torch (Figure 14) is a way of showing a
virtual or mixed entity A to another virtual or
mixed entity B. The torch can be implemented
as a highlight or a cursor. In Figure 14, an arrow
is used. The arrow belongs to the fictive world
and, therefore, exists in the fictive reference area
on a level with fictive mice and dogs. Each user
can manipulate his or her torch. Since the torch
is a pointing device, it should only be visible in
the proximity modeyou only point out things
to people close to you, and physical proximity
serves to delimit the number of addressees.
The torch is thus a way of establishing the
shared focus of attention between participants,
as required in section Activities. From the perspective of semantic roles, the torch is a way for

The Semiotics of Smart Appliances and Pervasive Computing

Figure 14. Using the torch

b
A
c

an agent (C) to increase the glue between another

agent (B) and a role in an activity that involves
the entity in focus (A). If a child points to a sausage in the presence of a dog, it strengthens the
desire of the dog to play the role of Agent in the
affordance of the sausage (namely, that of being
eaten by a dog).
The torch can also be used to move the focus
of the dog away from the affordance of cats
(being chased) to the affordances of sausages
(being eaten) or bogeymen (being chased away).
A mean way of using this technique is to reveal
the other players to the bogeyman and make him
hunt them.
The game ends after a fixed time or whenever
it is suspended by the players. Players can enter
and leave intermittently.
In section Starcatcher, we decided that the
ability to participate in activities is created by
spatial movement: if the participants are assembled
in the same place, they are able to participate in
the activity. But, as mentioned in Activities,
the activity will not execute unless the Agent
participant desires or is obligated to do it. How
do we provide these desires?

Designing Intelligent Agents

In this section, we will discuss how to implement the distribution of desires mentioned in the
previous section. Mice and children are clearly
competitors: how can we make the mice aware
of this and possibly begin oppose the children
(e.g., by supporting the bogeyman in catching
them)? How should the cats react toward chasing

mice: on the one hand, chasing them leads to the

desired action of eating them, but it also enables
the mice to flee to the sewers, which disables the
eating. We would also like to generate attitudes
between participants: who sides with whom? Will
the mice oppose the cats and side with the children
(although they compete for the candy)? Will the
dogs assume the role of helpers to the children,
even if helping out prevents them from chasing
cats? In short, will global narrative roles emerge
from the interplay between the actions?
Which methods should we use to make this
come about? We realize already that since the story
contains contradictions, we cannot use standard
logic for these inferences. Classical AI is thus
not an option. In order to choose a solution, let
us first look at our requirement lists, repeated as
follows for convenience:
1.

Intelligent technology is a kind of technology that is able to contribute positively to

activities whose manner of performance we
will intuitively call intelligent.
6. Agents should have intentions to act that are
understandable to the others (comprehensibility).
9. Agents should be able to enter and leave the
activity (intermittent participation).
10. An activity is executed if all participants are
able to fulfill their roles to some degree and
the Agent filler is strongly obligated and/or
strongly desires to fulfill his or her role.
(1) says that the algorithm must refer to networks of actors, not to the individual actor; (6)
says that the rules it uses must be known to the
human users.
One interesting possibility is the theory of
cognitive dissonance (Eskola, 1973). The theory
allow us to calculate the relation z between persons
A and C, if we know the relation x between A and
B and the relation y between B and C. It (regrettably) works well for international politics, so we
take the fight against terrorism as an example: if

575

The Semiotics of Smart Appliances and Pervasive Computing

Figure 15. Removing dissonance using y

Figure 16. Removing dissonance using y

C: Taleban
z

A: George

B: Al Qaeda

George W. Bush (A) opposes (x) Al-Qaeda (B)

and Al-Qaeda supports (y) the Taleban (C), then
George (A) opposes (z) the Taleban (C) (Figure 15).
In some cases, it is also possible to calculate z if we
know x and the inverse of y, y. If George opposes
Al-Qaeda and Taleban promotes Al-Qaeda, then
George opposes the Taleban (Figure 16).
If the Danish Prime Minister Anders Fogh
supports George Bush, and Bush supports the
policy of waging war against Iraq, then Fogh,
too, supports waging war against Iraq. We can
use the last observation for action selection, as
required by (10): when Fogh supports waging
war against Iraq, this means that he desires to
assume the Agent role in war activities (which,
in fact, he did).
The rules can be made to cover normal meansend planning: if occupying Iraq (means) promotes
(y) a stable and profitable supply of oil (end), which
George very supports (x), then George promotes
occupation of Iraq (z). Or, if occupation of Iraq,
which George supports (x), necessitates (y) that the
American arms budget is increased, then George
must support the increase, too (z).
We can, in fact, also leave persons out of the
equation: if the occupation of Iraq promotes (x) a
democratic government in the country, and democratic government in Iraq promotes (y) a more
peaceful world, then the occupation (indirectly)
promotes a more peaceful world.

576

C: Taleban

A: George

B: Al Qaeda

In all of these very real examples, the following two simple rules have been used ( is logical
equivalence):
18. zt = (xt-1 y t-1)
19. zt = (xt-1 yt-1)
Z at time t is calculated as the equivalence
of the x and y (or y) relations at time t-1. At the
same time, the rules provide explanations for the
attitudes: z holds because x and y hold. George
opposes the Taleban because they promote AlQaeda, and he opposes Al-Qaeda.
There is one problem, however. As we have
already seen, we cannot assume that such networks of actors and actions are consistent in good
narratives. Therefore, we often will encounter
the phenomenon that the same relation will be
calculated differently if we use two different
triangles that share one side. This is the problem
of overwriting. George Bush endorses an occupation that leads to casualties; therefore, George
ought to be in favor of the casualties (rightmost
triangle in Figure 17. However, he also likes his
voters, who are opposed to casualties. Therefore,
George ought to be opposed to casualties (leftmost
triangle). If the leftmost triangle in Figure 17 wins,
George may change his attitude to the occupation
in order to avoid dissonance (e.g., begin opposing
it). If the rightmost triangle wins, he has to support casualties (which is the strategy of his op-

The Semiotics of Smart Appliances and Pervasive Computing

Figure 17. Contradiction of attitudes

voters

casualties

George

occupation

ponent: the young suicide bombers are celebrated

as martyrs). Machines do everything in a fixed
order, so if the rightmost triangle is processed
last, it always will overwrite the influence from
the left triangle. In order to allow George Bush
to change his mind with respect to the casualties,
we introduce a random factor in the application
of the rules. Sometimes one triangle is chosen,
sometimes the other, and sometimes both. This has
two advantages: (1) all the triangles are allowed
to have an effect, and (2) the thinking of the
software agents is not momentary but proceeds
piecemeal, sometimes overlooking logical inferences in the first round.
However, endorsing the casualties is really no
option at all for George for deeper ethical reasons
than voter-tactical considerations. There are basic
ethical attitudes that are stronger than others and
should not easily be changed. In Western cultures,
the respect for human life is (officially) very high
and cannot easily be changed (publicly). Instead,
contradictions should be allowed to remain in
some cases. We therefore introduce a measure
of strength of the attitudes: weak, medium, and
strong.
We also make a distinction between relations
where the first A-term is a person and those
where it is an action. A person can have subjective desires that can be changed, but actions do
not desire anything. They can enable or cause
another action.

In summary, we make four modifications to

the original simple calculations.
20. Relations between actions only are allowed
to change from an unknown 0-state.
21. Relations between persons and between
persons and actions can change if they are
not strong.
22. The strength of the feelings of a person toward an action propagates to the actions or
persons that enable or oppose the action.
23. Rules are applied with a probability between
100% and 0%.
We distinguish thus between immutable
factual relations between actions, and mutable
subjective attitudes toward actions.
The reason for (22) is the following: if mice
strongly dislike to be eaten and being caught by
cats enables the eating, then mice should strongly
dislike being caught. Cats look quite differently
upon the subject: if they strongly like to eat mice
and catching them enables this, then cats should
be strongly in favor of catching mice.
(20-22) give the network a certain degree
of stability, because they restrict the possible
changes.
The following algorithm uses three simple
data structures. Network is a list of actors and
their type, <actor or action, type>, where type
can be P = participant and A = action. Relation
is a two-dimensional list representing the relation
between actors and actions: Relation[i,j] means
the relation from i to j. The relation can be +,
i supports j, 0, they are indifferent, or -, i
obstructs j.
Finally, Strength is another two-dimensional
list where Strength[i,j] represents the strength of
the relation from i to j. The strength can be +
strong, 0 medium, and - weak.
The rules for overwriting are as follows: a
non-zero relation can be overwritten if and only
if the source of the relation (the Atype below) is
an actor and the relation is not strong. If George

577

The Semiotics of Smart Appliances and Pervasive Computing

has very strong ethical feelings against casualties

in Figure 17, then the rightmost triangle cannot
overwrite it.
The different triangles are classified according to the type of their endpoints: thus, PAA is
a type where an actor, type P, has a relation to
an event, type A, that has a relation to another
event, type A.
It is not possible to evaluate the usefulness of
the individual triangle in isolation. The PAA case
corresponds to the proverb, If you have said A,
you must say B as well. It is not included in the
following algorithm, but it was in the beginning,
since it sounded at least as sensible as PPP, my
friends friend is my friend. In both cases, there
are exceptions. An example is George Bushs
dilemma where A = George, B = occupation, C
= casualties. Bush can evade the responsibility
for the casualties by claiming that they are an
unfortunate and unintended consequence of an
otherwise good deed. In the PPP case, one can
argue that the friend made a mistake in cultivating a disrespectable person and, thus, not extend
ones own friendship to this person.
PAA was deleted from the algorithm, not
because of its inherent qualities but because its
effect in the global network was detrimental, as
we shall see next.
The algorithm is as follows, written in pseudocode:

--- if you like me, I like you.

set the relation b/a to the relation a/b
end if
if x "0" then
repeat for all elements c of the network
if a = c or the strength of a/c + then next
repeat
--- strong relations cannot be changed!!
record the type of element c in Ctype

record Atype+Btype+Ctype in theCase

record relation b/c in y
record relation c/b in y
record relation a/c in z
if y 0 and chance() then
choose between two cases
case (thecase = "PPP" or thecase =

"PPA")
--- relations where a is an agent

set relation a/c to equiv(x, y)

set the strength of z to the strength of x
case (thecase = "AAP" or thecase = "AAA")

and z = 0
--- relations where a is an action
set relation a/c to equiv(x, y)
end choose
end if
if y 0 and chance() then
--- here we know the inverse of y
if (thecase = "PAP" or thecase = "PPA"
or thecase ="PPP" or thecase ="PAA")

RemoveDissonance (network, relations, strength)

repeat for all elements a of the network
record the type of element a in Atype

set the relation a/c to equiv(x, y)

if Atype = "P" and Chance() then

set the strength of z to the strength of x

-- agents like themselves.

set the relation of a to itself to +
end
repeat for all elements b of the network

record the type of element b in Btype

record the a/b relation in x

if Atype = "P" and Btype = "P" and the strength

of b/a +
and Chance() then

578

--- relations where a is an agent

then

end if
end if
end repeat
end if
end repeat
end repeat
end RemoveDissonance

The Semiotics of Smart Appliances and Pervasive Computing

Figure 18. Initial configuration

voters

George

casualties

Figure 19. George maintains his foreign policy

but has run foul of the voters
voters

casualties

George

occupation

occupation

Figure 20. George has changed his policy and

gained popularity
voters

casualties

George

occupation

The algorithm moves the network of attitudes

toward a consonant type by removing dissonances.
If a consonant version exists, it will arrive to it
through a number of iterations.
Georges dilemma can run as follows (with the
PAA possibility included): the seed is Figure 18:
George promotes his voters and strongly promotes
the occupation, the voters are against causalities, and occupation promotes causalities. After
three iterations with a rule probability of 50%,
the situation is as shown in Figure 19: George
still strongly promotes the occupation and, in
addition, the casualties, while he obstructs the
voters. The voters retaliate and strongly obstruct
George as well as the occupation and causalities.
At this time, Georges attitudes were changed in
the simulation so that he now strongly promotes
his voters. After a few iterations, this changes the
whole picture to Figure 20: now George strongly
obstructs occupation and casualties but has gained
the love of his voters.
Thus, change of action potentials is created by
a single change of attitude that ripples through the
network in a change of cause and effects.

If we run the dissonance algorithm on the

StarCatcher, we will see two teams opposing one
another and opposing everything the other team
does. There are no surprises, and this agrees with
the fact that StarCatcher is a simple zero-sum
game. Things are quite different in Bogeyman,
since it contains conflicts and contradictions. The
participants have their own agenda and create alliances to further their own interests. Bogeyman
requires strategic thinking of the children.
In the following, we make some comments
on initial experiments that use Figure 10 as seed.
The algorithm is being elaborated and refined at
the moment.
Figure 21 shows the attitudes developed after
four iterations.
Two main groups seem to be forming: cats,
dogs and children on one side; bogeyman and
mice on the other side. However, there are still
some conflicts: cats and mice like the bogeyman
but are opposed by him.
The simulation recorded a log of the attitude
changes, and this log can be used as explanations
for the attitudes. As the following examples show,
these explanations seem quite understandable:

Cats promote children because both oppose

mice. Cats oppose mice because cats want to
eat mice, which mice dont like. Children, on
their side, oppose mice, because the children
want to take candy, which the mice dont
want them to for the obvious reason that
then it is gone when the mice come to eat.

579

The Semiotics of Smart Appliances and Pervasive Computing

Figure 21. Bogeyman after four iterations

dogs

mice

cats

children

580

bogeyman

Cats promote the bogeyman because they

both oppose mice eating candy. The cats
reason for this is that they are enemies of
the mice, who like eating candycats are
against actions their enemies like to do. The
bogeyman, on the other hand, doesnt want
the mice to take candy, because it hinders him
in catching the childrenit is the childrens
picking up candy that enables the bogeyman
to catch them.
Dogs oppose the bogeyman because they
dont want him to catch children, which he
likes. The reason for the dogs opposition
is that they like to eat sausage, and this is
prevented if the bogeyman catches the children: if they are caught they cannot feed the
dogs.
Dogs promote children because neither of
them wants the bogeyman to catch children.
The reason for the dogs opposition to child
capture is that dogs like to eat sausage, which
is obstructed by the bogeyman catching the
children.
Dogs promote mice because both want the
dogs to catch cats. The mices reason is very
sensibly that dogs catching cats prevents cats
from catching mice.
Mice oppose children because children want
cats to eat mice, which the mice naturally

dont like. Children, on their part, want cats

to eat mice because children want candy,
and cats eating mice supports their candy
hunt since it puts a stop to mice eating the
candy, which prevents the children from
getting hold of it.
The bogeyman opposes dogs because he
doesnt want the children give sausage to
them, but the dogs like very much to be
fed. The reason why the bogeyman opposes
this sausage distribution is that it causes the
bogeyman to run away, which he does not
want, since running away prevents him from
catching the children, which is his goal in
life.
The bogeyman opposes mice because he
wants children to take candy, and mice hinder this. The bogeyman wants children to
take candy, because it enables him to catch
them, which he very much wants.

After the four iterations, most agents thus have

developed understandable attitudes toward most
actions and agents. However, a few of them turned
out to be somewhat strange. Here is an example:
it turned out that cats want dogs to catch them (!)
because they did not want mice to enter sewers.
Dogs catching cats prevents this. The reason why
cats dont want mice to enter the sewer is that they
want to catch the mice, and this is not possible
when the mice are in the sewer. The reason for the
error lies in the seed: we forgot to add that cats
strongly oppose being caught by dogs.
In earlier versions in which the PAA type was
allowed, we got really weird explanations; for
example, the bogeyman opposed children giving
fish to cats because he wanted to catch children,
and this prevented the children from giving fish
to cats. The triangle type is PAA that lets a person
transfer his or her attitude from an event to its
effect, which seems pretty unintelligible in our
case; there is no reason why the bogeyman should
oppose children giving fish to cats, even if he
indirectly prevents it by catching the children.

The Semiotics of Smart Appliances and Pervasive Computing

Evaluation of the Algorithm

The small experiment shows the following:
24. The dissonance algorithm does produce
understandable attitude changes and divides
the participants into groups that support or
help one another.
25. The explanations given for the attitude
changes are understandable.
26. The value of the individual rules must be
evaluated through their global effects on the
network through experiments. It was not
possible to predict that the PAA triangle in
itself would be detrimental.
Let us review our requirement list from the previous sections to see how the algorithm fares:
1.

Does it produce technology that is able to

contribute positively to networks of activities
whose manner of performance we will intuitively call intelligent? The intelligence
of the software agents is clearly a network
phenomenon since attitudes develop in the
context of other agents attitudes.
6. Agents should have intentions to act that are
understandable to the others (comprehensibility). The results show that the attitude
changes are understandable and that the
agents can provide reasonable explanations.
9. Agents should be able to enter and leave the
activity (intermittent participation). Since
the algorithm recursively tries to remove
dissonance in the network, addition or
deletion of agents will not spoil anything
but possibly may make the network move
toward other equilibrium states.
10. Does it conform to the rule that an activity
is executed if all participants are able to
fulfill their roles to some degree and the
Agent filler is strongly obligated and/or
strongly desires to fulfill his or her role. The

algorithm works by distributing desires of

the agents toward actions and other agents.
It thus produces one of the glue-components
that causes agents to act. The other, ability,
is produced by physical movements.
A drawback of the method is that the effects
of changes to the seed and the algorithm are difficult to predict, but this is the price you pay for
designing emergent games. The designer to some
degree loses control over the product.
Another problem could be that the software
agents are too intellectual in the sense that their
motivations are caused by reasoning with many
steps, which children may not be able to follow.
On the other hand, many adventure games have
plots that are more complicated.

Evaluation of the Framework

Did the framework give us good ideas as we hoped?
It turned out it was very productive in generating
ideas for playing with physical vs. fictive reality.
It was also very good at associating activities to
space and participants to activities in a simple
coherent way. Bogeyman took about an hour to
invent. In addition, the framework maintained
a foothold in the computational world and gave
indications of possible technological solutions.
It worked as a boundary object connecting the
aesthetic and the technical world.
The small simulation showed a possible way
of combining games of emergence and narrative
games of progression, a combination that was
judged impossible in Juul (2005).
Because of its robustness, it is very easy to
add or delete participants from the game, which
makes us hope that it can be developed into a
toolkit for the children themselves to modify or
build games.
The diagramming techniques clearly need
to be developed further. Small games are easy
to specify, but more complicated games need
conventions for abstraction and decomposition. It

581

The Semiotics of Smart Appliances and Pervasive Computing

also should be emphasized that we lack empirical

evidence that the games specified in this way are
actually fun to play!

Future Developments
The simulation does not deal with the balance
between reasoning about actions and executing them, although the conditions for executing
actions are well-defined (see (10) in the section
Activities).
The next step, therefore, is to orchestrate the
percolation of desires in the network and the
execution of these desires. One convention could
be that an action is not allowed to influence other
actions before it is executed; only when an action
is realized may it have effect on other actions. An
even stronger restriction is that only participants
who have experienced the execution of the action
are allowed to change their attitudes because of it.
In this way, the network changes would be much
easier to follow, and both solutions will reduce
the part of the network that the algorithm must
search and thereby decrease the complexity of the
algorithm, which regrettably is (n3).
However, the two conventions are probably too
strong. On the one hand, communication probably should count as a kind of execution: when
the Bogeyman tells the mice that he doesnt like
them to eat the candy, this should make the mice
reconsider their positive attitudes toward the bogeyman with the same force as if he had actually
prevented them from eating candy. Similarly, a
promise from the children to give sausage to dogs
should make the dogs side with the children as if
they had actually given sausages to the dogs. On
the other hand, some degree of secret reasoning is
a good literary trickit invites the reader/player
to reconstruct the possible reasons for unexpected
behavior.
But the whole play is, after all, staged for the
benefit of the children. In order for them to create
their tactics, it seems a good idea to inform the
children of the interactions between the software

582

agents so that they can understand what is going on

and make their plans accordingly. Literature and
films have the same problem: how do we inform
the reader/viewer of several concurrent happenings when we can only tell about one event at a
time? The solution is to let the characters meet,
tell stories about their adventures, and let the
reader/viewer overhear the storytelling. Should
we do a similar thing in the game so that when
the children meet a dog, it will tell them about the
feeling and atmosphere in the canine world?
But then again, maybe the children should be
required to seek the information themselves, at
least at advanced levels. The ultimate criterion
for deciding is still: is it fun?

Related Work
In this concluding section, we compare our
approach to a number of related approaches in
which notions like communicative acts, activity,
semantic roles, context, and pervasive games are
central:

The Language Action Perspective (LAP)

Community: This community is inspired
by the works of John Searle and Jrgen
Habermas on communicative actions. The
approach to artifact design in Winograd &
Flores (1986) also has been an inspiration
to the community. LAP basically views use
of IT as the execution of speech acts mediated by technology. The notion of speech
acts is also central to this chapter, but we
extend the concepts to also cover noncommunicative material acts and the relation
between the two. The LAP community has
emphasized the communicative type but to
a certain degree has failed to realize that
communicative actions are intertwined with
material acts in actual work processes. There
is, therefore, a need to build an integrated
theory that encompasses the two (Goldkuhl,

The Semiotics of Smart Appliances and Pervasive Computing

2001). The bias toward communicative

actions, for example, can be seen in the
DEMO method described in Dietz (2003).
It offers a finely grained typology of communicative acts (request, promise, decline,
quit, state, accept) but only distinguishes
between two material acts (deliver, sell).
Another problematic feature, which is also
noticeable in DEMO, stems from Winograd
& Flores (1986): conversation patterns are
primarily described in temporal terms: one
act comes before or after another act, like in
a finite state automaton. The problem is that
mere temporal sequence does not explain
the reason for the communicative pattern,
and it is difficult to relate to organizational
issues. The notion of glue in this chapter offers a more detailed method of description:
actions hang together because one action
changes the abilities, rights, and obligations
of an actor to participate in other activities.
Since qualifications and norms are central
to organizational analysis (Liu, 2000), our
framework seems easier to relate to organization theory.
Activity theory: Activity theory originated
from the dialectical materialist psychology
developed by Vygotsky and his students in
the Soviet Union in the beginning of the
20th century. Activity theory goes beyond
the popular human-machine dyad and insists on cultural and technical mediation
of human activity. Therefore, the unit of
analysis includes technical artefacts and
cultural organization, and the focus of activity theory is much wider than what has
been the core concern of past HCI research
(Bdker & Andersen, 2005). But activity
theory seems to have the opposite problem of
the LAP community: it emphasizes material
activities in which a subject applies a tool
to change some object, and only recently
has spoken and written discourse begun to
figure as mediators in activity theoretical

analyses of work, and the effort to explore

their role as mediators has been limited
(Wells, 2002). The chapter has borrowed the
notion of activity from activity theory but has
extended its three basic conceptssubject
(agent), mediator (instrument), and object
(theme)by means of the theory of semantic
roles.
Actor-network theory (ANT): ANT is not
a real theory but rather a number of related
methodologies and assumptions (Latour,
1999; Law, 1987) focused on empirical
ethnographic fieldwork. This chapter obviously is inspired by this tradition with its
emphasis on the significance of networks
and its (structuralistic) insistence that participants in the network primarily acquire
their properties from their position in the
network and not from inherent resources. We
deviate from the tradition by insisting that
it is possible to refine the vague concept of
actor into a set of more precisely defined roles
in a network. The roles we have chosen are
borrowed from the theory of semantic roles,
and the reason why we believe these roles
to be real is that they have been formalized
in natural languages over many thousands
of years by means of prepositions and case
inflexions. We thus view the roles as the
result of an evolutionary process in which
practical experience through the millennia
has distilled relevant distinctions. Another
deviation is that our framework is biased
toward design of pervasive technology,
which is not the case with ANT.
Object-role management (ORM): ORM
(Halpin, 1998) is an alternative to traditional
entity-relation modeling and object-oriented
modeling. ORM is a method for designing and querying database models at the
conceptual level, where the application is
described in terms readily understood by
users, rather than being recast in terms of
implementation data structures (Halpin,

583

The Semiotics of Smart Appliances and Pervasive Computing

584

1996). An ORM model consists of a set of

objects that play roles similar to actions in
Digital Habitats.
In our understanding of ORM, it resembles
a slot-and-filler grammar similar to ours and
to Fillmores Frame Semantics (Schalley,
2004).
ORM focuses on databases, design, and
querying, and it has strong formal properties
(e.g., formalized conversion to ER/OO/SQL
and nesting). It makes clear distinctions
between issues that belong to the conceptual level (how the users understand the
problem domain) and those that concern
the implementation (e.g., the OO concept of
inheritance tends to become very abstract
from the problem domain perspective).
Thus, ORM is well-suited to validate a
model together with domain experts that
know nothing or little about programming,
because the ORM model can be populated
with sample data in a way that reflects the
domain on the conceptual level. ORM is
supposedly superior to OO when it comes
to transformations over time, since it does
not use attributes (although facts can be
collapsed into attributes for easy overview),
which allows for model changes without
major restructuring. On the other hand, it
only handles the static properties of a model,
according to Halpin (1996, 66), since it is
the most stable part of the model (although
various ORM extensions have been proposed
for process and event modeling).
Compared to ORM, Digital Habitats is wellsuited to the handling of dynamic models,
and ORM lacks the integration of physical
and, to a large extent, pragmatic aspects.
However, ORM is simpler and has a more
formal and consistent design than Digital
Habitats.
Unified eventity representation (UER):
UER (Schalley, 2004) is a quite recent,
UML-based attempt to develop:

a representational framework for verbal

semantics [i.e., verbs, not speech] that is
formal and intuitive at the same time. This
means in effect proposing a framework that
is in principle computer processable on the
one hand, and yet on the other hand whose
representations reflect the wealth and flexibility of natural language in an intuitively
plausible way and in accordance with our
current knowledge about natural language.
(Schalley, 2004, p. 1)

Based on Unified Modeling Language

(UML), it combines computation-oriented
diagrams with a solid linguistic approach
to compositional semantics. It proposes a
diagram-based framework that is precise and
yet potentially underspecified. In the context
of analysis and design of pervasive games,
it seems too fine-grained; however, much
insight can be gained from the analyses and
discussions that are presented in Schalleys
(2004) proposal.
Context-awareness: Context-aware systems stand out as a fairly distinct field within
computer science. The Context Toolkit (Dey,
Salber, & Abowd, 2001) differs from Digital
Habitats in its technocentric orientation. It
is a set of concepts that provides concepts
and standards for things primarily at a
technical level. There is a set of higher-level
concepts, but these are developed on top of
the technical concepts rather than the other
way around.
Pervasive gaming theory: Walther (2005)
offers a set of concepts to classify and analyze pervasive games. They target important
distinctions and issues, and they may well
be used in a generative manner, but compared to Digital Habitats, they are much
more general in nature and do not support
detailed analysis and design.

The Semiotics of Smart Appliances and Pervasive Computing

Conclusion
Based on the comparisons in the previous section,
we can identify six areas in which the present
approach seems to present advantages:

It provides a description of action dependencies that is not a mere temporal sequence

but is relatable to organizational concepts
like qualifications and norms (as compared
to the LAP tradition).
It provides a better framework for describing
communicative as well as material acts plus
the way they hang together (as compared to
the LAP tradition and activity theory).
It provides an explicit link between human
activities and their spatial context (as compared to the technical literature on contextawareness).
It has an explicit dynamic model that precisely describes the conditions for executing
actions (as compared to the ORM methodology).
It offers a typology of participant roles,
based on linguistic evidence that reduces
complexity and, therefore, supports design
processes (as compared to Actor Network
Theory and pervasive gaming theory).

Finally, there are two features that we have

not touched upon in this chapter:

By using semantic roles, it encourages a

system architecture that is verbalizable (i.e.,
which automatically or semi-automatically
can produce understandable descriptions of
themselves) (Andersen, 2004b).
Technological artifacts are described according to their function in activities (Bdker &
Andersen, 2005). Some automatic systems
function as Agents since they initiate and
control the activity, others function as Instruments that do not do anything by themselves,
while still others function as Beneficiaries in

which other participants perform activities

on their behalf.

Acknowledgment
The authors wish to thank the anonymous reviewers for their detailed critical remarks and good
suggestions. Part of this work has been supported
by Center for Interactive Spaces, ISIS Katrinebjerg
(Project #122). Thanks to our colleagues in the
research group Frameworks for Understanding
Software-Based Systems, which has been a great
inspiration for this chapter.

References
Andersen, P. B. (2004a). Habitats: Staging life
COSIGN 2004: Comand art. In Proceedings of
putational Semiotics, Croatia.
Andersen, P. B. (2004b). Diagramming complex activities. In Proceedings of ALOIS 2004,
Linkping, Sweden. Retrieved from http://www.
vits.org/konferenser/alois2004/proceedings.asp
Andersen, P. B. (2004c). Anticipated activities
in maritime work, process control, and business
processes. In K. Liu (Ed.), Virtual, distributed
and flexible organizations. studies in organizational semiotics (pp. 35-60). Dordrecht: Kluwer
Academic Publishers.
Andersen, P. B. (2005). Things considered harmful. In Proceedings of ALOIS 2005, Limerick.
Andersen, P. B., & Nowack, P. (2002). Tangible
objects: Connecting informational and physical
space. In L. Qvortrup (Ed.), Virtual space: Spatiality of virtual inhabited 3D worlds (pp. 190-210).
London: Springer Publishers.
Andersen, P. B., & Nowack, P. (2004). Modeling
moving machines. In P. B. Andersen & L. Qvortrup (Eds.), Virtual applications: Applications

585

The Semiotics of Smart Appliances and Pervasive Computing

with virtual inhabited 3D worlds (pp. 189-211).

London: Springer Publishers.
Bardram, J. E., & Bossen, C. (2005). Mobility
work. The spatial dimension of collaboration
at a hospital. Journal of Computer Supported
Cooperative Work, 14(2), 131-160.
Bardram, J., & Hansen, T. R. (2004). The AWARE
architecture: Supporting context-mediated social
awareness in mobile cooperation. In Proceedings
of CSCW 2004, Chicago (pp. 192-201).
Blake, B.J. (2001). Case. Cambridge: Cambridge
University Press.
Bdker, S., & Andersen, P. B. (2005). Complex
20(4),
mediation. Human-Computer Interaction,
353
-
402.
Brynskov, M. (2004). Digital habitats. Masters
thesis, University of Aarhus, Aarhus, Denmark.
Brynskov, M., & Andersen, P. B. (2004). Habitats,
activities, and signs. In Proceedings of the 7th International Workshop on Organisational Semiotics,
-
151).
OS 2004, Setubal, Portugal (pp. 128
Brynskov, M., Christensen, B. G., Ludvigsen, M.,
Collins, A.-M., & Grnbk, K. (2005). Nomadic
play: A case study of participatory design with
children. In Proceedings of Interaction Design
and Children 2005 (IDC 2005), Boulder, CO.
Cabri, G., Ferrari, L., & Zambonelli, F. (2004).
Role-based approaches for engineering interactions in large-scale multi-agent systems. In C.
Lucena et al. (Eds.), SELMAS 2003, LNC 2940
(pp. 243-263). Berlin: Springer Publishers.
Csikszentmihalyi, M. (1990). Flow: The psychology of optimal experience. New York: HarperCollins.
Dennett, D. (1987). The intentional stance. Cambridge, MA: MIT Press.
Dennett, D. (1991). Consciousness explained.
Boston: Little, Brown.

586

Dey, A. K., Salber, D., & Abowd, G. D. (2001).

A conceptual framework and a toolkit for supporting the rapid prototyping of context-aware
applications. Human-Computer Interaction,
16(2/4), 97-166.
Dietz, J. L. G. (2003). Designing technical systems as social systems. In Proceedings of the
8th International Working Conference on the
Language-Action Perspective on Communication
Modelling (LAP 2003), Tilburg, The Netherlands
(pp. 187
-
207).
Eskola, A. (1973). Socialpsykologi [Social Psychology]. Copenhagen: Munksgaard.
Fillmore, C. J. (1968). The case for case. In E.
Bach & R. T. Harms (Eds.), Universals in linguistic theory (pp. 1-90). London: Holt, Rinehart
and Winston.
Fillmore, C. J. (1977). The case for case reopened.
In P. Cole & G. M. Sadock (Eds.), Syntax and
semantics: 8. Grammatical relations (pp. 59-81).
New York: Academic Press.
Gibson, E., & Rader, N. (1979). Attention: The
perceiver as performer. In G. Hale & M. Lewis
(Eds.), Attention and cognitive development (pp.
6-36). New York: Plenum Press.
Gibson, J. J. (1986). The ecological approach
to visual perception. Hillsdale, NJ: Lawrence
Erlbaum.
Goldkuhl, G. (2001). Communicative vs. material
actions: Instrumentality, sociality and comprehensibility [CMTO Research Papers no. 2001: 06].
Lindkping: Lindkping University.
Greimas, A. J. (1966). Smantique structurale.
Paris: Larousse.
Grice, H. P. (1975). Logic and conversation. In
P. Cole & J. Morgan (Eds.), Syntax and semantics: Vol. 3, Speech acts (pp. 43-58). New York:
Academic Press.

The Semiotics of Smart Appliances and Pervasive Computing

Haase, J., Musaeus, P., & Boisen, E. (2004).

Virtual reality and habitats for learning microsurgical skills. In P. B. Andersen & L. Qvortrup
(Eds.), Virtual applications: Applications with
virtual inhabited 3D worlds (pp. 29-48). London:
Springer Publishers.
Halpin, T. (1996, October). Business rules and
object role modeling. Database Programming
& Design, 9(10), 66-72.
Halpin, T. (1998). Object-role modeling. In P.
Bernus, K. Mertins, & G. Schmidt (Eds.), Handbook on architectures of information systems (pp.
81-104). Berlin: Springer-Verlag.
Juul, J. (2005). Half-real. Video games between
real rules and fictional worlds. Cambridge, MA:
MIT Press.
Kaplan, F., & Hafner, V. (2004). The challenges
of joint attention. In Proceedings of the 4th International Workshop on Epigenetic Robotics,
Genoa, Italy (pp. 67-74).
Koster, R. (2005). A theory of fun for game design.
Phoenix, AZ: Paraglyph.
Kristensen, B. B. (2002).
Associative modeling
and programming. In Proceedings of the 8th
International Conference on Object-Oriented
Information Systems (OOIS 2002), Montpellier,
France.
Kristensen, B. B. (2003). Associations: Abstractions over collaboration. Proceedings of the 2003
IEEE International Conference on Systems, Man
& Cybernetics, Washington, DC.
Latour, B. (1994). On technical mediation: Philosophy, sociology, genealogy. Common Knowledge, 3, 29-64.
Latour, B. (1999). Pandoras hope. Cambridge,
MA: Harvard University Press.
Law, J. (1987). Technology and heterogeneous
engineering: The case of Portuguese expansion.
In W. Bijker, T. P. Hughes, & T. Pinch (Eds.), The

social construction of technological systems (pp.

111-134).
Cambridge, MA: MIT Press.
Licklider, J. C. R. (1960). Man-computer symbiosis. IRE Transactions on Human Factors in
Electronics HFE-1, 4-11.
Lind, M. (1994). Modeling goals and functions
of complex industrial plants. Journal of Applied
Artificial Intelligence, 8, 259-283.
Liu, K. (2000). Semiotics in information systems
engineering. Cambridge: Cambridge University
Press.
Maes, P. (1994). Agents that reduce work and
information overload. Communications of the
ACM, 37(7), 31-40.
Maglio, P. P., & Campbell, C.S. (2003). Attentive agents. Communications of the ACM, 46(3),
47-51.
Mathiassen, L., Munk-Madsen, A., Nielsen, P.
A., & Stage, J. (2001). Object-oriented analysis
& design. Aalborg: Marko.
May, D., & Kristensen, B. B. (2003). Habitats for
the digitally pervasive world. In P. B. Andersen,
& L. Qvortrup (Eds.), Virtual applications: Applications with virtual inhabited 3D worlds (pp.
141-158). London: Springer.
May, D., Kristensen, B. B., & Nowack, P. (2001).
Tangible objects: Modeling in style. In Proceedings of the Second International Conference on
Generative Systems in the Electronic Arts (Second
Iteration: Emergence), Melbourne, Australia.
Piaget, J. (1962). Play, dreams and imitation in
childhood. New York: Norton.
Propp, V. (1968). Morphology of the folktale.
Austin: University of Texas Press.
Roth, G., & Dicke, U. (2005). Evolution of the
brain and intelligence. Trends in Cognitive Sciences, 9(5), 250-257.

587

The Semiotics of Smart Appliances and Pervasive Computing

Russell, S., & Norvig, P. (2003). Artificial intelligence: A modern approach. Upper Saddle River,
NJ: Prentice Hall.
Ryan, M. L. (1991). Possible worlds, artificial
intelligence and narrative theory. Bloomington:
Indiana University Press.
Ryle, G. (1970). The concept of mind. London:
Hutchinson.
Salen, K., & Zimmerman, E. (2004). Rules of
play. Cambridge, MA: MIT Press.
Sargent, R., Resnick, M, Martin, F., & Silverman,
B. (1996). Building and learning with programmable bricks. In Y. Kafai & M. Resnick (Eds.),
Constructionism in practice. Designing, thinking,
and learning in a digital world. Hillsdale, NJ:
Lawrence Erlbaum.

Sutton-Smith, B. (Ed.). (1979). Play and learning.

New York: Gardner.
Tomasello, M. (1995). Joint attention as social cognition. In C. Moore, & P. J. Dunham (Eds.), Joint
attention: Its origins and role in development (pp.
103-130). Hillsdale, NJ: Lawrence Erlbaum.
Tomasello, M., Kruger, A. C., & Ratner, H. H.
(1993). Cultural learning. Behavioral and Brain
Sciences, 16, 495-552.
Vygotsky, L. (1976). Play and its role in the mental
development of the child. In J. Bruner, A. Jolly,
& K. Sylva (Eds.), Play: Its role in development
and evolution. New York: Basic Books.
Walther, B. K. (2005). Atomic actionsmolecular
experience: Theory of pervasive gaming. ACM
Computers in Entertainment, 3(3), 7-13.

Schalley, A. (2004). Cognitive modeling and

verbal semantics: A representational framework
based on UML. Berlin: Mouton de Gruyter.

Weiser, M., & Brown, J. S. (1995). Designing

calm technology. Retrieved from http://www.ubiq.
com/hypertext/weiser/calmtech/calmtech.htm

Steels, L. (2003). Evolving grounded communication for robots. Trends in Cognitive Science,
7(7), 308-312.

Wells, G. (2002). The role of dialogue in activity

theory. Mind, Culture and Activity, 9(1), 43-66.
Winograd, T., & Flores, F. (1986). Understanding
computers and cognition. Norwood, NJ: Ablex.

This work was previously published in Semiotics and Intelligent Systems Development, edited by R. Gudwin, J. Queiroz, pp.
211-255, copyright 2007 by IGI Publishing, formerly known as Idea Group Publishing (an imprint of IGI Global).

588

589

Chapter 2.13

Analyzing the Factors

Influencing the Successful
Design and Uptake of
Interactive Systems
to Support Social Networks
in Urban Neighborhoods
Marcus Foth
Queensland University of Technology, Australia

ABSTRACT
In urban residential environments in Australia
and other developed countries, Internet access
is on the verge of becoming a ubiquitous utility
like gas or electricity. From an urban sociology
and community informatics perspective, this
article discusses new emerging social formations
of urban residents that are based on networked
individualism and the potential of Internet-based
systems to support them. It proposes that one of
the main reasons for the disappearance or nonexistence of urban residential communities is a lack
of appropriate opportunities and instruments to
encourage and support local interaction in urban
neighborhoods. The article challenges the view

that a mere reappropriation of applications used to

support dispersed virtual communities is adequate
to meet the place and proximity-based design
requirements that community networks in urban
neighborhoods pose. It argues that the key factors
influencing the successful design and uptake of
interactive systems to support social networks
in urban neighborhoods include the swarming
social behavior of urban dwellers; the dynamics
of their existing communicative ecology; and the
serendipitous, voluntary, and place-based quality of interaction between residents on the basis
of choice, like-mindedness, mutual interest and
support needs. Drawing on an analysis of these
factors, the conceptual design framework of a
prototype system the urban tribe incubator
is presented.

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Analyzing the Factors Influencing the Successful Design and Uptake of Interactive Systems

Introduction
The area of technology and human interaction
is cross-disciplinary and requires many different academic fields and design practices to work
together effectively in order to generate a better
understanding of the social context and human
factors in technology design, development, and
usage. This article focuses on the social communication aspects of this field and hopes to
establish a greater awareness of the contribution
that communit
y media and communication studies can deliver to
the field of human computer interaction. It seeks
to build a theoretical foundation for an analysis
of two interrelated issues, which are discussed
in turn.
First, the importance of place and the continued
purpose and relevance of urban neighborhoods are
established. New media and networked information and communication technologies have not led
to the diminishment of local place and proximity. However, they have given rise to new types
of social interaction and to new emerging social
formations. Understanding the nature and quality
of interaction in these new social formations can
inform the successful animation of neighborhood
community and sociability.
Second, appropriate opportunities and instruments to encourage and support local interaction
in urban neighborhood networks are not limited
to technology, but technology can be a key facilitator. Thus, system designers and engineers are
crucial allies to social scientists in the search for
hybrid methodologies that integrate community
development approaches with technology design.
The article questions whether it is sufficient to
appropriate tools originally designed for dispersed online (that is, virtual) communities in
the context of community networks (Schuler,
1996) for urban neighborhoods. Purpose-built
tools and instruments are required that afford

590

(a) interactive linkages between the residents

communicative ecologies of cyberspace and local place; and (b) personalized social networking
between proximate neighbors of choice. Such an
approach would allow the nonvirtual and placebased assets in a residents portfolio of sociability
to become more attractive. It would establish an
opportunity to create and to maintain local social
ties and, ultimately, to find out who is living next
door and who is personally compatible.
From the discussion of these issues, some of
the key factors influencing the successful design
and uptake of interactive systems to support social
networks in urban neighborhoods are derived.
Drawing on an analysis of these factors, the conceptual framework of a prototype system the
urban tribe incubator is presented.
This article seeks to set up the interdisciplinary conceptual foundation necessary to drive
a thorough theoretical and empirical investigation into the interaction of people, place, and
technology and the way they function together
to facilitate access to the social and cultural life
of cities. The purpose of this article is not only
to introduce and illustrate the issues at stake
and to present a design framework but also to
stimulate transfer and exchange of knowledge
across academic disciplines and especially to
invite discussion and comment from a broader
interdisciplinary audience. Supporting efforts to
build bridges between the social and engineering
sciences is paramount to the field of technology
and human interaction, and this article contributes
to the development of a dialogue between these
disciplines. An interdisciplinary approach that
brings together views and expertise from sociology, urban studies, interaction design, and related
disciplines will assist with efforts to facilitate
urban neighborhood community building, social
inclusion, public consultation and debate, fair
access to local information and services, urban
sustainability, and healthier local economies.

Analyzing the Factors Influencing the Successful Design and Uptake of Interactive Systems

Technical affordances and

new social formations in the
context of networked
individualism
The Internet has found its way into many households of urban dwellers in Australia and other
developed countries to the extent that Internet
access is on the verge of becoming a ubiquitous
utility like gas and electricity. The Internet has
advanced to become a communication tool that
coexists with other established communication
devices such as the telephone, short message
service (SMS), new media, and face-to-face
interaction. E-mail, instant messaging, online
chats, and other online applications are now
instrumental in establishing and maintaining
social ties with family, friends, coworkers, and
other peers, thus creating a private portfolio of
sociability (Castells, 2001, p. 132).
The Internet has entered peoples everyday life
and plays a significant role in the communication
pattern of urban residents. The Internet has not
substituted but supplemented off-line interaction
with online interaction (Fallows, 2004; Wellman
& Haythornthwaite, 2002). People still chat on the
phone and meet face-to-face. However, the Internet as well as mobile communication devices such
as mobile phones, laptops, and personal digital
assistants (PDA) allow people to maintain social
ties in different ways by taking advantage of new
features. The mobile phone introduced placeindependent communication, and the emerging
third- and fourth-generation (3G, 4G) mobile
telephony adds audiovisual telepresence. E-mail
and SMS afford asynchronous communication
and notification mechanisms. Online chats offer
broadcast-style, many-to-many communication,
whereas private chat rooms enable users to engage in multiple peer-to-peer dialogues. Instant
messaging tools combine the features of online
chat rooms with ambient awareness by adding
availability or other status information to a users
nickname (e.g., Jean|busy, Lucy|out to lunch).

However, these tools are used more often to

connect with family, friends, coworkers, and
peers and less with neighbors. The telephone has
evolved into a ubiquitous communication device,
but it has not contributed per se to overcoming
urban alienation. Sociologists such as Wellman
(2001, 2002) and Wellman et al. (2003) describe
how people construct their social networks with
the help of the telephone and other devices. Wellman argues that, while people become more accustomed with the features these tools offer, the
nature of the social ties that people establish and
maintain changes from door-to-door and placeto-place relationships to person-to-person and
role-to-role relationships. He creates a holistic
theoretical framework that builds on the dual nature in the interplay between community and the
individual. He describes the emerging qualities of
this behavior as networked individualism.
Residential areas, such as apartment buildings, townhouse complexes, master-planned
community sites, and the residents and tenants
of these units form the focal point in this article
to examine the interplay between people, place,
and technology. The results and findings of this
theoretical analysis will help to shed light on some
aspects of the community question, especially
the continued purpose and relevance of neighborhoods in urban habitation, by investigating
the ironic relationship between endemic urban
alienation and the widespread use of mobile and
ubiquitous communications technology by urban
dwellers that allows them to interact with each
other (Walmsley, 2000).
Before this technology became ubiquitous
and entered the everyday life of city dwellers,
predecessors and variations had been designed for
or had first become popular in workplace-based
environments in order to support communication
and collaboration among professionals. This was
followed later by their diffusion into everyday life
and their reappropriation for social use. The act of
reappropriation (e.g., from the professional use of
a pager to the social use of SMS) implies that there

591

Analyzing the Factors Influencing the Successful Design and Uptake of Interactive Systems

are opportunities to design and develop purposebuilt systems from the ground up, which, instead
of merely trying to make ends meet, take the
unique requirements into account of the social- and
place-based context in which they are used. Tools
to animate and network urban neighborhoods
require a consideration and treatment of notions
of sociability, place, privacy, and proximity in
order to take full advantage of the communicative opportunities that this environment offers its
inhabitants and the wider society.

Place matters:
Communication and
interaction in urban
neighborhoods
Tnnies (1887) idea of community as Gemeinschaft implies a well-connected, place-based,
collective, village-like community. However,
this notion of community represents an overly
romanticized image of community and ignores
more contemporary forms of community that have
been explored by recent sociological studies (Wellman, 2001, 2002). Gemeinschaft might resemble
Hobbiton in the Shire described by Tolkien (1966).
This communitarian notion (de Tocqueville, 2000;
Etzioni, 1995) is still referred to frequently in the
community development literature, although the
homogeneous, egalitarian, and all-encompassing
nature of Gemeinschaft is a utopian ideal that
is less and less compatible with contemporary
characteristics of community as social networks
in todays network society.
Before the advent of modern information and
communication technology, human interaction
was limited by the reach of the physical presence of self or the representations of self (e.g.,
letters and photographs) and available means
of transportation. The need to socialize and to
communicate was usually satisfied with family
members in the same household, with friends and
peers nearby, at work, or within the vicinity of the

592

neighborhood people lived in. Human relations

were door-to-door or place-to-place (Wellman,
2001). The fact that people residing in the immediate surroundings were known also established
a feeling of security, community identity, and a
sense of belonging a feeling that clashes with
the experience of living in todays high-density,
compact urban environments.
The invention and introduction of new information and communication technologies into society has usually been accompanied by foresights
that predict that people will be less dependent on
place and location. To an extent, this is true. The
phone was the first major invention to introduce
personal telepresence and to allow everybody to
communicate in real time with others outside their
own physical locality. Instead of being restricted
to people within proximity of oneself, the phone
enabled long-distance communication to maintain work and social relationships. However, it is
unlikely that anyone lifts the telephone handset
to introduce themselves to a neighbor nearby that
they have not met before.
The Internet affords both synchronous and
asynchronous applications that enable communication between one or multiple users, one-to-many
or many-to-many broadcasts to a closed group,
and public announcements to an open audience.
The abstract nature of Internet-mediated communication gave rise to the widespread use of
the metaphor cyberspace, which visualizes the
emergence of a new spatial dimension.
However, peoples bodies cannot be atomized
in the same way that their audiovisual representations can be digitized, mediated, and sent
across the world. Thus, people depend and will
continue to depend on place and locality and on
co-located face-to-face interaction. Bits and bytes
travel in the virtual space of flows spanned by
the Internet, but humans travel in the physical
space of flows that modern transportation affords.
Place and proximity continue to matter in every
socioeconomic context, because there are no Internet applications that can completely substitute

Analyzing the Factors Influencing the Successful Design and Uptake of Interactive Systems

real-time, co-located, face-to-face interaction.

This is evident by rising car and air travel sales
(Wellman, 2001), by people commuting to work
instead of working from home and by the formation of economic clusters, precincts, and hotspots
where industries based along the same value chain
co-locate to take advantage of synergy effects.
Florida rightly argues that the economy itself
increasingly takes form around real concentrations of people in real places (Florida, 2003, p.
4). In light of new urbanism (De Villiers, 1997)
and master-planned residential community sites
(Gleeson, 2004), his statement holds true not just
for the economy but for society, in general.
Attempts to bridge distance for the purpose of
more than just communication have seen initiatives such as telework and distance education,
yet they remain at the edge of mainstream usage
and have not replaced face-to-face interaction
(Dhanarajan, 2001; Gillespie & Richardson,
2004). To enable economic efficiencies, the goal of
Computer Supported Co-operative Work (CSCW)
and groupware applications is to supplement not
substitute place-based work practices.
Wellman (2002) points out that the dichotomies
of physical place and cyberspace or of online
and off-line are misleading. Even as the Internet
grows exponentially, place-based units such as
home, work, and school remain at the core of our
understanding of everyday life. The Internet and
other information and communication technology
add new qualities to the portfolio of communication tools available to us, enriching our communicative ecology and adding on to the variety of
media channels at our disposal. We do not rely on
the central location of traditional meeting places
anymore, such as the marketplace or town square,
in order to meet with friends and peers. Instead,
we use mobile communications technology that
we can carry around (e.g., mobile phone, SMS) or
ubiquitous communications technology that we
can access anywhere (e.g., wireless networks) not
to avoid but to negotiate on-the-fly meeting places
and venues anywhere and anytime. Teenagers, for

example, use their mobile phones to arrange meeting places on the spot; this could be the local caf,
the shopping mall, or someones home (Satchell,
2003). This emerging behavior introduces challenges to conventional understandings of place
and public places and opens up opportunities for
residential architecture, town planning, and urban
design (Castells, 2004; Florida, 2003; Grosz, 2001;
Horan, 2000; Mitchell, 2003; Oldenburg, 2001;
Walmsley, 2000).
In a lively online discussion about the continued purpose and relevance of neighborhood
communities, one participant (eric_brissette,
2004) illustrates the point that having less exposure to neighbors (as opposed to coworkers or
friends) does not mean that it is less likely that
there are, in fact, prospective friends living in
the neighborhood:
I guess it all depends on where you live. I live in
a rural town of about 10,000. Most people say
hello or good morning to you as you pass
them on the sidewalk. I cant say Ive known all
of my neighbors well, but I have at least spoken
with them enough to know a bit about who they
are. Visiting larger cities like Boston or New York
makes me feel weird. Nobody looks you in the eye,
and everyone seems constantly pissed off, almost
like everyone is scared of everyone else ... yet this
all seems perfectly normal to them. ... Chances are
good that there are people in your neighborhood
that share your [interests] or are at least [compatible] at the personality level who you wouldnt
normally interact with on a daily basis.
In todays networked society, it is questionable to project the image of the rural village and
use it as a best practice urban village model for
a city because of inherent differences between
both places and their inhabitants. Yet, the specific
characteristics of a city can give rise to a different model of urban village that acknowledges
the potential opportunities that this particular
environment offers its residents. For example,

593

Analyzing the Factors Influencing the Successful Design and Uptake of Interactive Systems

the simple fact that a city accommodates a larger

number of residents could offer the individual
greater choice and, thus, a chance to find the right
social interaction partners.
However, the motivation for and process of the
search itself remains to be examined. Getting to
know someone in their role as a neighbor is less
likely than getting to know them in their role as a
coworker or being the friend of a friend. Neighbors
may still be part of a residents social portfolio,
but the communication devices used to maintain
these ties are inherently place-independent and
ephemeral: A phone call or an e-mail does not
distinguish between close or distant friends.
Proximity does matter when it comes to physical encounters and face-to-face meetings. Most
frequent social ties, including online interaction,
are maintained with people who can easily be
reached physically; that is, they usually reside
within the same city, the surrounding suburbs,
or the same neighborhood (Horrigan, 2001; Horrigan et al., 2001). The majority of phone calls,
SMS, and e-mails helps the parties involved to
coordinate meetings or social gatherings (e.g., to
catch up over coffee in a caf nearby).
These ties are based primarily on common
friendship, workplace, or interest, and not on
shared locality. We may be introduced and
subsequently get along well with the friend of a
coworker who happens to live on the same street,
but it is unlikely that we would have found out
about that person without the coworker introducing us first.
Many urban neighborhoods are the result of
what town planners and developers call masterplanned communities. Traditional conceptual
models of community development limit action
to tangible places of public interaction such as
kindergartens, public schools, parks, libraries,
and so forth (Gleeson, 2004). This build-it-theywill-come approach lacks engagement with the
findings of recent community development research (Gilchrist, 2004; Pinkett, 2003). It ignores

594

both the human factors involved in urban renewal

and sociocultural neighborhood animation as
well as the potential that information and communication technology can offer urban residents
such things as online community networks and
location-based new media (Day & Schuler, 2004;
Rheingold, 2002).
Gilchrist points out that community development involves human horticulture rather than
social engineering (Gilchrist, 2000, p. 269).
Social encounters in urban neighborhoods cannot
be master planned. They are based on coincidence
and serendipity. Neighbors meet through friends
of friends, who happen to live close by; they
meet when walking the dogs or, in some cases,
when a local problem affects multiple residents
(Hampton, 2003). However, more often than not,
they do not meet at all, and even if they wanted
to, there is usually little opportunity beyond serendipity. Our preliminary results indicate that the
majority of residents surveyed believe, just like
Eric, that chances are good that there are people
in their neighborhood who share their interests
or with whom they are at least compatible at the
personality level, people they normally do not
interact with on a daily basis. For those who would
like to find out about them and who still believe
in good neighborhood relations, the question
remains: What can be done to avoid relying on
good fortune and fate? How can those who want
to, coax luck?
A step toward a more strategic approach to
develop urban neighborhoods encompasses online community networks (Schuler, 1996). Community networks are integrated online systems
designed for residential communities that, so far,
have usually been comprised of communication
tools such as mailing lists, discussion boards, and
newsletters. Ideally, community networks allow
residents to communicate and interact with other
users and to take advantage of the proximity to
other residents in the neighborhood. Thus, these
systems have the potential to build a bridge between virtual public spaces and physical public

Analyzing the Factors Influencing the Successful Design and Uptake of Interactive Systems

places and to foster network social capital and

neighborhood identity.

Community networks in
urban neighborhoods
Arnold (2003) states that for the ordinary citizen,
social interaction is the killer application of the
Internet (p. 83). This development has sparked
an increased interest among researchers from a
range of disciplines to investigate online communication and online communities (Preece, 2000).
Yet, the majority of the work undertaken so far in
this research field focuses on globally dispersed
online (virtual) communities and not on the use
of information and communication technology for
communities of place (Papadakis, 2004).
There is a small but growing body of literature
that reports on the use of information and communication technology for community development in place-based contexts, mostly within the
emerging discipline that Gurstein (2000, 2001)
terms community informatics. However, most of
these accounts investigate communities that are
in one way or another deprived (e.g., telecenters
or community access centers in rural and remote
locations; ICT for development and poverty reduction in developing countries). The transferability
of these studies to urban settings is questionable.
Urban dwellers may think of themselves as being
quite well-off and may lack common disadvantages, such as low income or unemployment. Such
instances of deprivation could contribute to shared
agony, which ultimately may help to establish a
collective need for change (Foth, 2004b) and, thus,
a reason to make use of technology for action
and change. In its absence, however, alternative
motivations to form neighborhood community
need to be found.
Today, the value of door-to-door and place-toplace relationships in urban neighborhoods seems
to be on the decline. Researchers and practitioners

endeavor to counter this trend through community

networking; that is, the application of Internetand Web-based tools in residential environments
to introduce and sustain local communication
and interaction among neighbors (Day, 2002).
Although the term is sometimes used broadly in
other contexts of community development and
community informatics, the focus in this article
is on urban neighborhoods.
A residential community comprises people
who live or stay in a geographically demarcated
area. Such communities are sometimes also
referred to as local communities, physically or
geographically based communities, or communities of place. Apart from the fact that members of
a residential community share the same location
or address, they are not bound necessarily by any
other common characteristic such as interest, age
group, or occupation. As such, residential communities are not communities or neighborhoods
a priori. An apartment complex might consist of
residents who do not know each other.
A range of research projects have been undertaken to examine whether online community
networks can facilitate the process of establishing neighborhood identity. These projects set
out to design and implement online community
networks for both large and small residential sites
with various aims and with varying degrees of
success (Arnold, 2003; Carroll & Rosson, 2003;
Cohill & Kavanaugh, 2000; De Cindio et al.,
2003; Hampton & Wellman, 2003; Meredyth et
al., 2004; Pinkett, 2003).
Reaching a critical mass of users is considered
to be one of the key criteria of success (Arnold et
al., 2003; Butler, 2001; Patterson & Kavanaugh,
2001) and has been reported as one of the most
common stumbling blocks: If you build it, they
will not necessarily come (Maloney-Krichmar
et al., 2002, p. 19). This statement seems to be
common sense; nonetheless, it provides the opportunity for a deeper analysis of the reasons and
motivations for urban residents to communicate,

595

Analyzing the Factors Influencing the Successful Design and Uptake of Interactive Systems

interact, and get together with other residents and

to participate actively in an urban neighborhood
network.
Dunbar (1996) suggests that the size of human social networks is limited for biological
and sociological reasons to a value of around 150
nodes. Barabsi (2003) and Watts (2003) provide
a more far-reaching overview of recent advances
in network theory and their impact on business,
science, and everyday life. Some ideas are crucial
in understanding community networks: They
usually increase or decrease in size, that is, social
network research and systems design need to find
ways to capture their dynamics. Their structure
is not random or chaotic but follows preferential
attachment (rich get richer) and fitness (fit get
richer). In the context of communities of place,
Jankowski and his colleagues support this thesis
with empirical research by pointing out that those
geographic communities already rich in social
capital may become richer thanks to community
networks, and those communities poor in social
capital may remain poor (Jankowski et al., 2001,
p. 113). Hampton & Wellman (2003) support this
notion by stating that connectivity seems to go
to the connected: greater social benefit from the
Internet accrues to those already well situated
socially (p. 283). Then, the next questions are,
what constitutes richness and fitness in urban
social settings, how do residents get rich (and
become a hub in their social network), and how
can community networks facilitate enrichment
in a fair and ethical manner?
The reasons and motivations for participation
in dispersed online (virtual) communities provide
further insight into the answers to these questions.
A person suffering from cancer might prefer
the expertise, empathy, and perhaps anonymity
available in an international online community
of cancer patients. Philatelists will find more
like-minded people in an appropriate virtual
community of interest, such as a newsgroup or
discussion board that is open to any Internet
user and that is not restricted to the residents of

596

just one apartment complex or one suburb. The

impossibility or impracticability of a face-to-face
exchange in a dispersed online community does
not usually impact negatively upon the value that
participants derive from such online interactions.
The large number of active online communities
tells its own tale.
The core characteristic of such dispersed
online communities is their collective nature;
that is, they accumulate participants who share
a common interest, profession, or support need
with an entity that acts as a collective group with
a shared purpose. The tools that are used to support online communities, including mailing lists,
newsletters, discussion boards, and so forth, are
more closely designed toward a many-to-many
broadcast approach instead of a peer-to-peer
networking approach. They assume a pre-existing
motivation to participate in and use the virtual
space. In the case of shared interest, profession,
or support need, that may be the case. However,
in the case of residents of urban neighborhoods,
the only shared attribute is place and colocation.
Apart from occasions where an item of discussion
or a topic of interest relates directly to the shared
place that residents coinhabit, most interaction
is located within place but not necessarily about
place. Thus, place and proximity are insufficient
attributes to attract residents to a community
network and to sustain it. Furthermore, a reappropriation of the tools used to support online
(virtual) communities in the context of urban
neighborhood networks opens up further issues,
because a community of place is inherently different from a dispersed community of interest.
In addition, connectivity per se does not ensure
community, and proximity does not ensure neighborliness (Foth, 2003, 2004b).
The unique selling proposition that could give
online community networks for urban neighborhoods a competitive advantage over dispersed
online communities is proximity. Community
networks allow residents to interact online and to
take and continue online interaction off-line, in

Analyzing the Factors Influencing the Successful Design and Uptake of Interactive Systems

real life, and face to face with other residents who

live in the same location. As such, they can be an
effective tool for local community engagement
and activism, if the community faces a shared
problem or a common enemy that provides the
required motivation for residents to come together.
Hampton (2003) describes the experience with
residents in Netville, who faced the prospect of
losing broadband Internet access, which previously had been provided to them free of charge.
The issue and the presence of a common enemy;
that is, the Internet service provider, unified
residents in community activism to advocate for
a continuation of the service, and the traffic in
the online community network (in the form of an
electronic mailing list) increased significantly.
The unifying vigor of a common problem or issue
can (temporarily) transform a certain number of
residents into a residential collective and, thus,
sustain an online community network (Foth &
Brereton, 2004).
In the absence of a common enemy, a shared
purpose or a pre-existing village-like atmosphere,
are there other reasons and motivations for social
encounters to occur and for the formation of
residential networks in urban neighborhoods?
Examining existing urban communities may help
to answer this question. Watters (2003) describes
the emergence of clusters of under-35-year-old
urban dwellers mostly in America but also in
other parts of the word as urban tribes. They
represent a social network, a swarming group of
friends who live in the same city and who are all
connected with each other through strong and
weak ties. The interaction between members of
urban tribes is facilitated through the use of mobile phones, e-mail, and face-to-face gatherings.
Watters (2003) does not mention the use of neighborhood or similar ICT-supported networks, but
his account of the behavior of urban tribes allows
one to imagine a new generation of purpose-built
interactive community networks for residents in
urban neighborhoods.

The urban tribe incubator:

Networking serendipitous
social encounters in Urban
neighborhoods
The previous section discussed the conditions
under which residents might ultimately engage
in neighborhood community networks and, thus,
talk to people within their vicinity. In order for
these conditions to emerge, competitive tools
need to be designed that allow residents to find
out who is living around them and that facilitate
local communication and interaction that so
far has relied on coincidence and serendipity.
However, conventional community networks do
not necessarily address these needs. They are
very delicate, organic entities. They thrive only
in favorable circumstances (e.g., similar demographic and professional orientation) with special
nurturing (e.g., free Internet access) (Hampton
& Wellman, 2003; Kavanaugh et al., 2003), and
chances are high that, otherwise, they may fail
(Arnold et al., 2003).
The findings of these sociological studies
provide essential insights for a new design methodology that can guide the successful development of interactive systems and devices that can
stimulate local interaction and animate urban
neighborhoods. A prototype system of an urban
tribe incubator is currently being developed and
tested in three urban residential sites in Australia
(Foth, 2004a). Action research (Hearn & Foth,
2005) and participatory design (Churchill et
al., 2004; Greenbaum & Kyng, 1991; Schuler &
Namioka, 1993) play crucial roles in iteratively
constructing and testing a successful prototype.
The participation of residents in the design and
development is essential to integrate the range
of communication channels that they use and to
allow residents to take social ownership of the
system.
The previous discussion of the factors influencing systems that support social networks in
urban neighborhoods gives rise to a set of design

597

Analyzing the Factors Influencing the Successful Design and Uptake of Interactive Systems

considerations that are being integrated into the

design of the urban tribe incubator prototype.
These are now discussed in turn.

Size, Growth, and Critical Mass

Popular services and functions in conventional
community networking systems include electronic
newsletters, mailing lists, and discussion boards.
In order to keep these systems interesting and appealing, content needs to be generated either by
a systems administrator or a delegate but ideally
by the community of users itself. Thus, a critical
mass of users is required to maintain an ongoing
supply of discussion board postings and responses,
mailing submissions, and newsletter contributions. It requires residents to invest a reasonable
amount of time and effort in order to collectively
sustain the systems viability.
The urban tribe incubator may include such
collective, broadcast-style, many-to-many functions, but the core will be a residents directory
that does not require maintenance on a regular
basis, unless details have changed and need to
be updated. A residents personal profile may
comprise information about skills, trade, interests,
hobbies, and contact details. The profile becomes
the virtual representation of a potential node that
invites other residents to link to and from. The
system does not require users to use the directory on a regular basis in order to interact with
all other users. Rather, the system allows users
to opt in and opt out as they please and as a need
arises by facilitating personalized networking;
that is, voluntarily initiating contact and building
social ties with people of their choice. Thus, the
directory becomes the catalyst for personalized
peer-to-peer social networks to form.
The size and growth of the directory itself is
in no linear relation to the size and growth of an
individual residents social network. The system
acknowledges different levels of social richness
and fitness, and thus, the point of saturation remains a personal preference. If an individuals

598

personal limit of social saturation is reached, he

or she can opt out. In conventional community
networks, for example, users usually cannot control how many people will respond to their posting
on a discussion board: It may be none, or it may
set off an avalanche of responses. In an instant
messenger application, however, users remain
in control of the social network with which they
engage their private buddy lists.

Diversity, Individualism, and Choice

The urban tribe incubator is not designed to host
an online community of a particular interest or
support need but rather allows for the diversity
of individual residents with different interests
and support needs to find each other and to form
smaller social clusters. The system presents
residents with choice in relation to the number
and characteristics of communication partners
and modes of interaction. It provides easy and
convenient ways for residents to identify birds
of a feather; that is, to find like-minded people
with common interests or support needs.
The system raises awareness among residents
of who is living around them in order to facilitate
peer-to-peer connections. The resident directory
that links to individual profiles allows residents
to choose what personal information they publish
online or whether to keep certain information
private or only available upon request. The goal
of a resident directory is not to facilitate residents
initiating virtual contact first (although it can be
used in this way) but rather to simplify the process
of strengthening serendipitous social encounters
that happen while walking the dog. Without an
urban tribe incubator, such informal contacts that
have the potential to develop into rich interaction
may remain superficial and transitory.
The system does not require residents to keep
communication within the system but allows them
to move it to other synchronous or asynchronous
communication platforms and devices. Having
access to an online directory, a resident is able to

Analyzing the Factors Influencing the Successful Design and Uptake of Interactive Systems

maintain contact with a new acquaintance and to

integrate this contact into their established usage
of existing personal peer-to-peer communication
devices, such as instant messenger, e-mail, SMS,
and online chat.

Privacy and Social Control

In order to safeguard privacy, residents have
control over their personal information and the
scope of their online engagement. Enhanced local sociability is welcomed by most residents but
must not come at the cost of losing security and
control of the voluntary and selective nature of
ones social networks. Our preliminary results
are encouraging insofar as residents seem to be
trusting their (yet personally mostly unknown)
neighbors with personal details such as name,
phone numbers, e-mail addresses, photo, occupation, interests, hobbies, and so forth. In our survey,
the majority of residents indicated that they are
willing to share this kind of personal information
online with other residents in the building.
Nevertheless, issues of privacy and social
control have to be translated into appropriate
terms and conditions that govern the usage of
the system and the interaction among residents
of the building. It is imperative to ensure that
residents have the chance to opt in and opt out
at any time without missing out on any essential
information. Hence, it is worthwhile to consider
supplementing official online communication
channels with public announcements on neighborhood pinboards in prominent places within
the building (e.g., parking lot entry, reception or
entrance area, managers office door, elevators)
in order to provide alternative ways to access
community information.

Network of Networks, Identity,

and Sense of Belonging
The urban tribe incubator may resemble more the
networked nature of, for example, an online dating

site than the collective nature of, for example, an

online discussion board. What may emerge from
this process of personalized networking (or online
dating) is a complex web of social networks that
span the anonymous void of the building complex,
a web of urban tribes (Watters, 2003). Social hubs
will continue to play a crucial role as their bridging
links (Kavanaugh et al., 2003) connect different
social networks and establish connectivity in the
sense of community and solidarity. Drawing on
viral marketing strategies (Godin, 2001; Goldsmith, 2002), the incubator allows individuals
to cross-invite and introduce peers to the other
networks in which they participate, both inside
and outside the neighborhood. The feeling of a
neighborhood identity and a sense of belonging
can only emerge if bridging social links between
members of different urban tribes contributes to
the formation of a mesh-work of urban tribes that
is networked to the edge of chaos (Gilchrist,
2000, p. 264). In this context, identity and a sense
of belonging are not derived from the collective
feeling of being co-located in the same place but
from the feeling of being connected to a group of
friends who are part of a greater group of peers
living close by.

Conclusion and outlook

The design considerations presented here will
guide the development of the core prototype system. We then envision to extend this core with more
sophisticated features that, for example, allow
users to produce and to exchange creative content
(photos, audio, video) through sociocultural animation (Foth, 2006) and that simplify the tasks of
organizing and managing social gatherings, such
as calendaring, inviting, RSVPs, synchronizing
with SMS and e-mail, and so forth. As well, in
this environment, the social aspects of the urban
tribe incubator can be combined with managerial
features that allow apartment owners to interact
with the body corporate and tenants with the

599

Analyzing the Factors Influencing the Successful Design and Uptake of Interactive Systems

on-site management. In this role, the system can

manage rates and rent payments, entry notices,
mailings and notifications, personalized information on contractors, and house rules, thus adding
further value to the system and encouraging uptake
and usage. Cross-platform compatibility is key.
As such, the urban tribe incubator is anticipated
to be a technical framework that can be accessed
not only on the home or office computer but also
on mobile and other devices.
The future holds interesting outlooks for
platform developments. New urbanism, urban
renewal, and the move toward more and more
compact cities create opportunities to rethink the
communicative paradigm of apartment complexes
and vertical real estate as well as the sociological qualities of the office environment in which
most social software is accessed. The kitchen is
associated with the preparation of food, which is
an essential part of ones social life, as opposed
to the office, which is the center of professional
life. Hence, modern residential architecture often links the kitchen area with the living room
in order to form one seamless space that can be
re-purposed for entertainment and leisure. In
this context, the much-scorned Internet fridge
might see a revival as an integrated local communication hub that combines the functionality
of a simple touch-screen display interface, a
ubiquitous instant messenger, and a synchronized
resident buddy list with location-aware services
and groupware functionality. The rationale for
choosing the fridge is not based on the inherent
cooling functionality of the fridge itself, but its
position and prominence within the environment
of many urban homes.
The article contributes to substantiating a
new zeitgeist of designing residential community
networks for urban neighborhoods, which is characterized by combining current understandings
of social networks inherent in Wellmans theory
of networked individualism with the affordances
of ubiquitous communication devices and applications for personalized place-based networking

600

such as the Internet, instant messengers, and

mobile phones. Putnam (2000) argues that the
Internet will not automatically offset the decline
in more conventional forms of social capital,
but that it has that potential. In fact, it is hard to
imagine solving our contemporary civic dilemmas
without computer-mediated communication (p.
180). If online community networks for residential
communities are designed to include features that
cater to both collective and network interaction,
then they have the potential to contribute to the
creation of neighborhood identity and to increase
network capital and social capital in urban environments (Florida, 2003; Huysman & Wulf, 2004;
Quan-Haase et al., 2002; Watters, 2003). Thus,
they may prove to be a milestone in the quest to
animate urban neighborhoods, to revive forms of
civic engagement in society and to enact global
connectivity for local action, in order to move
from the vision of the global village to a new
understanding of the urban village.

References
Arnold, M. (2003). Intranets, community, and
social capital: The case of Williams Bay. Bulletin
of Science, Technology & Society, 23(2), 78-87.
Arnold, M., Gibbs, M. R., & Wright, P. (2003).
Intranets and local community: Yes, an intranet
is all very well, but do we still get free beer and
a barbeque? In M. Huysman, E. Wenger, & V.
Wulf (Eds.), Proceedings of the First International
Conference on Communities and Technologies
(pp. 185-204). Amsterdam, NL: Kluwer Academic
Publishers.
Barabsi, A.-L. (2003). Linked: How everything
is connected to everything else and what it means
for business, science, and everyday life. New
York: Plume.
Butler, B. S. (2001). Membership size, communication activity, and sustainability. Information
System Research, 12(4), 346-362.

Analyzing the Factors Influencing the Successful Design and Uptake of Interactive Systems

Carroll, J. M., & Rosson, M. B. (2003). A trajectory for community networks. The Information
Society, 19(5), 381-394.

Dunbar, R. I. M. (1996). Grooming, gossip, and the

evolution of language. Cambridge, MA: Harvard
University Press.

Castells, M. (2001). The Internet galaxy: Reflections on the Internet, business, and society.
Oxford: Oxford University Press.

eric_brissette. (2004, September 3). Personal Comment. Retrieved December 14,

2005, from http://slashdot.org/comments.pl?cid
=10147964&sid=120406

Castells, M. (2004). Space of flows, space of places:

Materials for a theory of urbanism in the information age. In S. Graham (Ed.), The cybercities
reader (pp. 82-93). London: Routledge.
Churchill, E., Girgensohn, A., Nelson, L., & Lee,
A. (2004). Blending digital and physical spaces
for ubiquitous community participation. Communications of the ACM, 47(2), 39-44.
Cohill, A. M., & Kavanaugh, A.L. (Eds.). (2000).
Community networks: Lessons from Blacksburg, Virginia (2nd ed.). Norwood, MA: Artech
House.
Day, P. (2002). Designing democratic community
networks: Involving communities through civil
participation. In M. Tanabe, P. van den Besselaar,
& T. Ishida (Eds.), Digital cities II: Second Kyoto
workshop on digital cities, LNCS Vol. 2362 (pp.
86-100). Heidelberg, Germany: Springer.
Day, P., & Schuler, D. (Eds.). (2004). Community
practice in the network society: Local action/
global interaction. London: Routledge.
De Cindio, F., Gentile, O., Grew, P., & Redolfi, D.
(2003). Community networks: Rules of behavior
and social structure. The Information Society,
19(5), 395-406.
de Tocqueville, A. (2000). Democracy in America
(H. C. Mansfield & D. Winthrop, Trans.). Chicago:
University of Chicago Press.
De Villiers, P. (1997). New urbanism: A critical
review. Australian Planner, 34(1), 30-34.
Dhanarajan, G. (2001). Distance education: Promise, performance and potential. Open Learning,
16(1), 61-68.

Etzioni, A. (1995). The spirit of community: Rights,

responsibilities, and the communitarian agenda.
London: Fontana Press.
Fallows, D. (2004). The Internet and daily life.
Washington, DC: Pew Internet & American Life
Project.
Florida, R. L. (2003). Cities and the creative class.
City and Community, 2(1), 3-19.
Foth, M. (2003). Connectivity does not ensure
community: On social capital, networks and communities of place. In S. Marshall, & W. Taylor
(Eds.), Proceedings of the 5th International Information Technology in Regional Areas (ITiRA)
Conference 2003 (pp. 31-39). Rockhampton, QLD:
Central Queensland University Press.
Foth, M. (2004a). Animating personalised networking in a student apartment complex through
participatory design. In A. Bond, A. Clement, F.
de Cindio, D. Schuler, & P. van den Besselaar
(Eds.), Proceedings of the Participatory Design
Conference, Toronto, Canada, July 27-31 (Vol.
2, pp. 175-178). Palo Alto, CA: CPSR.
Foth, M. (2004b, September 29-October 1). Designing networks for sustainable neighbourhoods:
A case study of a student apartment complex. In
G. Johanson, & L. Stillman (Eds.), Community
Informatics Research Network (CIRN) 2004 Colloquium and Conference Proceedings (Vol. 1, pp.
161-172), Prato, Italy.
Foth, M. (2006). Sociocultural animation. In S.
Marshall, W. Taylor, & X. Yu (Eds.), Encyclopedia of developing regional communities with

601

Analyzing the Factors Influencing the Successful Design and Uptake of Interactive Systems

information and communication technology (pp.

640-645). Hershey, PA: Idea Group Reference.

communications technologies. Hershey, PA: Idea

Group Publishing.

Foth, M., & Brereton, M. (2004, November 2024). Enabling local interaction and personalised
networking in residential communities through
action research and participatory design. In P.
Hyland, & L. Vrazalic (Eds.), Proceedings of
OZCHI 2004: Supporting Community Interaction.
Wollongong, NSW: University of Wollongong.

Gurstein, M. (2001). Community informatics,

community networks and strategies for flexible
networking. In L. Keeble, & B. D. Loader (Eds.),
Community informatics: Shaping computer-mediated social relations (pp. 263-283). New York:
Routledge.

Gilchrist, A. (2000). The well-connected community: Networking to the edge of chaos. Community Development Journal, 35(3), 264-275.

Hampton, K. N. (2003). Grieving for a lost network: Collective action in a wired suburb. The
Information Society, 19(5), 1-13.

Gilchrist, A. (2004). The well-connected community: A networking approach to community

development. Bristol, UK: Policy Press.

Hampton, K. N., & Wellman, B. (2003). Neighboring in netville: How the Internet supports
community and social capital in a wired suburb.
City and Community, 2(4), 277-311.

Gillespie, A., & Richardson, R. (2004). Teleworking and the city: Myths of workplace transcendence and travel reduction. In S. Graham (Ed.),
The cybercities reader (pp. 212-218). London:
Routledge.

Hearn, G., & Foth, M. (2005). Action research in

the design of new media and ICT systems. In K.
Kwansah-Aidoo (Ed.), Topical issues in communications and media research (pp. 79-94). New
York: Nova Science.

Gleeson, B. (2004). Deprogramming planning:

Collaboration and inclusion in new urban development. Urban Policy and Research, 22(3),
315-322.

Horan, T. A. (2000). Digital places: Building

our city of bits. Washington, DC: Urban Land
Institute (ULI).

Godin, S. (2001). Unleashing the ideavirus. New

York: Hyperion.
Goldsmith, R. (2002). Viral marketing: Get your
audience to do your marketing for you. London:
Pearson Education.
Greenbaum, J. M., & Kyng, M. (Eds.). (1991).
Design at work: Cooperative design of computer
systems. Hillsdale, NJ: Lawrence Erlbaum Associates.
Grosz, E. A. (2001). Architecture from the outside:
Essays on virtual and real space. Cambridge,
MA: MIT Press.
Gurstein, M. (Ed.). (2000). Community informatics: Enabling communities with information and

602

Horrigan, J. B. (2001). Cities online: Urban development and the Internet. Washington, DC: Pew
Internet & American Life Project.
Horrigan, J. B., Rainie, L., & Fox, S. (2001).
Online communities: Networks that nurture longdistance relationships and local ties. Washington,
DC: Pew Internet & American Life Project.
Huysman, M., & Wulf, V. (Eds.). (2004). Social
capital and information technology. Cambridge,
MA: MIT Press.
Jankowski, N. W., Van Selm, M., & Hollander, E.
(2001). On crafting a study of digital community
networks: Theoretical and methodological considerations. In L. Keeble, & B. D. Loader (Eds.),
Community informatics: Shaping computer-mediated social relations (pp. 101-117). New York:
Routledge.

Analyzing the Factors Influencing the Successful Design and Uptake of Interactive Systems

Kavanaugh, A. L., Reese, D. D., Carroll, J. M.,

& Rosson, M. B. (2003). Weak ties in networked
communities. In M. Huysman, E. Wenger, & V.
Wulf (Eds.), Proceedings of the First International
Conference on Communities and Technologies
(pp. 265-286). Amsterdam, NL: Kluwer Academic
Publishers.
Maloney-Krichmar, D., Abras, C., & Preece, J.
(2002, June 6-8). Revitalizing an online community. Paper presented at the International Symposium
on Technology and Society (ISTAS) Social
Implications of Information and Communication
Technology, Raleigh, North Carolina.
Meredyth, D., Ewing, S., & Thomas, J. (2004).
Neighbourhood renewal and government by community. International Journal of Cultural Policy,
10(1), 85-101.
Mitchell, W. J. (2003). Me++: The cyborg self
and the networked city. Cambridge, MA: MIT
Press.
Oldenburg, R. (2001). Celebrating the third place.
New York: Marlowe & Co.
Papadakis, M. C. (2004). Computer-mediated
communities: A bibliography on information,
communication, and computational technologies and communities of place (No. P10446.004).
Arlington, VA: SRI International.
Patterson, S. J., & Kavanaugh, A. L. (2001).
Building a sustainable community network: An
application of critical mass theory. The Electronic
Journal of Communication, 11(2).
Pinkett, R. D. (2003). Community technology
and community building: Early results from the
creating community connections project. The
Information Society, 19(5), 365-379.
Preece, J. (2000). Online communities: Designing usability, supporting sociability. Chichester:
John Wiley.

Putnam, R. D. (2000). Bowling alone: The collapse and revival of American community. New
York: Simon & Schuster.
Quan-Haase, A., Wellman, B., Witte, J. C., &
Hampton, K. N. (2002). Capitalizing on the Net:
Social contact, civic engagement, and sense of
community. In B. Wellman, & C. A. Haythornthwaite (Eds.), The Internet in everyday life (pp.
291-324). Oxford: Blackwell.
Rheingold, H. (2002). Smart mobs: The next social
revolution. Cambridge, MA: Perseus.
Satchell, C. (2003, November 26-28). The swarm:
Facilitating fluidity and control in young peoples
use of mobile phones. In S. A. Viller, & P. Wyeth
(Eds.), Proceedings of OZCHI 2003: New directions in interaction, information environments,
media and technology. Brisbane, QLD: University
of Queensland.
Schuler, D. (1996). New community networks:
Wired for change. New York: ACM Press.
Schuler, D., & Namioka, A. (Eds.). (1993). Participatory design: Principles and practices. Hillsdale,
NJ: Lawrence Erlbaum Associates.
Tolkien, J. R. R. (1966). The lord of the rings (2nd
ed.). London: Allen & Unwin.
Tnnies, F. (1887). Gemeinschaft und gesellschaft
(3rd ed.). Darmstadt, Germany: Wissenschaftliche
Buchgesellschaft.
Walmsley, D. J. (2000). Community, place and
cyberspace. Australian Geographer, 31(1), 5-19.
Watters, E. (2003). Urban tribes: Are friends the
new family? London: Bloomsbury.
Watts, D. J. (2003). Six degrees: The science of a
connected age. New York: Norton.
Wellman, B. (2001). Physical place and cyberplace:
The rise of personalized networking. International
Journal of Urban and Regional Research, 25(2),
227-252.

603

Analyzing the Factors Influencing the Successful Design and Uptake of Interactive Systems

Wellman, B. (2002). Little boxes, glocalization,

and networked individualism. In M. Tanabe, P.
van den Besselaar, & T. Ishida (Eds.), Digital
cities II: Second Kyoto Workshop on Digital
Cities, LNCS Vol. 2362 (pp. 10-25). Heidelberg,
Germany: Springer.

Wellman, B., Quan-Haase, A., Boase, J., Chen,

W., Hampton, K. N., Daz de Isla Gmez, I., et al.
(2003). The social affordances of the Internet for
networked individualism. Journal of ComputerMediated Communication, 8(3).

Wellman, B., & Haythornthwaite, C. A. (Eds.).

(2002). The Internet in everyday life. Oxford:
Blackwell.

This work was previously published in International Journal of Technology and Human Interaction 2(2), edited by B. Stahl, pp.
65-82, copyright 2006 by IGI Publishing, formerly known as Idea Group Publishing (an imprint of IGI Global).

604

605

Chapter 2.14

Different Levels of Information

Systems Designers Forms of
Thought and Potential for
Human-Centered Design
Hannakaisa Isomki
University of Lapland, Finland

ABSTRACT
This article describes a study clarifying information systems (IS) designers conceptions of human
users of IS by drawing on in-depth interviews
with 20 designers. The designers lived experiences in their work build up a continuum of
levels of thought from more limited conceptions
to more comprehensive ones reflecting variations
of the designers situated knowledge related to
human-centred design. The resulting forms of
thought indicate three different but associated
levels in conceptualising users. The separatist
form of thought provides designers predominantly
with technical perspectives and a capability for
objectifying things. The functional form of
thought focuses on external task information and
task productivity, nevertheless, with the help of
positive emotions. The holistic form of thought
provides designers with competence of human-

centred information systems development (ISD).

Furthermore, the author hopes that understanding
the IS designers tendencies to conceptualise human users facilitates the mutual communication
between users and designers.
Introduction
As information systems (IS) increasingly pervade
all aspects of everyday life, of utmost importance
is how applications of IS are adjusted to human
action. In particular, in current information
systems development (ISD) it is essential to take
into account human characteristics and behaviour; that is, to humanise IS (Sterling, 1974).
In the same vein, Checkland (1981) argues that
ISD should be seen as a form of enquiry within
which IS designers understandings regulate an
operationalisation of their intellectual framework
into a set of guidelines for investigation that require

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Different Levels of Information Systems Designers' Forms of Thought and Potential

particular methods and techniques for building

the system. Regarding the humanisation of IS, a
notion concerning the nature of the human being
is a crucial element of the intellectual framework.
As a consequence, within this kind of enquiry,
the way humans are taken into account in ISD
is dependent on the operationalisation of the
IS designers conceptualisations of users. With
respect to human-centeredness, attention should
be paid to the fundamental qualities of people
without any explicit or implicit domination of the
other elements of IS, such as data, formal models
and technical appliances, or managerial belief
systems that treat humans instrumentally. This
is necessary in order to conceptualise humans in
their own right, and thus avoid the reduction of
humans to something that exists only in relation
to particular instrumental needs and purposes
(cf. Buber, 1993).
Of essential importance is the nature of IS
designers insights into human characteristics
and behaviour that are essential with respect to
the IS-user relationship. The most crucial insight
regarding human-centred design is to be able to
conceptualise users as active subjects comprised of
physical, cognitive, emotional, social and cultural
qualities, an insight which is the prerequisite for
design that promotes subsequent user acceptance
and satisfaction. Yet conspicuously absent from
contemporary IS literature are empirical studies
investigating IS designers conceptions of the
human users, which have been studied more
intensively two decades ago when the systems
designers inadequate view of the user has been
stated to be one reason for the behavioural problems often experienced while implementing IS
(Bostrom & Heinen, 1977; Dagwell & Weber,
1983). Also, the lack of knowledge of human
needs and motivation on the part of the systems
designers has been claimed to cause IS implementation failures (Hawgood, Land & Mumford,
1978). Further, Hedberg and Mumford (1975) have
defined the nature of the view of human being
held by systems designers as an essential factor

606

in the IS design process. The systems designers

view of the user is also included in some studies
as one of the targets of value choices during the
ISD process (Kumar & Bjrn-Andersen, 1990;
Kumar & Welke, 1984) and is therefore defined
as a value rather than an insight in these studies.
Dagwell and Weber (1983), in their replication
study, rely on Hedberg-Mumfords definition of the
concept but also refer to Kling (1980). we know
very little about the perceptions that computer
specialists have of the users they serve and the
ways in which they translate these perceptions
into concrete designs (p. 47). Bostrom & Heinen
(1977), in turn, define systems designers assumptions of people as one of the system designers
implicit theories or frames of reference. These
previous works do not take an explicit stance
toward the definition of the concept conception,
and do not align the nature of conceptions in
detail. For instance, from where do conceptions
derive their origins, and what is the nature of those
conceptions? In a more recent study, Orlikowski
and Gash (1994) discuss their definition of the
IS designers views. They elaborate the concept
frame of reference by comparing it to the concept schema (Neisser, 1976, pp. 9-11), shared
cognitive structures or cognitive maps (Eden,
1992, pp. 261-262), frames (Goffman, 1974,
pp. 10-11), interpretative frames (Bartunek &
Moch, 1987, p. 484), thought worlds (Dougherty,
1992, p. 179), interpretative schemes (Giddens,
1984, pp. 29-30), scripts (Gioia, 1986, p. 50),
paradigms (Kuhn, 1970, p. 43), and mental
models (Argyris & Schn, 1978). They end up
by defining their own meaning for the concept
frames as a general concept of shared cognitive
structures, not especially regarding humans.
This article describes a study which aims to
clarify IS designers conceptions of users of IS
by drawing on in-depth interviews with 20 IS
designers. The analytical choices carried out in
this study regard IS designers conceptions of
users as experiences inherent in their lifeworlds,
particularly during the different phases of ISD.

Different Levels of Information Systems Designers' Forms of Thought and Potential

The lived experiences build up conceptions that

form a structure of meaning, which incorporates
a continuum of levels from more limited understandings to more comprehensive notions; that is,
different levels of thought reflecting variations of
the designers situated, practical knowledge. In
this way the development of IS is understood as
knowledge work. It is an intellectual and personal
process which takes its form according to the
conceptions of the performers of the process. IS
designers are then applying the ISD methodologies
according to their own observations and thinking
(Avison & Fitzgerald, 1994; Hirschheim et al,
1995; Mathiassen, 1998). Then the most important
tool for ISD, and a key resource in contemporary
IT companies, is the IS designers thought and
insight (Nonaka & Takeuchi, 1995; Quinn, 1992).
Particularly with respect to the humanisation of
IS, designers conceptualisations of the human
users are seen as knowledge that reflects the
designers competence in humanising IS. In this
way, IS designers conceptions may be seen as
intellectual capital that mirrors the know-how,
practices, and accumulated expertise of practitioners within a particular profession (Kogut &
Zander, 1992).
In what follows, first, the assumptions informing this study are presented by introducing the
interpretative approach referred to as phenomenography. Second, the resulted forms of thought
are presented and discussed. The IS designers

conceptualisations presenting their mental

schemes of the human user result in three forms
of thought, revealing both context-centred and
human-centred understandings of what humans
are. These three levels of understanding indicate
that IS designers tend to conceptualise uses in
terms of technology, business and work, and that
seldom are users taken into account according to
their human qualities. Finally, the different levels
of conceptualisation are discussed in relation to
human-centred ISD.
Different Levels of
Understanding
This study merges with the principles of phenomenography, which is a qualitatively oriented
method of empirical research for investigating the
different ways in which people experience aspects
of reality (Marton, 1981; Marton & Booth, 1997).
Essentially, phenomenography is about individual
meaning construction, which results in a conception. The primary focus is on the structure of the
meaning of conceptions, which are seen in the
light of the phenomenological notion according to which person and world are inextricably
related through a persons lived experience of
the world (e.g., Husserl, 1995; Merleau-Ponty,
1962). Our intentionality is seen as qualitatively
varying foci on the horizon of our life worlds.

Figure 1. The analytical distinctions of an experience (Marton & Booth, 1997)

EXPErIENcE
structural
aspect

External
horizon

referential
aspect

Internal
horizon

607

Different Levels of Information Systems Designers' Forms of Thought and Potential

While experiencing the world, individuals form

conceptions, including qualitative dissimilarities,
which are inherent in the intertwined referential
and structural aspects of an experience. Different levels are due to the way the structural aspect
and the referential aspect merge with each other.
Then an experience is specified by the analytical
distinctions of a structural aspect and a referential
aspect (Figure 1). The structural aspect denotes
how a particular phenomenon is both discerned
from its environment and how the phenomenons
parts relate to each other as well as to the whole
phenomenon. That which surrounds the phenomenon experienced, including its contours, is its
external horizon. The parts and their relationships,
together with the contours of the phenomenon, are
its internal horizon. The referential aspect signifies
the meaning of the conception. These two aspects
are dialectically intertwined and occur simultaneously within an experience. Thus, people create
conceptions with respect to the structural aspects
external and internal horizons of a phenomenon
that are dialectically merged with the referential
aspect of that particular phenomenon.
On the one hand, conceptions differ in terms
of content, and on the other hand, they differ
in terms of the extent of the form that a certain
phenomenon is experienced, as a part of that phenomenon, or more as a whole. When detached parts
of a phenomenon are the focus of thought instead
of relating the parts meaningfully to the whole
phenomenon, the meaning of the phenomenon is
understood in a way that refers to a more narrow
view. Respectively, when the focus of thought is
more comprehensive regarding the whole meaning
of a phenomenon instead of on separate parts of
it or even the surroundings of the phenomenon,
the more explanatory conceptions are. Further,
the more explanatory power conceptions have,
the better they support competent action with
respect to the phenomenon in question (Sandberg,
2000). Based on these differences, conceptions
form a structure of meaning, which incorporates
a continuum of levels from more limited under-

608

standings to more comprehensive notions. The

qualitative variation in the ways that IS designers conceptualise users reflects their different
conceptions of users, and simultaneously forms
different levels in the designers understandings,
reflecting variations of their situated, practical
knowledge as forms of thought. These forms of
thought, in turn, suggest different levels of competence in humanising IS, because the subjective
conceptualisations of IS designers refer to their
intention of action (Slj, 1994).
The interview method is accomplished with
respect to phenomenographic principles. To
promote multiple interpretations within individual designers, first, opening questions with
varying perspectives into the different phases
of ISD, such as planning, design, implementation, use and maintenance, was incorporated in
the interview framework. Second, to sustain the
connection between the designers reflection and
the actual work within ISD phases while the indepth interviews aimed at achieving mutual and
authentic understanding that proceeded from the
interviewees expressions. The interviews were
taped at the designers workplaces to maintain
the work practice orientation and to facilitate the
expression of the connection between the respondents immediate experiences and the subsequent
conception. Second, the respondents represent a
variety of geographical location, age, gender, educational background and work experience. They
came from various different application areas of
IS practice, such as e-commerce, tele and media
communications, groupware, health care systems,
office automation and insurance systems.
The analysis of the interview data was carried
out against the phenomenographical theory. Since
phenomenography does not offer grounds for
defining the content of conceptions within data, a
coding paradigm was developed from the data in
order to facilitate the identification and categorisation of meanings in the data (Glaser & Strauss,
1967; Tesch, 1990; Strauss & Corbin, 1990). The
development of the coding paradigm was based

Different Levels of Information Systems Designers' Forms of Thought and Potential

on the phenomenographical principle regarding

the intentionality of conceptions. Conceptions
are context-dependent and every experience is
described in content-loaded terminology (Slj,
1994,). That is, the descriptions are carried out in
terms of the nature of the situational experiences
in question. Consequently, the meanings are to be
found in accordance with the underlying assumptions concerning the intentional nature of ISD. In
addition, since it is assumed in phenomenography
that the meanings of the respondents mental acts
exist in the data and are constitutive of the data
(Walsh, 1994), the way meanings are understood
in this study should also be in accordance with the
types of intentionality existing in the data. Thus,
the coding paradigm was finalised by bracketing
away any preconceived ideas of what the IS designers views might be like (Francis, 1993), and
letting the analysis be informed by the underlying
assumptions of intentionality in ISD.
First, the data includes utterances that describe
various actions and objectives concerning ISD.
These expressions indicate intentionality as
defined by Hirschheim et al. (1995, p. 16). They
state that IS development is intentional, to the
extent it reflects a planned change. It is based on
developers intentions to change object systems
towards desirable ends, and go on to say that
(1995, p. 17) intentions in systems development
are expressed by objectives. These are related to
general value-orientations and represent what one
ought to do or what is good. From this it can be
concluded, in the first place, that intentionality in
ISD is expressed by intentional action. That is to
say, IS designers descriptions of the actions and
means they are involved with when developing an
IS reveal the meanings they give to the phenomena
they deal with concerning ISD. This notion is in
accordance with the principle of contextuality in
phenomenography, which denotes that peoples
conceptualisations are not detachable, either from
their context or the content of the task at hand.
This stance also reinforces the interpretative

nature of phenomenographical analysis in that

the researcher must see the designers action
as inherently meaningful (cf. Schwandt, 2000).
In the second place, as Hirschheim et al. (1995)
point out, intentions are expressed by objectives
of ISD. Consequently, it is an appropriate way to
define that the way the IS designers understand
the human user of an IS is revealed through
descriptions in which the respondents focus
of reflection is on the objectives of ISD. That is
to say, in addition to the actions and means the
designers refer to, the IS designers intentions
to change object systems toward desirable ends
reveal the meanings they give to the phenomena
they deal with concerning ISD. These desirable
ends or objectives represent the things that are
regarded most important in ISD. In this way, the
IS designers descriptions of action, means and
objectives also implicitly indicate value orientations included in the process of ISD. Therefore,
the described actions and objectives represent the
things that are regarded important, and thus reveal
the referential aspect in terms of intentionality as
an implied value orientation. This means that the
initial referential aspects of conceptions may be
found in utterances in which the designers refer
to their way and means of building systems and
the objectives of their actions.
Second, the data includes descriptions in which
the respondents thoughts are attached to human
objects. These descriptions of people indicated
human features and also value orientations toward
people. Often these descriptions also included
expressions which indicated emotionally toned
reactions. These kinds of expressions within the
data indicate intentionality that is in accordance
with Uljens (1991), who states that the process of
qualitative individuation of a mental act has been
done when an object and a psychological mode,
referred to as an attitude, is shown. In other words,
how a particular object of thought is experienced
denotes the respondents attitudes toward the phenomenon that is being reflected on. In brief, the
inherent meaning of an utterance may be seen as

609

Different Levels of Information Systems Designers' Forms of Thought and Potential

the correlation between the what- and how-aspects

in that they are not detachable from each other,
but are interrelated in a particular logical way,
indicating what a particular phenomenon is, in
what it is revealed, and what kind of values and
attitudes are related to it. As described above, the
search for the meanings in the data, data analysis,
was initiated by establishing a coding paradigm,
which suggests that the meanings in the data are
found in utterances in which the designers refer
to their actions, means, and objectives concerning
ISD, as well as to human characteristics.
The subsequent analysis procedures followed
the idea, firstly, of iterating between the meaning
(referential aspect) of single statements, their surrounding statements, and the data as a whole, and
second, iterating between the features that reveal
different levels in these meanings (structural aspect). The analysis appreciates the phenomenological notion of Lebenswelt in that the continuum
of levels from more limited forms of thought
to more comprehensive understandings reflects
associative connections within the designers
conceptions rather than direct, law-like logical
relations between them (cf. Husserl, 1995). The
analysis revealed the designers conceptions as
different levels of understanding, which appear
as the separatist, functional, and holistic forms of
thought. These forms of thought reveal three different levels of understandings incorporating both
context-centred and human-centred notions. In the
descriptions associated with the context-centred
conceptions, the designers focus of reflection is on
technology, work, and business. The human-centred conceptions deal with knowledge, emotions,
and the designers selves. The following description
delineates the specific contents of the conceptions
as parts of the three forms of thought by highlighting some exemplary conceptions.

610

The Separatist Form of

Thought
The most partial way in which the IS designers
conceptualise humans is through the separatist
form of thought. It demonstrates how IS designers
see humans within the affordances and constraints
of contemporary IS and their development as separated from fluid and coherent interactions. Within
this form of thought, the user is positioned outside
the IS designers awareness through objectivist
conceptualisations. An objectivist pattern is evident in the context-centred separatist conceptions,
which reflect understandings according to which
reality exists independent of humans and can thus
be understood independent of humans (cf. Lakoff,
1987; Orlikowski & Baroudi, 1991). In this case,
when discussing human-centred issues of ISD,
the designers focus of reflection is directed to
technology, job titles, and market mechanisms.
For example, within the conception of the human
being displaced by technology IS designers refer
to humans in terms of technology:
R:What are these requirements and wishes like?
Could you tell me more about them?
D16: Well, because it is a question of lets say
a feedback channel that our company offers as
a product to its clients, it means that if the client
purchases, for instance, a datanet-based customer
network, they have datanet and router accesses
through which they operate between their networks
and use the whole telecommunication network.
Then there are a lot of this kind of usability issues, response times and load percentages, or in
a way, how it (telecommunication network) sort
of behaves, what happens there.
In the above interview extract, the designer
considers the customers needs as a piece of software a feedback channel and the main
point that emerges is how this item of software
works with the functions of a telecommunications
network. The designers train of thought becomes

Different Levels of Information Systems Designers' Forms of Thought and Potential

focussed on technology instead of human-centred

issues and needs, such as how the software is built
in regard to the humans that will be using it. In
the same vein, within the conception of the human being as a market the designers make use
of expressions which show their intention is to
build products that are profitable and, therefore
easy to sell. Yet they do not base their intentions
upon human features, such as spontaneous and
mood-related online behaviour that could be a
prerequisite for selling their products (e.g., Hoffman & Novak, 1996; Bellman et al, 1999):
D5: It is more reasonable to develop a mass
product which has a lot of users. The point here
is that then it can be copied and sold.
In contrast to notions that emphasise understanding human consumption behaviour, the IS
designers adhere to the idea that humans are simply
a featureless mass of consumers who form a market
for IT products. Because the above conception
does not incorporate any human characteristics,
but refers to a mass market, it is thereby making a clear distinction between the market and
the features of the people assumed to form that
market. For this reason, the conception appears
as objectivist. This conception also implies a
predisposition according to which the current
development of IS as an industry is that of a rational institution which produces mass culture by
reducing humans to members of a mass (cf. Slater,
1997). Furthermore, humans become separated
attitudinally from IS and their development due
to a presumed lack of technological knowledge,
and thus are forced to encounter disparaging
attitudes. The separatist human-centred conception of the technology-illiterate human bein
produces accounts according to which the most
distinct characteristic of humans is that they are
ignorant of technology, specifically computers,
software and ISD methodologies. In particular,
this illiteracy is seen as a contrast to the literacy
of the IS designers:

R: Have you ever wondered why people behave in

that way that they cannot say what they want
from the system?
D17: I think that its because they dont know how
these [IS] are defined. If one doesnt know these
methods, one cant do it. That is the biggest reason,
not that they arent willing to say what they want
but they dont have the know- how.
Beath and Orlikowski (1994) report similar
findings in their analysis of a relatively new representative of the ISD methodologies rationalist
tradition, information engineering (IE). According
to the analysis, the IE text creates and sustains
both implicitly and explicitly a dichotomy between
users and IS designers by characterising the users as technologically ignorant in regard to the
use of technology. When operationalised, these
characterisations are likely to generate nonviable
and unsatisfactory interactions between users and
IS designers. It seems also that the designers do
not consider the weaknesses in users knowledge
and thought as an issue that deserves to be taken
into account in design. However, when humans are
included in the design considerations, the weaknesses in peoples thinking should be understood
as natural flaws in human cognitive behaviour that
can be appropriately guided, or even prevented
by adequate design (Kirs, Pflughoeft, & Kroeck,
2001; Norman, 1989; Robillard, 1999). The idea
then is that designers should have awareness of and
be able to recognise these erroneous tendencies in
users in order to carry out IS planning and design
with the express aim of preventing people from
committing faulty actions during computer use,
rather than conceptualising users as technologically ignorant.
Moreover, negative emotions and physical
stress symptoms have the effect of separating
humans from IS. The conception of the computer
anguished human being reveals views acknowledging that IS cause negative emotional arousal
in users. These reactions are manifested as negative attitudes, resistance, fear and discomfort in

611

Different Levels of Information Systems Designers' Forms of Thought and Potential

situations where people are confronted by plans

for the future use of computers or in situations in
which individuals are using computers:
R: How in your mind do people learn to use
software?
D6: I have also met users who have so much
fear of the user interface that they dont dare
to explore or try anything, they just do what is
familiar and safe.
These conceptualisations are consistent with
statements concerning the widespread existence
of technophobia (Brosnan, 1998a). Besides being
an obviously unpleasant and undesired experience,
negative emotions, such as anxiety and fear, make
peoples behaviour withdrawn and elusive by
narrowing their action (Fredrickson & Branigan,
2001), as well as decreasing the quality of their
performance (Brosnan, 1998b).
In brief, within the separatist form of thought,
the human being becomes separated from viable
interactions with both the IS designers themselves
and IS. This is due to a tendency to an objectivist conceptualisation, which blurs the designers
thought to such an extent that people are no longer
recognised as humans. The overall narrative style
of this form of thought is reminiscent of the style
of a nomothetic science reflecting technical, strategic views aiming at controlling the IS-related
social system with the technical system (cf. Deetz,
1996). Further, people become separated from the
development of IS due to disparaging attitudes
inherent in designers assumptions that users are
technologically ignorant. Moreover, this form of
thought brings to the fore human characteristics,
such as negative emotions, which are seen as an
obstacle to a viable IS-user relationship.

The Functional Form of

Thought
The functional form of thought consists of conceptualisations in which humans are seen to act
in an insubstantial manner, adapting to the external functions of technology, work tasks and the
way the IS designers themselves use computers.
Within this adaptive response, positive emotions
are required in order to create and sustain viable
interactions with IS. In this way, the IS-user
relationship is seen as functional: the action of
people is seen as determined by their external
environment, and the role of human emotion is to
facilitate this process. The different conceptions
that build up this form of thought reveal a behaviourist understanding of the human being.
For instance, the conception of the invisible
human being denotes humans as using IS in an
insubstantial manner. Typical of this conception
is the belief that there is a user who uses an IS.
Yet the user is not characterised further but is
assumed just to use the system:
R: If you think of a situation where you are creating an application, who do you think youre
making it to?
D16: Hm..
R: Do you think of certain types of people or how
does it show that you are making it for people?
D16: I dont think of particular types of people
but I think that the human being is in some sense
always a part of the system. If it is a system that
has a user interface so there must be somebody
who uses it. Even if it is a system that runs by
timer initiation, there must be a user interface,
too, for setting the timer parameters in the system,
so there must be somebody to use it, too. To my
mind there is always someone using the systems,
they (systems) are not fully automated.
A functioning relation between people and IS
is thus acknowledged, but this does not include
any features originating from the mental, social

612

Different Levels of Information Systems Designers' Forms of Thought and Potential

or cultural human modes of being. In other words,

humans and their behaviour are understood as
purely physical-organic responses to technology, as established in the tenets of Skinnerian
behaviourism (Skinner, 1938, 1991). Similarly,
the IS designers conception of the human being
behind the process of work denotes humans in
conformity with behaviourist thinking. Within
this conception individuals are seen in terms of
their work tasks or organisational work processes.
Characteristic of these conceptualisations is that
the people performing the tasks are not portrayed
further, but are assumed merely to use IS according to the external task flows:
R: How would you define users needs?
D8: They consist of the utilising organisations
needs at all levels, beginning with what the people
need in order to continually do their work tasks,
and ending with the things that the organisation
expects from the system, what can be abstracted
from the process and be used to develop and
control action.
Here, human action is seen as a series of
direct responses to external work tasks issued to
people. Zuboffs (1988) well-known distinction
between automating work and informating
work highlights the difference between implied
behaviourist and nonbehaviourist assumptions
concerning human action in computerised work
(pp. 9-10). Automating work refers to deploying
technology in ways that increase the self-acting
and self-regulating capacities of technical systems, which are expected to minimise human
intervention. Because human intervention is
minimised and machines perform the work tasks,
interactions between individuals and computers
become determined by the structure and sequence
of computerised workflows to which, in turn,
humans are supposed to respond. Zuboffs term
of automating work, thus, implies a behaviourist
assumption of humans and their behaviour.

Quite the opposite is suggested by the term

informating work, which adds to the automating
view of work in that information technology can
be used to automate, but at the same time, it has
the ability to translate the automated activities
into a form that renders work processes, objects,
events and behaviours visible, knowable and
sharable for people (Zuboff, 1988). That is to say,
within the interaction of humans and computers,
people actively observe, interpret and share the
information which is mediated to them by IS.
They do not just respond like marionettes to the
information offered by IS, but actively construct
their own conceptions of the computer-mediated
tasks they are given and act according to their
own interpretations of the particular situation.
Thus, in order to accomplish fluid and coherent
interaction designs between humans and computers in regard to particular tasks, the users
mental models, especially those concerning the
tasks submitted to them, should also be designed
(Norman, 1989; Preece, 1994).
Also typical of the functionalist form of
thought is that the role of human emotion is
to facilitate peoples adaptation to technology.
Within the conception of The techno-enthusiast
human being the designers depict positive emotions, such as enthusiasm, as essential features
in humans:
R: Do you think there are common features in those
people for whom you have built systems?
D17: Well, at least during very recent years, it
has been enthusiasm.
In particular, positive emotional reactions
in people are seen to be induced by technology. Positive feelings are especially seen as a
prerequisite for the successful use of IS. These
conceptualisations reveal a functional understanding of positive emotions. Whereas negative
emotions are associated with specific tendencies,
such as an urge to escape or to avoid disquieting
things, positive emotions seem to spark changes

613

Different Levels of Information Systems Designers' Forms of Thought and Potential

in cognitive activity in addition to producing

behavioural tendencies (Fredrickson & Branigan,
2001). Therefore, the IS designers accounts of
positive emotions as a prerequisite for the use of
computers imply an understanding of the role of
human emotional features in promoting successful functioning.
To sum up, within the functional form of
thought humans and their behaviour are understood from a behaviourist stance, which renders
human substance only as physical and organic by
nature, denoting that the movements of people can
be explained by the laws of mechanics (Wilenius,
1978). However, this form of thought adds to the
previous separatist way of thinking in so far as
humans are actually depicted as performing tasks
with computers, whereas in the separatist form of
thought the conceptualisations either totally omit
human features or humans are seen as unable to
use computers. In addition, the human emotional
feature that is recognised in this form of thought
appears as positive even though functional
in nature. This way of thinking acknowledges
humans as users, and therefore is more developed
than the previous separatist form of thought.

The Holistic Form of Thought

The most developed form of thought by which the
IS designers conceptualise humans as users of IS is
the one characterised as holistic. Its holistic quality is revealed in several ways. First, unlike the
preceding forms of thought, the designers recognise a number of human characteristics. Second,
these observed human features are often seen to
coexist or intertwine with each other. Third, these
conceptualisations suggest that the relationship
between users and designers, as well as the ISuser relation, is a reciprocal process, including
characteristics typical of human behaviour.
To begin with, the conception of the human
being reflected in technology reveals the specific

614

goal of constructing computer interfaces with human-like features: the interaction between people
and computers is then envisaged as enriched with
dialogues conveying both the rational and emotional meanings of the information in question
(e.g., Nakazawa, Mukai, Watanuki & Miyoshi,
2001). Respectively, the depictions of various human features in technology reveal understandings
suggesting human features built into technology
render the interaction between users and IS as
resembling the interplay of cognitive, emotional
and social aspects that occur between humans:
R: What kind of user interface do you think that
people would want to use?
D4: I strongly believe that 3D interfaces are coming. They could offer kind of human-like facial
features as agents, which would bring a human
sense to the systems. The third dimension could
also be utilised so that interfaces become tangible
and accessible.
Further, the context-centred conception of
the human being as an organisational learner,
which highlights people as organisations which
learn about their own work processes, refers indirectly to learning, which stresses both cognitive
and social human features. Collective cognitive
features are referred to as an organisations ability
to form new insights into its work processes and
to guide the deployment of IS effectively (Robey,
Boudreau & Rose, 2000). A social dimension is
also implied when it is assumed that people learn
as an organisation:
D8: Needs are prone to change rapidly, especially
after the implementation of the system, because
they teach an organisation a lot about itself, and
an organisations self-knowledge increases and
usually needs change in a more clever direction.
Then there very quickly happens a sort of learning leap, which is often experienced as if the
system is not valid at all although it is a question

Different Levels of Information Systems Designers' Forms of Thought and Potential

of the organisations increased knowledge of its

own activity.
Within the conception of the knowledge
sharing human being the designers open up
their view of learning by specifying mutual
understanding between users and designers as
essential. It is important to be able to explain
things so that we understand each other. The
capability of taking anothers perspectives into
account form the core of this conception, which
highlights knowledge sharing as a particularly
important instance within the processes of organisational learning. Knowledge sharing is the
link between individual and group learning, and
signifies the expansion of individuals cognitive maps into shared understandings (Crossan,
Lane & White, 1999). In particular, the ability
to take the perspective of others into account
is an indispensable prerequisite for knowledge
sharing (Boland & Tenkasi, 1995). Buber (1993)
ascertains that, in order to be able to take others
perspectives into account fully, one has to treat
others as equal human beings and respect the
current circumstances of others. In these kinds
of relationships positive emothional features,
such as care and joy, need to be acknowledged
and combined with cognitive and social abilities
(Fredrickson & Branigan, 2001)
Moreover, the conception of the emotionally
coping human being refers to an ability to regulate in a successful way both negative and positive
subjective feelings in computerised situations.
In this way, the designers see emotional coping
in the light of positive outcomes (cf. Folkman &
Moskowitz, 2000, pp.648-649):
D8: a skilful user always has such peace of
mind and attitude. She or he kind of has a better
tolerance for stress, and an ability to cope with
contradictions in a better way than others. For
some reason this kind of attitude leads to a particular resourcefulness and an ability to utilise
the system in a more natural way, compared to a

person who has some negative emotional features,

fear or hostility towards the system, and who then
ends up having difficulties with the system due to
her/his heavy attitude.
A cognitive aspect is seen as inherent in emotional coping in that it requires that individuals
recognise their different emotional experiences.
However, in addition to these internal cognitiveaffective features, emotion regulation refers to the
external social and cultural factors that redirect,
control, and shape emotional arousal in such a
way that an individual is able to act adaptively
in emotionally activating situations (Pulkkinen,
1996). While ISD is often seen as a stressful
process which requires an ability to endure changing emotional experiences, such as interest and
frustration (Newman & Noble, 1990) in recurrent situations of failure and subsequent success
(Robey & Newman, 1996), it is understandable
that the designers regard people who are able to
regulate their emotions successfully as skilful.
Briefly, the holistic form of thought is comprised of conceptualisations that regard humans as
cognitive, emotional, social and cultural creatures.
The conceptions belonging to this form of thought
embody similar basic human modes of being, as
shown above. However, the aforementioned basic
modes of being emerge in these conceptions as
different behavioural affordances. The cognitive
mode of being is seen as intellect, reasoning,
learning, reflection, understanding and awareness
of something. Similarly, the emotional mode of
being is conceptualised as empathy, stress, tranquillity, commitment, contentment and a feeling
of mastery. Further, the social mode of being is
referred to as a need for communication, group
learning, interpersonal power and connection, as
well as knowledge sharing. These behavioural
affordances are seen as incorporated in technology, appearing between humans, or within the
interaction of humans and IS.

615

Different Levels of Information Systems Designers' Forms of Thought and Potential

The Forms of Thought

in ISD
The IS designers forms of thought revealed in
the results of this study are regarded as important
tools for ISD, and are seen to have implications
for the ways that humans are taken into account
as users within the different situations of ISD.
These different situations refer to the phases of
ISD such as planning, design, implementation, use
and maintenance. The phases are cyclical and intertwining (e.g., Beynon-Davies, Carne, Mackay,
& Tudhope, 1999). Planning refers to initiation
and requirements analysis actions, including client
contacts and definition of user requirements. During this phase the greatest degree of interaction
occurs between users and designers (Newman &
Noble, 1990). In order to accomplish requirements
analysis, the designers should understand many
human issues in addition to technical ones (Holtzblatt & Beyer, 1995). Design denotes procedures
where requirements are refined and turned into
specifications and finally software. Then technical
reliability and maintainability of the system, user
interfaces applicability for the intended purpose
of the system, as well as the aesthetical appearance of the system, are designed (Smith, 1997).
Winograd (1995) emphasises that, in addition to
technical requirements, the properties of a user
interface should meet with the social, cognitive
and aesthetic needs of people. Especially within
new ubiquitous technological environments, the
design of IS-user relationship should focus, in
addition to social and cultural features, on individuals perceptual, cognitive and emotional
space (Stephanidis, 2001).
How would the designers then perform according to their forms of thought? The strength of the
designers utilising a separatist form of thought
would be technical knowledge, especially the
ability to fluently conceptualise issues of design
in accordance with objective definitions, a skill
that is needed in creating formal specifications.
However, the validity of objectifying design issues

616

is dependent on the focus of such definitions. From

a human-centred perspective, valid definitions
would require being theoretically sensitive to
human activity and deriving second-order conceptions from that activity (see Walsham, 1995),
rather than creating objectivist conceptualisations,
which overlook humans and their behaviour. An
obvious disutility would be a tendency to treat
users as technologically ignorant, which implies
incompetence in social relationships with users.
The designers embracing the functional form
of thought possess technical knowledge, and value
such knowledge in users. They tend to focus on
formal job descriptions, external work tasks and
individuals task productivity. A deficit from a
human-centred perspective would be the tendency
to overlook human issues and to focus instead on
the functional purposes of IS; that is, external task
information regarding an organizations process
improvements. Often such conceptualisations
are regarded to yield Tayloristic designs, which
underestimate the social context. However, they
possess competence in functional and technical
systems design.Their strength would be increased
social competence to fulfil the demand for mutual understanding, which is regarded of utmost
importance in ISD (cf. Heng, Traut, & Fischer,
1999).
The designers building upon the holistic
form of thought emphasise clients satisfaction,
which ensures sustainable customer relationships
and regard mutual understanding during ISD
as essential between users and designers. Their
strength would be increased social competence
fulfil the demand for mutual understanding,
which is regarded of utmost importance in ISD
(e.g., Klein & Hirschheim, 1993; Lyytinen &
Ngwenyama, 1992). It seems also likely that they
have competence in IS planning which aims at
the improvement of organisational processes and
are identified as functional, such as sales and purchasing processes, and emphasise mutual understanding. Also, they understand how to maintain
customership instead of just visioning economic

Different Levels of Information Systems Designers' Forms of Thought and Potential

gains or focusing on peoples task productivity.

Besides possessing technical competence, these
holistic designers would be able to consolidate
definitions of formal and external work tasks
into human issues. A particularly significant
capability would be to understand the process
of organisational learning, which is essential in
order to adjust the evolving requirements during
the process of ISD. Moreover, they value balanced
emotional behaviour, and thus intuitively grasp
the possible dangers of relying on superfluous
emotional behaviour.
With respect to the humanisation of IS, a
holistic conception is required in ISD. It is then
assumed that the human being is actualised in
intertwined physical, cognitive, emotional, social
and cultural qualities, and that these qualities
are fundamentally different. Without the simultaneous existence of all of the qualities, it is not
possible to consider a creature as a human. Yet
the qualities cannot be reduced from one quality
to another, but rather need to be understood as a
whole (Rauhala, 1983). Considering the human
being as an actor, as a user of an IS, the whole
of a human being is understood as an active
subject adjoining to IS. Then the IS-user relationship consists of human action involving explicit
and tacit affordances that emerge dynamically
in the interaction between humans and IS. In
other words, the static characteristics of humans
and technology take on a new form within their
intertwining activity, which is shaped according to the affordances that, on the one hand, the
human substance embodies, and which, on the
other hand, the properties of IS support or ignore.
Consequently, understanding humans and their
behaviour as users of IS requires insight into these
emerging human experiences appearing within the
affordances and constraints of contemporary IS
and their development. Especially at present when
the IS are no longer merely
tools for personal and
professional instrumental productivity, but also
(re)constituting and mediating different social
structures and practices (e.g., Orlikowski, 1992;

Orlikowski, 2000), IS acts as social spaces that are

important growing social and cultural reference
points for users and, thus also for IS designers.
Design that takes into account the consequences
of the form and functions of IS to usersl social
qualities, such as self-identity, is indeed and
necessity if contemporary IS development aims
at high-quality and usable systems (Greenhill &
Isomki, 2005).
In summary, the resulting forms of thought
indicate three different but associated levels
of intellectual competence in conceptualising
humans as users of IS. The separatist form of
thought provides designers predominantly with
technical perspectives and a capability for objectifying things. However, it is worth noticing
that the validity of objectifying design issues is
dependent on the focus of such definitions. From
a human-centred perspective, valid definitions
would require being theoretically sensitive to human activity and deriving abstracted conceptions
from that activity rather than creating objectivist
conceptualisations, which overlook humans and
their behaviour. The functional form of thought
focuses on external task information and task
productivity, nevertheless, with the help of positive
emotions. The holistic form of thought provides
designers with competence of human-centred
ISD, even though all the aspects of the richness
of the human condition are not revealed. It seems
the designers are intellectually more oriented
toward designing IS for objectified, streamlined
organisational processes consisting of external
work tasks, and that this orientation challenges
the human-centred orientations.

References
Avison, D.E., & Fitzgerald, G. (1994). Information
systems development. In W. Currie & R. Galliers (Eds.), Rethinking management information
systems: An interdisciplinary perspective (pp.
250-278). Oxford: Oxford University Press.

617

Different Levels of Information Systems Designers' Forms of Thought and Potential

Beath, C.M., & Orlikowski, W. (1994). The

contradictory structure of systems development
methodologies: Deconstructing the IS-user relationship in information engineering. Information
Systems Research, 5(4), 350-377.
Bellman, S., Lohse, G.L., & Jordan, E.J. (1999).
Predictors of online buying behavior. Communications of the ACM, 42(12), 32-38.
Beynon-Davies, P., Carne, C., Mackay, H., & Tudhope, D. (1999). Rapid application development
(RAD): An empirical review. European Journal
of Information Systems, 8, 211-223.
Boland, R.J., & Tenkasi, R.V. (1995).
Perspective
making and perspective taking in communities of
knowing. Organization Science, 6(4), 350-372.
Bostrom, R.P., & Heinen, J.S. (1977). MIS

problems and failures: A socio-technical perspective.

Part I: The causes. MIS Quarterly, 1(3), 17-32.
Brosnan, M. (1998a). Technophobia. The psychological impact of information technology.
London: Routledge.
Brosnan, M. (1998b). The impact of computer anxiety and self-efficacy upon performance. Journal
of Computer Assisted Learning, 14, 223-234.
Buber, M. (1993). Sin ja min [I and Thou].
Juva: WSOY.
Checkland, P. (1981). Systems thinking, systems
practice. Chichester: Wiley.
Cotterman, W.W., & Kumar, K. (1989). User cube:
A taxonomy of end users. Communications of the
ACM, 32(11), 1313-1320.
Crossan, M.M., Lane, H.W., & White, R.E. (1999).
An organizational learning framework: From
intuition to institution. Academy of Management
Review, 24(3), 522-537.
Dagwell, R., & Weber, R. (1983). System designers user models: A comparative study and

618

methodological critique. Communications of the

ACM, 26(11), 987-997.
Deetz, S. (1996). Describing differences in approaches to organization science: Rethinking
Burrell and Morgan and their legacy. Organization
Science, 7(2), 191-207.
Folkman, S., & Moskowitz, J.T. (2000). Positive
affect and the other side of coping. American
Psychologist, 55(6), 647-654.
Francis, H. (1993). Advancing phenomenography: Questions of method. Nordisk Pedagogik,
13, 68-75.
Fredrickson, B.L., & Branigan, C. (2001). Positive emotions. In T. Mayne & G. Bonanno (Eds.),
Emotions: Current issues and future directions
(pp. 123-151). New York: Guilford Press.
Glaser, B.G., & Strauss, A.L. (1967). The discovery of grounded theory. Strategies for qualitative
research. London: Weidenfeld and Nicolson.
Greenhill, A., & Isomki, H. (2005).
Incorporating
self into Web information system design.
In A.
Pirhonen, H. Isomki, C. Roast, & P. Saariluoma
(Eds.), Future interaction design (pp. 52-66).
London: Springer-Verlag.
Hawgood, L., Land, F., & Mumford, E. (1978). A
participative approach to forward planning and
systems change. In G. Bracchi, & P.C. Lockermann (Eds.), Information systems methodology.
Proceedings of the 2nd Conference of the European
Cooperation in Informatics (pp. 39-61), Venice,
Italy. Springer-Verlag.
Hedberg, B., & Mumford, E. (1975). The design
of computer systems: Mans vision of man as an
integral part of the system design process. In E.
Mumford, & H. Sackman (Eds.), Human choice
Amsterdam: North
and computers (pp. 31-59).
Holland.
Heng, M.S.H., Traut, E.M., & Fischer, S.J. (1999).
Organisational champions of IT innovation. Ac-

Different Levels of Information Systems Designers' Forms of Thought and Potential

counting, Management and Information Technology, 9(3), 193-222.

Hirschheim, R., & Klein, H.K. (1989). Four paradigms of information systems development. Communications of the ACM, 32(10), 1199-1216.
Hirschheim, R., Klein, H.K., & Lyytinen, K.
(1995). Information systems development and
data modeling. Conceptual and philosophical
foundations. Cambridge University Press.
Hoffman, D.L., & Novak, T.P. (1996). Marketing
in hypermedia computer-mediated environments:
conceptual foundations. Journal of Marketing,
60(3), 50-68.
Holtzblatt, K., & Beyer, H.R. (1995). Requirements
gathering: The human factor. Communications of
the ACM 38(5), 31-32.
Husserl, E. (1995). Fenomenologian idea. Viisi
luentoa [The phenomenological idea. Five lectures] (
Himanka, Hmlinen & Sivenius, Trans.).
Helsinki: Loki-kirjat.
Kirs, P.J., Pflughoeft, K., & Kroeck, G. (2001).
A process model cognitive biasing effects in
information systems development and usage.
Information & Management, 38, 153-165.
Klein, H.K., & Hirschheim, R.A. (1993). The

application of neo-humanist principles in information systems development. In D.E. Avison, T.E.

Kendall, & J.J. DeGross (Eds.), Human, organizational, and social dimensions of information
systems development (pp. 263-280). Amsterdam:
Elsevier.
Kling, R. (1977). The Organizational Context of
User-Centered Software Designs. MIS Quarterly
1(4), 41-52.
Kogut, B., & Zander, U. (1992). Knowledge of
the firm, combinative capabilities and the replication of technology. Organization Science,
3(5), 383-397.

Kumar, K., & Bjrn-Andersen, N. (1990).

A
cross-cultural comparison of IS designer values.
Communications of the ACM, 33(5), 528-538.
Kumar, K., & Welke, J. (1984). Implementation
failure and system developer values: Assumptions,
truisms and empirical evidence. In Proceedings of
the 5th International Conference on Information
Systems (pp. 1-12), Tucson, AZ.
Lakoff, G. (1987). Women, fire and dangerous
things. University of Chicago Press.
Marton, F. (1981). Phenomenography: Describing
conceptions of the world around us. Instructional
Science, 10, 177-200.
Marton, F., & Booth, S. (1997). Learning and
awareness. Mahwah, NJ: Lawrence Erlbaum.
Mathiassen, L. 1998. Reflective systems development. Scandinavian Journal of Information
Systems, 10(1/2), 67-118.
Merleau-Ponty, M. (1962). Phenomenology of
perception. London: Routledge.
Nakazawa, M., Mukai, T., Watanuki, K., & Miyoshi, H. (2001). Anthropomorphic

agent and multimodal interface for nonverbal communication.

In N. Avouris, & N. Fakotakis (Eds.), Advances in
human-computer interaction I. Proceedings of the
PC HCI 2001 (pp. 360-365), Athens, Greece.
Newman, M., & Noble, F.(1990). User involvement
as an interaction process: A case study. Information Systems Research, 1(1), 89-110.
Nonaka, I., & Takeuchi, H. (1995). The knowledge-creating company: How Japanese companies create the dynamics of innovation. Oxford
University Press.
Norman, D.A. (1989). Miten avata mahdottomia
ovia? Tuotesuunnittelun

salakarit [The psychology of everyday things]. Jyvskyl:

Gummerus.
Orlikowski, W.J. (1992). The duality of technology: Rethinking the concept of technology

619

Different Levels of Information Systems Designers' Forms of Thought and Potential

in organizations. Organization Science, 3(3),

398-427.
Orlikowski W.J. (2000). Using technology and
constituting structures: A practice lens for studying technology in organizations. Organization
Science, 11(4), 404-428.
Orlikowski, W.J., & Baroudi, J.J. (1991). Studying

information technology in organizations: Research approaches and assumptions. Information

Systems Research, 2(1), 1-28.
Orlikowski, W.J., & Gash, D.C. (1994). Technological frames: Making sense of information
technology in organizations. ACM Transactions
on Information Systems, 12(2), 174-207.
Preece, J. (1994). Human-computer interaction.
Harlow, UK: Addison-Wesley.
Pulkkinen, L. (1996). Female and male personality
styles: A typological and developmental analysis.
Journal of Personality and Social Psychology,
70(6), 1288-1306.
Quinn, J.B. (1992). The intelligent enterprise: A
new paradigm. Academy of Management Executive, 6(4), 48-63.
Rauhala, L. (1983). Ihmisksitys ihmistyss [The
conception of the human being in human work].
Helsinki: Gaudeamus.
Robey, D., Boudreau, M.C., & Rose, G.M. (2000).
Information technology and organizational
learning: A review and assessment of research.
Accounting, Management & Information Technology, 10(1), 125-155.
Robey, D., & Newman, M. (1996). Sequential
patterns in information systems development: An
application of a social process model. ACM Transactions of information systems, 14(1), 30-63.
Robillard, P.N. (1999). The role of knowledge in
software development. Communications of the
ACM, 42(1), 87-92.

620

Slj, R. (1994). Minding action. Conceiving

the world vs. participating in cultural practices.
Nordisk Pedagogik, 14, 71-80.
Sandberg, J., (2000). Understanding human
competence at work: An interprettive approach,
Academy of Management Journal 43(1), 9-25.
Schwandt, T., (2000). Three epistemological
stances for qualitative inquiry. Interpretivism,
hermeneutics, and social constructionism. In
Denzin, N.K. &Y.S. Lincoln (Eds.) The handbook
of qualitative research (2nd ed.). Thousand Oaks,
CA: Sage, 189-213.
Skinner, B.F. (1991). The behavior of organisms:
An experimental analysis. Acton, MA: Copley.
Originally published in 1938.
Slater, D. (1997). Consumer culture and modernity. Malden, MA: Blackwell.
Smith, A. (1997). Human computer factors: A
study of users and information systems. London:
McGraw-Hill.
Stephanidis, C. (2001). Human-computer interaction in the age of the disappearing computer. In
N. Avouris, & N. Fakotakis (Eds.), Advances in
human-computer interaction I. Proceedings of the
PC HCI 2001 (pp. 15-22), Athens, Greece.
Sterling, T.D. (1974). Guidelines for humanizing
computerized information systems: A report from
Stanley House. Communications of the ACM,
17(11), 609-613.
Strauss, A., & Corbin, J. (1990). Basics of qualitative research: Grounded theory procedures and
techniques. Newbury Park, CA: Sage Publications.
Tesch, R. (1990). Qualitative research: Analysis
types and software tools. New York: Falmer
Press.
Uljens, M. (1991). Phenomenography: A qualitative approach in educational research. In L.
Syrjl, & J. Merenheimo (Eds.), Kasvatustut-

Different Levels of Information Systems Designers' Forms of Thought and Potential

kimuksen laadullisia lhestymistapoja. Oulun

yliopiston kasvatustieteiden tiedekunnan opetusmonisteita ja selosteita 39 (pp. 80107).
Walsh, E. (1994). Phenomenographic analysis
of interview transcripts. In J.A. Bowden, & E.
Walsh (Eds.), Phenomenographic research:
Variations in method (pp. 17-30). The Warburton
Symposium. Melbourne: The Royal Melbourne
Institute of Technology.
Walsham, G. (1995). Interpretive case studies in IS
research: Nature and method. European Journal
of Information Systems, 4(2), 74-81.
Wilenius, R. (1978). Ihminen, luonto ja tekniikka
[The human being, nature and technology]. Jy
vskyl: Gummerus.
Winograd, T. (1995). From programming environments to environments for designing. Communications of ACM, 38(6), 65-74.
Zuboff, S. (1988). In the age of the smart machine:
The future of work and power. New York: Basic
Books.

This work was previously published in International Journal of Technology and Human Interaction 3(1), edited by B. Stahl, pp.
30-48, copyright 2007 by IGI Publishing, formerly known as Idea Group Publishing (an imprint of IGI Global).

621

622

Chapter 2.15

The Think Aloud Method and

User Interface Design
M.W.M. Jaspers
University of Amsterdam, The Netherlands

INTRODUCTION
Daily use of computer systems often has been
hampered by poorly designed user interfaces.
Since the functionality of a computer system is
made available through its user interface, its design has a huge influence on the usability of these
systems (Carroll, 2002; Preece, 2002). From the
users perspective, the user interface is the only
visible and, hence, most important part of the
computer system; thus, it receives high priority
in designing computer systems.
A plea for human-oriented design in which the
potentials of computer systems are tuned to the
intended user in the context of their utilization
has been made (Rossen & Carroll, 2002).
An analysis of the strategies that humans use
in performing tasks that are to be computer-supported is a key issue in human-oriented design
of user interfaces. Good interface design thus
requires a deep understanding of how humans
perform a task that finally will be computersupported. These insights then may be used to
design a user interface that directly refers to their
information processing activities. A variety of

methodologies and techniques can be applied to

analyze end users information processing activities in the context of a specific task environment
among user-centered design methodologies. More
specifically, cognitive engineering techniques are
promoted to improve computer systems usability
(Gerhardt-Powels, 1996; Stary & Peschl, 1998).
Cognitive engineering as a field aims at understanding the fundamental principles behind
human activities that are relevant in the context
of designing a system that supports these activities (Stary & Peschl, 1998). The ultimate goal is
to develop end versions of computer systems that
support users of these systems to the maximum in
performing tasks in such a way that the intended
tasks can be accomplished with minimal cognitive
effort. Empirical research has indeed shown that
cognitively engineered interfaces are considered
superior by users in terms of supporting task
performance, workload, and satisfaction, compared to non-cognitively engineered interfaces
(Gerhardt-Powels, 1996). Methods such as the
think aloud method, verbal protocol analysis,
or cognitive task analysis are used to analyze in
detail the way in which humans perform tasks,

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

The Think Aloud Method and User Interface Design

mostly in interaction with a prototype computer

system.

BACKGROUND
In this section, we describe how the think aloud
method can be used to analyze a users task
behavior in daily life situations or in interaction
with a computer system and how these insights
may be used to improve the design of computer
systems. Thereafter, we will go into the pros and
cons of the think aloud method.

The Think Aloud Method

Thinking aloud is a method that requires subjects
to talk aloud while solving a problem or performing a task (Ericsson & Simon, 1993). This method
traditionally had applications in psychological and
educational research on cognitive processes. It
is based on the idea that one can observe human
thought processes that take place in consciousness. Thinking aloud, therefore, may be used to
know more about these cognitive processes and
to build computer systems on the basis of these
insights. Overall, the method consists of (1) collecting think aloud reports in a systematic way
and (2) analyzing these reports to gain a deeper
understanding of the cognitive processes that
take place in tackling a problem. These reports
are collected by instructing subjects to solve a
problem while thinking aloud; that is, stating
directly what they think. The data so gathered
are very direct; there is no delay. These verbal
utterances are transcribed, resulting in verbal
protocols, which require substantial analysis and
interpretation to gain deep insight into the way
subjects perform tasks (Deffner, 1990).

The Use of the Think Aloud Method

in Computer System Design
In designing computer systems, the think aloud
method can be used in two ways: (1) to analyze
users task behaviors in (simulated) working
practices, after which a computer system is actually built that will support the user in executing
similar tasks in future; or (2) to reveal usability
problems that a user encounters in interaction
with a (prototype) computer system that already
supports the user in performing certain tasks.
In both situations, the identification and selection of a representative sample of (potential) end
users is crucial. The subject sample should consist
of persons who are representative of those end users who will actually use the system in the future.
This requires a clearly defined user profile, which
describes the range of relevant skills of system
users. Computer expertise, roles of subjects in the
workplace, and a persons expertise in the domain
of work that the computer system will support are
useful dimensions in this respect (Kushnirek &
Patel, 2004). A questionnaire may be given either
before or after the session to obtain this information. As the think aloud method provides a rich
source of data, a small sample of subjects (eight
to 10) suffices to gain a thorough understanding of task behavior (Ericsson & Simon, 1993)
or to identify the main usability problems with
a computer system (Boren & Ramey, 2000). A
representative sample of the tasks to be used in
the think aloud study is likewise essential. Tasks
should be selected that end users are expected to
perform while using the (future) computer system. This requirement asks for a careful design
of tasks to be used in the study to assure that
tasks are realistic and representative of daily life
situations. It is recommended that task cases be
developed from real-life task examples (Kushnirek
& Patel, 2004).
Instructions to the subjects about the task at
hand should be given routinely. The instruction
on thinking aloud is straightforward. The essence

623

The Think Aloud Method and User Interface Design

is that the subject performs the task at hand, possibly supported by a computer, and says out loud
what comes to mind.
A typical instruction would be, I will give
you a task. Please keep talking out loud while
performing the task. Although most people do
not have much difficulty rendering their thoughts,
they should be given an opportunity to practice
talking aloud while performing an example task.
Example tasks should not be too different from the
target task. As soon as the subject is working on the
task, the role of the instructor is a restrained one.
Interference should occur only when the subject
stops talking. Then, the instructor should prompt
the subject by the following instruction: Keep
on talking (Ericsson & Simon, 1993).
Full audiotaping and/or videorecording of
the subjects concurrent utterances during task
performance and, if relevant, videorecording of
the computer screens are required to capture all
the verbal data and user/computer interactions
in detail. After the session has been recorded, it
has to be transcribed. Typing out complete verbal
protocols is inevitable to be able to analyze the data
in detail (Dix et al., 1998). Videorecordings may
be viewed informally, or they may be analyzed
formally to understand fully the way the subject
performed the task or to detect the type and number
of user-computer interaction problems.
The use of computer-supported tools that are
able to link the verbal transcriptions to the cor-

responding video sequences may be considered

to facilitate the analysis of the video data (Preece,
2002).
Prior to analyzing the audio and/or video data,
it is usually necessary to develop a coding scheme
to identify step-by-step how the subject tackled
the task and/or to identify specific user/computer
interaction problems in detail. Coding schemes
may be developed bottom-up or top-down. In a
bottom-up procedure, one would use part of the
protocols to generate codes by taking every new
occurrence of a cognitive subprocess code. For
example, one could assign the code guessing to
the following verbal statements: Could it be
X? or Lets try X. The remaining protocols
then would be analyzed by using this coding
scheme. An excerpt from a coded verbal protocol
is given in Figure 1. Note that the verbal protocol
is marked up with annotations from the coding
scheme. Otherwise, categories in the coding
scheme may be developed top-down, for example,
from examination of categories of interactions
from the human/computer interaction literature
(Kushnirek & Patel, 2004). Before it is applied,
a coding scheme must be evaluated on its intercoder reliability.
To prevent experimenter bias, it is best to leave
the actual coding of the protocols to a minimum of
two independent coders. Correspondence among
codes assigned by different coders to the same

Figure 1. Excerpt from a coded verbal protocol for analyzing humans task behavior

624

Code

Verbal protocol segment

Explanation

NPSCR04

How can I exit this screen?

Navigation problem screen04

MBT012

What does this button mean?

Meaning of button012

RTACT002

It has been busy a very long time

Response time after action002

VSSACT006

What is it doing now?

Visibility of system status after action006

MSSACT009

What does fatal error098 mean?

Meaning of system feedback after action009

The Think Aloud Method and User Interface Design

verbal statements must be found, for which the

Kappa mostly is used (Altman, 1991).
The coded protocols and/or videos can be
compiled and summarized in various ways,
depending on the goal of the study. If the goal is
to gain a deep insight into the way humans perform a certain task in order to use these insights
for developing a computer system to support
task performance, then the protocol and video
analyses can be used as input for a cognitive task
model. Based on this model, a first version of a
computer system then may be designed. If the
aim is to evaluate the usability of a (prototype)
computer system, the results may summarize any
type and number of usability problems revealed.
If the computer system under study is still under
development, these insights then may be used to
better the system.

PROS AND CONS OF THE THINK

ALOUD METHOD
The think aloud method, preferably used in
combination with audio- and/or videorecording,
is one of the most useful methods to gain a deep
understanding of the way humans perform tasks
and of the specific user problems that occur in
interaction with a computer system. As opposed to
other inquiry techniques, the think aloud method
requires little expertise, while it provides detailed
insights regarding human task behavior and/or
user problems with a computer system (Preece,
2002). On the other hand, the information provided
by the subjects is subjective and may be selective.
Therefore, a careful selection of the subjects who
will participate and the tasks that will be used in
the study is crucial. In addition, the usefulness
of the think aloud method is highly dependent
on the effectiveness of the recording method. For
instance, with audiotaping only, it may be difficult
to record information that is relevant to identify
step-by-step what the subjects were doing while

performing a task, whether computer-supported

or not (Preece, 2002).
Another factor distinguishing the think aloud
method from other inquiry techniques is the
promptness of the response it provides. The think
aloud method records the subjects task behavior
at the time of performing the task. Other inquiry
techniques, such as interviews and questionnaires, rely on the subjects recollection of events
afterwards. Subjects may not be aware of what
they actually are doing while performing a task
or interacting with a computer, which limits the
usefulness of evaluation measures that rely on
retrospective self-reports (Boren & Ramey, 2000;
Preece, 2002). The advantage of thinking aloud,
whether audio- or videotaped, as a data eliciting
method includes the fact that the resulting reports
provide a detailed account of the whole process
of a subject executing a task.
Although using the think aloud method is
rather straightforward and requires little expertise, analyzing the verbal protocols can be very
time-consuming and requires that studies are
well planned in order to avoid wasting time (Dix
et al., 1998).
The think aloud method has been criticized,
particularly with respect to the validity and
completeness of the reports it generates (Boren
& Ramey, 2000; Goguen & Linde, 1993).
An argument made against the use of the think
aloud method as a tool for system design is that
humans do not have access to their own mental
processes and, therefore, cannot be asked to report
on these. With this notion, verbalizing thoughts is
viewed as a cognitive process on its own. Since
humans are poor at dividing attention between
two different tasks (i.e., performing the task under
consideration and verbalizing their thoughts), it is
argued that thinking aloud may lead to incomplete
reports (Nisbett & Wilson, 1997).
However, this critique seems to bear on some
types of tasks that subjects are asked to perform
in certain think aloud studies. As Ericsson and
Simon (1993) point out, in general, talking out

625

The Think Aloud Method and User Interface Design

loud does not interfere with task performance

and, therefore, does not lead to much disturbance
of the thought processes. If reasoning takes place
in verbal form, then verbalizing thoughts is easy
and uses no extra human memory capacity.
However, if the information is nonverbal and
complicated, verbalization will not only cost
time but also extra human memory capacity.
Verbalization of thoughts then becomes a cognitive process by itself. This will cause the report
of the original task processing to be incomplete,
and sometimes, it even may disrupt this process
(Ericsson & Simon, 1993). Therefore, the think
aloud method only may be used on a restricted set
of tasks. Tasks for which the information can be
reproduced verbally and for which no information
is asked that is not directly used by the subject
in performing the task under attention are suitable for introspection by the think aloud method
(Boren & Ramey, 2000).
The fact that the experimenter may interrupt
the subject during task behavior is considered another source of error, leading to distorted reports
(Goguen & Linde, 1993). It has been shown, however, that as long as the experimenter minimizes
interventions in the process of verbalizing and
merely reminds the subject to keep talking when
a subject stops verbalizing his or her thoughts, the
ongoing cognitive processes are no more disturbed
than by other inspection techniques (Ericsson &
Simon, 1993).
The think aloud method, if applied under prescribed conditions and preferably in combination
with audio- and/or videorecording, is a valuable
information source of human task/behavior and,
as such, a useful technique in designing and
evaluating computer systems.

FUTURE TRENDS
The think aloud method is propagated and far more
often used as a method for system usability testing
than as a user requirements eliciting method. In

626

evaluating (prototype) computer systems, thinking aloud is used to gain insight into end users
usability problems in interaction with a system
to better the design of these systems. The use of
think aloud and video analyses, however, may be
helpful not merely in evaluating the usability of
(prototype) computer systems but also in analyzing in detail how end users tackle tasks in daily
life that in the end will be computer supported.
The outcomes of these kinds of analyses may
be used to develop a first version of a computer
system that directly and fully supports users in
performing these kinds of tasks. Such an approach
may reduce the time spent in iterative design of
the system, as the manner in which potential
end users process tasks is taken into account in
building the system.
Although a deep understanding of users task
behaviors in daily settings is indispensable in
designing intuitive systems, we should keep in
mind that the implementation of computer applications in real-life settings may change and
may have unforeseen consequences for work
practices. So, besides involving potential user
groups in an early phase of system design and in
usability testing, it is crucial to gain insight into
how these systems may change these work practices to evaluate whether and how these systems
are being used. This adds to our understanding
of why systems may or may not be adopted into
routine practice.
Today, a plea for qualitative studies for studying a variety of human and contextual factors that
likewise may influence system appraisal is made
in literature (Aarts et al., 2004; Ammenwerth et
al., 2003; Berg et al., 1998; Orlikowski, 2000; Patton, 2002). In this context, sociotechnical system
design approaches are promoted (Aarts et al., 2004;
Berg et al., 1998; Orlikowski, 2000). Sociotechnical system design approaches are concerned not
only with human/computer interaction aspects of
system design but also take psychological, social,
technical, and organizational aspects of system
design into consideration. These approaches

The Think Aloud Method and User Interface Design

take an even broader view of system design and

implementation than cognitive engineering approachesthe organization is viewed as a system
with people and technology as components within
this system. With sociotechnical system design
approaches, it can be determined which changes
are necessary and beneficial to the system as a
whole, and these insights then may be used to decide on the actions to effect these changes (Aarts
et al., 2004; Berg et al., 1998). This process of
change never stops; even when the implementation of a computer system is formally finished,
users still will ask for system improvements
to fit their particular requirements or interests
(Orlikowski, 2000).

References

CONCLUSION

Berg, M., Langenberg, C., Van Der Berg, I., &

Kwakkernaat, J. (1998). Considerations for sociotechnical design: Experiences with an electronic
patient record in a clinical context. International
Journal of Medical Informatics, 52(3), 243-251.

The use of the think aloud method may aid in

designing intuitive computer interfaces, because
using thinking aloud provides us with a more
thorough understanding of work practices than
do conventional techniques such as interviews
and questionnaires.
Until now, thinking aloud was used mostly to
evaluate prototype computer systems. Thinking
aloud, however, likewise may be used in an earlier
phase of system design, even before a first version of the system is available. It then can be used
to elicit every step taken by potential end users
to process a task in daily work settings. These
insights then may be used as input to the design
of a computer systems first version.
Development, fine-tuning, testing, and final
implementation of computer systems take a lot
of time and resources. User involvement in the
whole life cycle of information systems is crucial,
because only when we really try to understand
end users needs and the way they work, think,
and communicate with each other in daily practice
can we hope to improve computer systems.

Aarts, J., Doorewaard, H., & Berg, M. (2004).

Understanding implementation: The case of a
computerized physician order entry system in a
large Dutch university medical center. Journal of
the American Medical Informatics Association,
11(3), 207-216.
Altman, D.G. (1991). Practical statistics for medical research. London: Chapman and Hall.
Ammenwerth, E., Grber, S., Herrmann, G.,
Brkle, T., & Knig, J. (2003). Evaluation of health
information systemsProblems and challenges.
International Journal of Medical Informatics,
71(2-3), 125-135.

Boren, M.T., & Ramey, J. (2000). Thinking

aloud: Reconciling theory and practice. IEEE
Transactions on Professional Communication,
43(3), 261-278.
Carroll, J.M. (2002). Human computer interaction in the new millennium. New York: Addison
Wesley.
Deffner, G. (1990). Verbal protocols as a research
tool in human factors. Proceedings of the Human
Factors Society, 34t h Annual Meeting.
Dix, A., Finlay, J., Abowd, G., & Beale, R. (1998).
Human computer interaction. London: Prentice
Hall.
Ericsson, K.A., & Simon, H.A. (1993). Protocol
analysis: Verbal reports as data. Cambridge,
MA: MIT Press.
Gerhardt-Powals, J. (1996). Cognitive engineering principles for enhancing human computer

627

The Think Aloud Method and User Interface Design

performance. International Journal of Human

Computer Interaction, 8(2), 189-211.

KEY TERMS

Goguen, J.A., & Linde, C.L. (1993). Techniques

for requirements elicitation. Proceedings of the
IEEE International Conference on Requirements
Engineering.

Cognitive Engineering: A field aiming at

understanding the fundamental principles behind
human activities that are relevant in context of designing a system that supports these activities.

Kaplan, B. (2001). Evaluating informatics applicationsSome alternative approaches: Theory,

social interactionism, and call for methodological pluralism. International Journal of Medical
Informatics, 64(1), 39-56.

Cognitive Task Analysis: The study of the

way people perform tasks cognitively.

Kushnirek, A.W., & Patel V.L. (2004). Cognitive

and usability engineering methods for the evaluation of clinical information systems. Journal of
Biomedical Informatics, 37, 56-76.
Nielsen, J. (1993). Usability engineering. Toronto:
Academic Press.
Nisbett, R.E., & Wilson, T.D. (1997). Telling more
than we know: Verbal reports on mental processes.
Psychological Review, 3, 231-241.
Orlikowski, W.J. (2000). Using technology and
constituting structures: A practice lens for studying technology in organizations. Organizational
Science, 11, 404-428.
Patton, M.Q. (2002). Qualitative research and
evaluation methods. Thousand Oaks, CA: Sage
Publications.
Preece, J. Rogers, Y., & Sharp, H. (2002). Interaction design: Beyond human computer interaction.
New York: Wiley.
Rossen, M.B., & Carroll, J.M. (2002). Usability
engineering. New York: Academic Press.

Cognitive Task Model: A model representing the cognitive behavior of people performing
a certain task.
Sociotechnical System Design Approach:
System design approach that focuses on a sociological understanding of the complex practices in
which a computer system is to function.
Think Aloud Method: A method that requires
subjects to talk aloud while solving a problem or
performing a task.
User Profile: A description of the range of relevant skills of potential end users of a system.
Verbal Protocol: Transcription of the verbal
utterances of a test person performing a certain
task.
Verbal Protocol Analysis: Systematic
analysis of the transcribed verbal utterances to
develop a model of the subjects task behavior
that then may be used as input to system design
specifications.
Video Analysis: Analysis of videorecordings of the user/computer interactions with the
aim to detect usability problems of the computer
system.

Stary, C., & Peschl, M.F. (1998). Representation

still matters: Cognitive engineering and user
interface design. Behavior and Information
Technology, 17(6), 338-360.

This work was previously published in Encyclopedia of Human Computer Interaction, edited by C. Ghaoui, pp. 597-602, copyright 2006 by Information Science Reference, formerly known as Idea Group Reference (an imprint of IGI Global).
628

629

Chapter 2.16

Lessons Learned in
Designing Ubiquitous
Augmented Reality
User Interfaces
Christian Sandor
Technische Universitt Mnchen, Germany
Gudrun Klinker
Technische Universitt Mnchen, Germany

Introduction
In recent years, a number of prototypical demonstrators have shown that augmented reality has the
potential to improve manual work processes as
much as desktop computers and office tools have
improved administrative work (Azuma et al., 2001;
Ong & Nee, 2004). Yet, it seems that the classical
concept of augmented reality is not enough (see
also http://www.ismar05.org/IAR). Stakeholders
in industry and medicine are reluctant to adopt
it wholeheartedly due to current limitations of
head-mounted display technology and due to
the overall dangers involved in overwhelming a
users view of the real world with virtual information. It is more likely that moderate amounts
of augmented reality will be integrated into a

more general interaction environment with many

displays and devices, involving tangible, immersive, wearable, and hybrid concepts of ubiquitous
and wearable computing. We call this emerging
paradigm ubiquitous augmented reality (UAR)
(MacWilliams, 2005; Sandor, 2005; Sandor &
Klinker, 2005).
It is not yet clear which UAR-based humancomputer interaction techniques will be most
suitable for users to simultaneously work within
an environment that combines real and virtual
elements. Their success is influenced by a large
number of design parameters. The overall design
space is vast and difficult to understand.
In Munich, we have worked on a number of
applications for manufacturing, medicine, architecture, exterior construction, sports, and enter-

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Lessons Learned in Designing Ubiquitous Augmented Reality User Interfaces

tainment (a complete list of projects can be found

at http://ar.in.tum.de/Chair/ProjectsOverview).
Although many of these projects were designed
in the short-term context of one semester student
courses or theses, they provided insight into
different aspects of design options, illustrating
trade-offs for a number of design parameters. In
this chapter, we propose a systematic approach
toward identifying, exploring, and selecting
design parameters at the example of three of our
projects, PAARTI (Echtler et al., 2003), FataMorgana (Klinker et al., 2002), and a monitoring
tool (Kulas, Sandor, & Klinker, 2004).
Using a systematic approach of enumerating
and exploring a defined space of design options
is useful, yet not always feasible. In many cases,
the dimensionality of the design space is not
known a-priori but rather has to be determined
as part of the design process. To cover the variety of aspects involved in finding an acceptable
solution for a given application scenario, experts
with diverse backgrounds (computer science,
sensing and display technologies, human factors,
psychology, and the application domain) have to
collaborate. Due to the highly immersive nature
of UAR-based user interfaces, it is difficult for
these experts to evaluate the impact of various
design options without trying them. Authoring
tools and an interactively configurable framework are needed to help experts quickly set up
approximate demonstrators of novel concepts,
similar to back-of-the-envelope calculations
and sketches. We have explored how to provide
such first-step support to teams of user interface
designers (Sandor, 2005). In this chapter, we report
on lessons learned on generating authoring tools
and a framework for immersive user interfaces
for UAR scenarios.
By reading this chapter, readers should
understand the rationale and the concepts for
defining a scheme of different classes of design
considerations that need to be taken into account
when designing UAR-based interfaces. Readers
should see how, for classes with finite numbers of

630

design considerations, systematic approaches can

be used to analyze such design options. For less
well-defined application scenarios, the chapter
presents authoring tools and a framework for
exploring interaction concepts. Finally, a report
on lessons learned from implementing such tools
and from discussing them within expert teams of
user interface designers is intended to provide
an indication of progress made thus far and next
steps to be taken.

Background
In this section, we provide an overview of the
current use of UAR-related interaction techniques
and general approaches toward systematizing the
exploration of design options.

User Interface Techniques for

Ubiquitous Augmented Reality
User interfaces in UAR are inspired by related
fields, such as virtual reality (VR) (Bowman,
Kruijff, LaViola, & Poupyrev, 2004), attentive
user interfaces (AUIs) (Vertegaal, 2003), and
tangible user interfaces (TUIs) (Ishii & Ullmer,
1997). Several interaction techniques for VR have
been adapted to UAR: for example the World-inMiniature (Bell, Hllerer, & Feiner, 2002), pinch
gloves for system control (Piekarski, 2002), and
a flexible pointer to grasp virtual objects that
are beyond arms reach (Olwal & Feiner, 2003).
The core idea of TUIs is to use everyday items
as input and output simultaneously. This idea has
also been applied to UAR (Kato, Billinghurst,
Poupyrev, Tetsutani, & Tachibana, 2001; Klinker,
Stricker, & Reiners, 1999; MacWilliams et al.,
2003). Ideas from AUIs have been used in UAR
interfaces by using head tracking (Olwal, Benko,
& Feiner, 2003) and eye tracking (Novak, Sandor,
& Klinker, 2004).

Lessons Learned in Designing Ubiquitous Augmented Reality User Interfaces

Implementing New Interaction

Techniques
To develop new interaction techniques and visualizations for UAR, several software infrastructures
have been created to simplify the development
of new interaction techniques by programmers:
distributed frameworks, dataflow architectures,
user interface management systems, scene graphbased frameworks, a variety of class libraries, and
finally scripting languages. A detailed discussion
can be found in Sandor (2005).
For novice users, several desktop tools for
authoring augmented reality content have been
developed: PowerSpace (Haringer & Regenbrecht, 2002), DART (MacIntyre, Gandy, Dow, &
Bolter, 2004), and MARS (Gven & Feiner, 2003).
Several systems exist that follow an immersive
authoring approach (Lee, Nelles, Billinghurst,
& Kim, 2004; Poupyrev et al., 2001). Piekarski
describes a mobile augmented reality system that
can be used to capture the geometries of real objects (Piekarski, 2002). Several hybrid authoring
approaches combine immersive authoring with
desktop authoring (Olwal & Feiner, 2004; Zauner,
Haller, Brandl, & Hartmann, 2003).

Design Optimization for HighDimensional Design Spaces

One of the most difficult issues in designing novel
interaction techniques for UAR is the wealth of
criteria that are potentially involved in finding
an optimal solution. We divide such criteria into
three classes: criteria pertaining to the task(s) that
need to be executed, the knowledge and skills
of the user, and the current state-of-the-art of
technology. Figure 1 illustrates the classes and
their relationships.

Figure 1. Design criteria classified according to

tasks, systems, and users

Classes of Design Criteria

Task-specific criteria are related to the

requirements of specified tasks in an application. According to principles of software
engineering, they are determined from
scenarios and use cases, taking the environmental setting and the required technical
quality into account. Yet, they may change
over time due to changing work processes,
which may indirectly depend on evolving
technology.
System-specific criteria are defined by the
state-of-the art of engineering-related parameters of sensing and display devices and
computer systems. Due to evolving technology, these criteria have to be continuously
re-evaluated, resulting in ever-changing
optimal system configurations (Klinker et
al., 1999).
User-specific criteria depend on ergonomic issues and the cultural background
of usershuman factors and anthropology.
They describe current working conditions,
habits (working culture), and educational
background, as well as specific user-related
restrictions.

631

Lessons Learned in Designing Ubiquitous Augmented Reality User Interfaces

Criteria Reduction through

Inter-Class Constraints
Finding an overall optimal system that works
perfectly with respect to all criteria seems to be
impossible. We have thus adopted the approach
of selecting specific criteria of one or two classes
to impose constraints on design options in other
classes. In this section, we analyze the relationship between the classes of criteria from user-,
system-, and task-centric specifications. Later we
illustrate the exploitation of such constraints in
specific examples.
The relationships between task and system
requirements are described by the edge linking the
task and system nodes in Figure 1. From the taskperspective, they are described as the functional
and non-functional requirements of software
systems. From the system-perspective, they need
to be matched with the currently available technical options. Trade-offs must be made to obtain
pragmatically implementable solutionswith an
eye toward upcoming requirements and technical
developments. Later, we will present an example
of casting such trade-offs.
The relationships between task and user requirements are described by the edge linking the
task and user nodes in Figure 1. This case does
not involve any considerations of currently available technology. Thus, options that are discussed
here should hold true now, as well as 100 years
ago or 100 years in the future. They are analyzed
by disciplines such as task analysis and system
ergonomics (Bubb, 1993). Yet, they can provide
significant constraints upon todays technically
achievable system configurations. We will present
an example of analyzing how a user (car designer)
physically behaves with respect to a number of
tasks geared toward analyzing and comparing
different automotive designs.
The relationships between system and user
are described by the edge linking the system and
user nodes in Figure 1. From the user-perspective,
they are described as usability criteria, evaluat-

632

ing how users perform, given a specific technical

systemin comparison to other technical options.
From the system-perspective, they describe user
requirements that need to be satisfied with currently available technical means. Later, we show
an example of how a specific technical device can
be evaluated with respect to specific physical user
skills in using such a device.

Dealing with Ill-Defined Design

Spaces
By applying inter-class constraints on a user
interface, the design space can often be reduced
considerably. The next step in our proposed design
process is to explore the reduced design space with
interactive tools that encourage collaboration. In
this section, we first give the rationale for our interactive tools. Then, we proceed by highlighting
the problems that occur when using this approach.
Finally, we give an outlook of how we elaborate
on these concepts within this chapter.
To further explore the design space, collaboration between researchers with different
backgrounds is imperative to yield a solution
that is well balanced according to our three
main classes: user, task, and system. Thinking
about this problem led to the invention of a new
development process: Jam Sessions. The name
Jam Sessions was inspired by the spontaneous
collaboration of Jazz musicians that is also named
Jam Sessions. However, in our case we collaborate
on user interface elements, instead of music. In
Jam Sessions, development takes place at system
runtime, next to a running system. This allows
playful exploration of user interface ideas. Our
experience with Jam Sessions was first presented
in (MacWilliams et al., 2003); we have already
discussed these from a software engineering
(MacWilliams, 2005) and user interface (Sandor,
2005) perspective.
To support this development process, interactive tools for novices are an important ingredient,
since they foster the interdisciplinary collabora-

Lessons Learned in Designing Ubiquitous Augmented Reality User Interfaces

tion with other researchers. Desirable would be a

set of generic tools that can be applied in all Jam
Sessionsindependent of the user interface to
be developed. Although we have achieved this
for programmers, for novices this is yet an unsolved problem. We go in line with several other
research tools that allow modifying only a quite
limited amount of user interface functionality.
Since these tools are customized toward the user
interface that has to be built, most projects require
writing new tools. Thus, a sophisticated software
infrastructure that quickly allows building new
tools is very useful.
The final section of this chapter describes a
complex user interface that we have designed in
Jam Sessions. Additionally, we first describe briefly our software infrastructure and elaborate on the
tools that we have created for this project.

Projects Using Inter-Class

Constraints
This section presents three examples of analyzing design options by exploring inter-class
constraints.

PAARTI
In the PAARTI project (practical applications of
augmented reality in technical integration), we
have developed an intelligent welding gun with
BMW that is now being used on a regular basis
to weld studs in the prototype production of cars
(Echtler et al., 2003). It exemplifies the systematic exploitation of constraints between task and
system criteria.
The task was to assist welders in positioning
the tip of a welding gun with very high precision
at some hundred predefined welding locations on
a car body. The main system design issue was to
find an immersive solution with maximal precision. An AR-based system would need a display
(D), a tracking sensor (S), and some markers (M)

that needed to be installed in the environment on

the user or on the welding gun in a manner that
would yield maximal precision. As a fourth option, we considered the case that one of the objects
(esp.: markers) would not be necessary at all. The
result was the definition of a 3-dimensional design
space, S M D, with each dimension spanning
a range of four options. In total, there were 43 =
64 solutions that needed to be considered.
According to an analysis of all options, the
highest precision could be achieved by using an
outside-in tracking arrangement with sensors
placed in the welding environment and markers
attached to the welding gun. A small display was
attached to the welding gun. The visualization
used a notch and bead metaphor of real guns,
consisting of several concentric rings. A sphere
was positioned three-dimensionally at the next
welding location. Welders were requested to
capture the sphere within the concentric rings by
moving the gun (and the display) to the appropriate location (see Figure 2(a)).
The left diagram of Figure 2(b) shows the
optimal setup. The center diagram of Figure 2(b)
indicates the design space of S M = 16 options of
placing markers and sensors in the environment,
with the fixed display dimension D of placing the
display on the welding gun. The right diagram
presents the analysis pertaining to the achievable
precision for this setup. This setup clearly outperforms the classical AR-arrangement involving a
head-mounted display (Echtler et al., 2003).

FataMorgana
In the FataMorgana project, we have developed
an AR-based prototypical demonstrator for
designers at BMW, helping them compare real
mock-ups of new car designs with virtual models
(Klinker et al., 2002). The rationale for building
this system was that, although the importance of
digital car models is increasing, designers have
not yet committed wholeheartedly to a VR-based
approach but rather prefer relying on physical

633

Lessons Learned in Designing Ubiquitous Augmented Reality User Interfaces

Figure 2. Part of the design space for the intelligent welding gun (Reprinted with permission of Springer
Science and Business Media, Echtler et al., 2003, Figures 17.6, 17.7, and 17.8)

(a) Intelligent welding gun

feasability

3
4

none
room

sensors

user
tool
none

room

user

tool

markers

(b) Display attached to the welding gun

mock-ups. One of the reasons may be that special

viewing arrangements such as projection walls
do not permit people to view the digital models
within a real environment. AR can help alleviate
this problem by placing virtual cars next to real
(mock-up) cars.
Here we present this project as an example of
a systematic analysis of the relationships between
tasks and user actions. The underlying thesis is
that users (designers) behave in specific ways in
order to achieve tasks. If a system is expected

634

to support users in achieving their tasks, it has

to be designed to function well within the range
of typical actions performed by the user. To this
end, we have subdivided the task in a set of different approaches and asked a designer to act out
each of these tasks within the real car presentation environment. We recorded the designers
motions with a camera that was attached to his
head (Figure 3).

Lessons Learned in Designing Ubiquitous Augmented Reality User Interfaces

Figure 3. Five different tasks resulting in different user head motions ( 2002 IEEE, Klinker et al.,
2002)

(a) Turning

(b) Overview

(c) Detail

(d) Discuss

(e) Compare

Turning: The car designer remains in a

fixed location and looks at the car rotating
on a turn table.
Overview: The car designer performs an
overview evaluation of the car, by walking
around and turning his head to change the
lighting conditions.
Detail: The car designer focuses on a specific
detail of the car, such as a character line on
the side of the car or the shape of the front
spoiler.
Discuss: The car designer discusses the car
under evaluation with a colleague.
Compare: The car designer compares two
cars, for example, an existing car and a new
design.

For each scenario, we determined the typical

angular range of head rotations, as well as the
range of positional changes. Combined with a
projection of the field of view onto the environment, this gave us an indication how markers had
to be laid out in the room in order to guarantee

that enough of them were clearly visible during

all user actions.

Monitoring Tool for Determining

Usage Patterns of Novel Interaction
Techniques
We have developed a monitoring tool (Kulas et
al., 2004) to evaluate the usability of novel input
techniques and devices. The monitoring tool allows us to systematically analyze relationships
between user and system criteria, analyzing
whether a system is well tailored to the physiological and cognitive skills of its users.
We have used the monitoring tool to evaluate the usability of a novel input device called
TouchGlove that was developed at Columbia
University (Blasko & Feiner, 2002). It consists of
a touch-sensitive plate (similar to a touchpad in
a laptop) that is attached to the center of a users
palm. It is sensitive to single-finger input, measuring 2D location and pressure. In the evaluation
setup, we have compared two techniques of using

635

Lessons Learned in Designing Ubiquitous Augmented Reality User Interfaces

the TouchGlove to select items from a menu. In

the first case, users were asked to make a linear
gesture with their fingertip on the TouchGlove to
select items from a regular pull-down menu. In
the second case, the TouchGlove was combined
with a gyroscope to select items from a pie menu.
Users were asked to rotate their hands around
their wrists, generating only a tapping signal on
the touchpad to signal start and end of the gesture
rotating the pie menu.
During a usability evaluation, the user is placed
at a suitable distance from a usability engineer.
The engineer enters observations into a usability
logging system and also monitors what the user
actually sees on screen. Simultaneously, he also
monitors real-time visualizations of measured
usability data. The tool provides immediate
feedback during an interactive tryout session,
thereby supporting Jam Sessions.

Interactive Tools for

Collaborative Design Space
Explorations
This section presents our tools for supporting
Jam Sessions. First, we give a brief overview of
our tools. Second, we present an interdisciplinary
research project, CAR, which uses them. We close
with a description of the underlying real-time
development environment.

Overview of Tools
To support Jam Sessions, we have created a toolbox
of lightweight and flexible tools. They form the
basic building blocks which user interface development teams can use to generate, experience, and
test their novel interaction techniques.
The tools use AR, TUI, and WIMP interaction
paradigms and are designed to support a number
of tasks. The first task focuses on monitoring the
user. The second task involves the configuration
of dataflow networks. UAR systems need to

636

communicate in real-time with many sensing and

display devices, requiring a distributed system
approach. A dataflow network connects such
devices and components. We provide tools that
allow modifying these dataflow graphs during
runtime. Another task is related to the adjustment
of dialog control, i.e., the control of the high-level
behaviour of a user interface. Tools that enable
developers to specify dialog control quickly speed
up the development process significantly. The
final task involves the creation of context-aware
animations. Conventional animations have time
as the only parameter that changes the appearance of graphical elements. However, for mobile
systems a variety of research projects (e.g., a
context-aware World-in-Miniature [Bell et al.,
2002]) have explored animations that change their
appearance according to context.
We have developed six tools, T1T6, in support of these tasks. T1 collects and evaluates
usability data during system runtime. T2 uses
an augmented reality visualization to shows a
users visual focus of attention in a combination
of head and eye tracking (Novak et al., 2004)
(see Figure 9). T3 is a graphical editor, DIVE to
adjust dataflow networks (MacWilliams et al.,
2003; Pustka, 2003) (see Figure 10[a]). T4 is an
immersive visual programming environment
(Sandor, Olwal, Bell, & Feiner, 2005). T5 is a User
Interface Controller Editor, UIC, to graphically
specify dialog control by composing Petri nets
(Hilliges, Sandor, & Klinker, 2004) (see Figure
10(b)). T6 is a collection of tools to experiment
with context-aware mobile augmented reality
user interfaces.
Figure 4 classifies our tools with respect to
the user interface paradigms they employ and
the tasks they address (Sandor, 2005). It shows
that we sometimes developed several tools, addressing the same task, using different interaction
paradigms. This reflects our goal of exploring and
comparing design options for our own tools as
much as for the interaction techniques that will
be developed with them.

Lessons Learned in Designing Ubiquitous Augmented Reality User Interfaces

Figure 4. Classification of implemented tools. Development tasks are addressed with tools that use different user interface paradigms
User interface paradigm

Augmented
reality

WIMP

Tangible

Monitoring
the user

Configuring dataflow
networks

CAR
CAR is an industry-sponsored multi-disciplinary project to investigate issues pertaining to
the design of augmented reality user interfaces
in cars. CAR has used most of the tools T1-T6 to
investigate several user interface questions.

Motivation
In CAR, we have investigated a variety of questions: How can information be presented efficiently across several displays that can be found
in a modern car (e.g., the dashboard, the board
computer, and heads-up displays (HUDs))? How
can we prevent that information displayed in a
HUD is blocking the drivers view in crucial
situations? Since a wealth of input modalities

Specifying

Creating con-

dialog control

text-aware animations

can be used by a car driver (tactile, speech, head

and hand gestures, eye motion), which modalities
should be used for which tasks?
In a multi-disciplinary UI design team, we have
discussed, for example, how to present a navigation
map on a HUD. Where should it be placed? How
large should it be? What level of detail should it
provide? Should it be a two-dimensional map or
a tilted view onto a three-dimensional environmental model (WIM)? If so, which viewing angle
should be selected? Will the angle, as well as the
position of the WIM and the size and zoom factor
adapt to sensor parameters, such as the current
position of the car while approaching a critical
traffic area in a town?

637

Lessons Learned in Designing Ubiquitous Augmented Reality User Interfaces

Physical Setup
We have set up a simulator for studying car
navigation metaphors in traffic scenes (Figure
5). It consists of two separate areas: a simulation
control area (large table with a tracked toy car)
and a simulation experience area (person sitting
at the small table with a movable computer monitor in the front and a stationary large projection
screen in the back). In the simulation control area,
members of the design team can move one or
more toy cars on the city map to simulate traffic
situations, thereby controlling a traffic simulator
via a tangible object. The simulation experience
area represents the cockpit of a car and the driver.
The picture projected on the large screen in the
front displays the view a driver would have when
sitting in the toy car. The monitor in front of the
driver provides a mock-up for the visualizations
to be displayed in a HUD. Further monitors can
be added at run-time, if more than one view is
needed.
The room is equipped with an outside-in
optical tracking system (http://www.ar-tracking.

Figure 5
. Physical setup

638

de). The cameras track the toy car, the computer

monitor, and the user (simulating a car driver).
Each tracked object is equipped with a marker
consisting of a rigid, three-dimensional arrangement of reflective spheres.
Information is presented on several devices
and surfaces in the room: A projector at the
ceiling projects a birds eye view of a city onto
the large, stationary table on the right. Another
projector presents the current, egocentric view of
a virtual car driver sitting in the toy car on the
large screen at the front wall. A third, locationdependent visualization of the driving scenario
is shown on the mobile computer monitorour
substitute for a HUD.
The system provides tools for a team of design experts with diverse backgrounds to jointly
explore various options to present a map (Figure
6).

Lessons Learned in Designing Ubiquitous Augmented Reality User Interfaces

Figure 6
. Discussion of user interface options for car navigation in a design team

(a)

Controlling the Context-Aware

Adjustment of a Navigation Map
It is not yet clear how navigational aids are best
presented within a drivers field of view. In the
CAR project, we have experimented with various options of placing and orienting a map in a
HUD.
Figure 7 shows how our system provides
designers with a tangible objecta platethat
is correlated with the orientation (tilt) and zoom
of a 3D map on the HUD. When the user moves
the tangible plane, the 3D map is turned and
zoomed accordingly on the HUD. Figure 6(a)
shows a member of the design team experiment
with different map orientations.
The position and size of a map in a HUD may
have to depend on various parameters that depend
on the driving context, such as the current position
of the car relative to its destination, the drivers
viewing direction, and immanent dangers in the
environment. Interface designers need to explore
schemes for the display system to automatically
adapt to context parameters. Figure 8 shows an
interactive sketching tool for designers to describe functional dependencies between context

(b)

parameters and display options. Figure 9 shows

first steps toward using tracked head and eye motions to provide a context-dependent interaction
scheme (Novak et al., 2004). Figure 6(a) shows
the head and eye tracking device. We are in the
process of analyzing context-dependent information presentation further. First user studies of
selected issues are presented in Tnnis, Sandor,
Klinker, Lange, and Bubb (2005).

Real-Time Development
Environment for Interaction Design
The tools presented in earlier sections were geared
toward the immediate use by non-programming
user interface experts. They mainly address the
customization of a set of functionalities and filters,
linking context measurements to information
presentation schemes. In order to add new functionality to a system, the development team must
also be able to modify the underlying network of
components, and its dataflow scheme. Tools T3
and T5 provide such support.
All system configuration tools are based on
DWARF (distributed wearable augmented reality
framework) (Bauer et al., 2001) and AVANT-

639

Lessons Learned in Designing Ubiquitous Augmented Reality User Interfaces

Figure 7
. Tangible interaction for adjustment of a three-dimensional map

Figure 8
. Sketching the context visualization function: (a) Staircase function; (b) linear function

(a)

(b)

Figure 9. Attentive user interface, visualizing a drivers eye and head motions: (a) The DWARFs
Interface Visualization Environment for managing distributed components; (b) The User Interface
Controller for specifying dialog control

(a)

GUARDE (Sandor, 2005; Sandor & Klinker,

2005). DWARF is the underlying infrastructure
that connects a set of distributed components.
AVANTGUARDE is composed of DWARF components that address the specific requirements for
user interfaces in UAR.
640

(b)

DWARFs interactive visualization environment (MacWilliams et al., 2003) (tool T3, Figure
10[a]) enables developers to monitor and modify
the dataflow network of distributed components.
However, since this requires substantial knowl-

Lessons Learned in Designing Ubiquitous Augmented Reality User Interfaces

Figure 10
. Tools for programmers used in CAR: (a) The DWARFs interactive visualization environment
for managing distributed components; (b) the user interface controller for specifying dialog control

(b)

(a)

edge of DWARF and distributed programming,

novices have difficulties to use this tool.
The core component of AVANTGUARDE is a
Petri net-based dialog control management system
(Hilliges et al., 2004) (Tool T5, Figure 10[b]).
We have developed a visual programming environment that eases the modification of the Petri
nets (and accordingly the user interface) during
system runtime. However, it is still too difficult
to use for non-programming design experts, since
understanding Petri nets requires knowledge in
computer science.

Conclusion
In PAARTI and FataMorgana, we have learned
that the reduction of criteria through inter-class
constraints is a valuable approach for designing

user interfaces. The crucial issue of this method

is to determine the most important constraints
by talking with domain experts. The subsequent
systematic design space exploration is straightforward.
We have presented an example for the interclass constraint of user and system: the evaluation
of the TouchGlove input device. In this project, we
have observed the importance of immediate feedback through interactive tools. Our first prototype
of the TouchGlove had a loose contact. While we
conducted the usability study with the first user,
we immediately spotted the problem and solved
it. This saved us a lot of valuable time.
Our tool-based approach for further design
space explorations has been applied successfully
in several projects. The idea of providing user
interface developers with a toolbox of flexible,
lightweight tools seems feasible. However, one

641

Lessons Learned in Designing Ubiquitous Augmented Reality User Interfaces

problem has to be pointed out: when creating a

variety of tools, a supporting real-time development environment is imperative. Otherwise, too
much development time has to be allocated to
tool creationleaving little time for the actual
use of the tools. In this respect, we have successfully built our tools on top of DWARF and
AVANTGUARDE.
The combination of tools with different user
interface paradigms turned out to be a valuable
idea. We have made two important observations:
first, there seems to be a trade-off between ease
of use for a tool and the complexity of results
that can be accomplished with it. WIMP tools
can be used to model more complex interactions,
whereas ease of use is greater with tools that have
a tangible user interface or an augmented reality
user interface. Second, the combination of tools
with different paradigms opens new possibilities
of interaction design that would not be possible
with tools employing a single paradigm. Interaction designers are typically not fluent in complex
programming tasks, so their involvement with
easy to use tools yields important benefits.
Ideally, it would be enough to create one
generic tool that novices can use to explore the
design space of UAR user interfaces. Our first
prototype toward this goal has been published
in Sandor et al. (2005).
This tool seems to be very easy to use, as it
employs only direct manipulation of real world
objectsno conventional programming is required at all. However, the ceiling of the tool
(i.e., what can be achieved with it) is quite low,
since our system supports only a fixed, and very
limited number of operations. We are exploring
how we can extend it to allow users to specify new
operations at runtime. While we anticipate using
programming-by-demonstration to address a carefully planned universe of possibilities, supporting
arbitrary operations through demonstration and
generalization is an open problem.
The CAR project also showed us that for design
space explorations, rapid prototyping is more im-

642

portant than realism for finding new interaction

techniques. Though, for the thorough evaluation
of these new concepts, formal usability studies
within a realistic environment are still necessary.
We have conducted a first study in this respect
(Tnnis et al., 2005).

References
Azuma, R., Baillot, Y., Behringer, R., Feiner, S.,
Julier, S., & MacIntyre, B. (2001). Recent

advances
in augmented reality. IEEE Computer Graphics
and Applications, 21(6), 34-47.
Bauer, M., Brgge, B., Klinker, G., MacWilliams,
A., Reicher, T., Riss, S., et al. (2001).

Design of a
component-based augmented reality framework.
In ISAR01: Proceedings of the International
Symposium on Augmented Reality (pp. 45-54).
New York.
Bell, B., Hllerer, T., & Feiner, S. (2002). An

annotated situation-awareness aid for augmented

reality. In UIST 02: Proceedings of the 15th Annual
ACM Symposium on User interface Software and
Technology (pp. 213-216). Paris: ACM Press.
Blasko, G., & Feiner, S. (2002). A menu interface
for wearable computing. In ISWC 02: Proceedings of the 6th IEEE International Symposium on
Wearable Computers (pp. 164-165).
Bowman, D. A., Kruijff, E., LaViola, J. J., &
Poupyrev, I. (2004). 3D user interfaces: Theory
and practice. Redwood City, CA: Addison Wesley
Longman Publishing Co.
Bubb, H. (1993). Systemergonomische Gestaltung. In H. Schmidtke (Ed.), Ergonomie (1st ed.,
pp. 390-420). Mnchen, Germany: Carl Hanser.
Echtler, F., Sturm, F., Kindermann, K., Klinker,
G., Stilla, J., Trilk, J., et al.
(2003). The intelligent
welding gun: Augmented reality for experimental
vehicle construction. In S. Ong & A. Nee (Eds.),

Lessons Learned in Designing Ubiquitous Augmented Reality User Interfaces

Virtual and augmented reality applications in

manufacturing. London: Springer Verlag.
Gven, S., & Feiner, S. (2003). A hypermedia
authoring tool for augmented and virtual reality.
New Review of Hypermedia, 9(1), 89-116.
Haringer, M., & Regenbrecht, H. (2002). A pragmatic approach to augmented reality authoring.
In ISMAR02: Proceedings of the IEEE and
ACM International Symposium on Mixed and
Augmented Reality, Darmstadt, Germany (pp.
237-245).
Hilliges, O., Sandor, C., & Klinker, G. (2004).
A lightweight approach for experimenting with
tangible interaction metaphors. In MU3I 04:
Proceedings of the International Workshop on
Multi-user and Ubiquitous User Interfaces,
Funchal, Madeira, Spain.
Ishii, H., & Ullmer, B. (1997). Tangible bits: Towards seamless interfaces between people, bits,
and atoms. In CHI97: Proceedings of the SIGCHI
Conference on Human Factors in Computing
Systems. Atlanta, GA: ACM.
Kato, H., Billinghurst, M., Poupyrev, I., Tetsutani,
N., & Tachibana, K. (2001). Tangible augmented
reality for human computer interaction. In ProNagoya, Japan.
ceedings of Nicograph 2001,
Klinker, G., Dutoit, A., Bauer, M., Bayer, J., Novak,
V., & Matzke, D. (2002). Fata

Morgana: A presentation system for product design. In ISMAR 02:

Proceedings of the IEEE and ACM International
Symposium on Mixed and Augmented Reality,
Darmstadt, Germany.
Klinker, G., Stricker, D., & Reiners, D. (1999).
Augmented reality: A balancing act between high
quality and real-time constraints. In ISMR 99:
Proceedings of the 1st International Symposium on
Mixed Reality, Yokohama, Japan (pp. 325-346).
Kulas, C., Sandor, C., & Klinker, G. (2004). To
wards a development methodology for augmented
reality user interfaces. In MIXER 04: Proceed-

ings of the International Workshop Exploring

the Design and Engineering of Mixed Reality
Systems, Funchal, Madeira, Spain.
Lee, G. A., Nelles, C., Billinghurst, M., & Kim,
G. J. (2004). Immersive authoring of tangible
augmented reality applications. In ISMAR 04:
Proceedings of the IEEE and ACM International
Symposium on Mixed and Augmented Reality
(pp. 172-181). Arlington, VA: IEEE Computer
Society.
MacIntyre, B., Gandy, M., Dow, S., & Bolter, J. D.
(2004). Dart: A toolkit for rapid design exploration
of augmented reality experiences. In UIST 04:
Proceedings of the 17th Annual ACM Symposium
on User Interface Software and Technology, Santa
Fe, NM (pp. 197-206).
MacWilliams, A. (2005). A decentralized adaptive architecture for ubiquitous augmented reality systems. PhD thesis, Technische Universitt
Mnchen, Germany.
MacWilliams, A., Sandor, C., Wagner, M., Bauer,
M., Klinker, G., & Brgge, B. (2003). Herding
sheep: Live system development for distributed
augmented reality. In ISMAR 03: Proceedings
of the IEEE and ACM International Symposium
on Mixed and Augmented Reality, Tokyo, Japan
(pp. 123-132).
Novak, V., Sandor, C., & Klinker, G. (2004).
An
AR workbench for experimenting with attentive
user interfaces. In ISMAR 04: Proceedings of
IEEE and ACM International Symposium on
Mixed and Augmented Reality, Arlington, VA
(pp. 284-285).
Olwal, A., Benko, H., & Feiner, S. (2003). SenseShapes: Using statistical geometry for object selection in a multimodal augmented reality system.
In ISMAR 03: Proceedings of the 2nd IEEE and
ACM International Symposium on Mixed and
Augmented Reality (pp. 300-301). Washington,
DC: IEEE Computer Society.

643

Lessons Learned in Designing Ubiquitous Augmented Reality User Interfaces

Olwal, A., & Feiner, S. (2003). The flexible

pointerAn interaction technique for selection
in augmented and virtual reality. In UIST 03:
Proceedings of the 24th Annual ACM Symposium
on User Interface Software and Technology,
Vancouver, BC (pp. 81-82).
Olwal, A., & Feiner, S. (2004). Unit: Modular
development of distributed interaction techniques for highly interactive user interfaces. In
GRAPHITE 04: International Conference on
Computer Graphics and Interactive Techniques
(pp. 131-138). Singapore: ACM Press.
Ong, S., & Nee, A. (2004). Virtual and augmented
reality applications in manufacturing. London:
Springer Verlag.
Piekarski, W. (2002). Interactive 3D modelling in

thesis,
outdoor augmented reality worlds. PhD
University of South Australia.
Poupyrev, I., Tan, D. S., Billinghurst, M., Kato, H.,
Regenbrecht, H., & Tetsutani, N. (2001). Tiles:

A
mixed reality authoring interface. In INTERACT
01: The 7th Conference on Human-Computer
Interaction, Tokyo, Japan (pp. 334-341).
Pustka, D. (2003). Visualizing distributed systems
of dynamically cooperating services. Unpublished masters thesis, Technische Universitt
Mnchen.

Sandor, C. (2005). A software toolkit and authoring tools for user interfaces in ubiquitous augmented reality. PhD thesis, Technische Universitt
Mnchen, Germany.
Sandor, C., & Klinker, G. (2005). A rapid prototyping software infrastructure for user interfaces
in ubiquitous augmented reality. Personal Ubiquitous Computing, 9(3), 169-185.
Sandor, C., Olwal, A., Bell, B., & Feiner, S.
(2005). Immersive mixed-reality configuration of hybrid user interfaces. In ISMAR 05:
Proceedings of IEEE and ACM International
Symposium on Mixed and Augmented Reality,
Vienna, Austria.
Tnnis, M., Sandor, C., Klinker, G., Lange, C.,
& Bubb, H. (2005).
Experimental evaluation of
an augmented reality visualization for directing
a car drivers attention. In ISMAR 05: Proceedings of IEEE and ACM International Symposium on Mixed and Augmented Reality, Vienna,
Austria.
Vertegaal, R. (2003). Attentive user interfaces.
Communications of ACM, Special Issue on Attentive User Interfaces, 46(3).
Zauner, J., Haller, M., Brandl, A., & Hartmann,
W. (2003).
Authoring of a mixed reality assembly
instructor for hierarchical structures. In ISMAR
03: Proceedings of the IEEE and ACM International Symposium on Mixed and Augmented
Reality, Tokyo, Japan (pp. 237-246).

This work was previously published in Emerging Technologies of Augmented Reality: Interfaces and Design, edited by M.
Haller, B. Thomas, and M. Billinghurs, pp. 218-235, copyright 2007 by IGI Publishing, formerly known as Idea Group Publishing (an imprint of IGI Global).

644

645

Chapter 2.17

Social Network Structures in

Open Source Software
Development Teams
Yuan Long
Colorado State University-Pueblo, USA
Keng Siau
University of Nebraska-Lincoln, USA

Abstract

INTRODUCTION

Drawing on social network theories and previous

studies, this research examines the dynamics of
social network structures in open source software
(OSS) teams. Three projects were selected from
SourceForge.net in terms of their similarities
as well as their differences. Monthly data were
extracted from the bug tracking systems in order
to achieve a longitudinal view of the interaction
pattern of each project. Social network analysis
was used to generate the indices of social structure. The finding suggests that the interaction
pattern of OSS projects evolves from a single hub
at the beginning to a core/periphery model as the
projects move forward.

The information system development arena has

seen many revolutions and evolutions. We have
witnessed the movement from structured development to object-oriented (OO) development.
Modeling methods, such as data flow diagram
and entity relationship diagram, are facing new
OO modeling languages, such as the unified
modeling language (UML) (see Siau & Cao,
2001; Siau, Erickson, & Lee, 2005; Siau & Loo,
2006) and OO methodologies, such as unified
process (UP). The latest development includes
agile modeling (see Erickson, Lyytinen, & Siau,
2005), extreme programming, and OSS development. While many of these changes are related
to systems development paradigms, methodologies, methods, and techniques, the phenomenon
of OSS development entails a different structure
for software development teams.

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Social Network Structures in Open Source Software Development Teams

Unlike conventional software projects, the

participants of OSS projects are volunteers. They
are self-selected based on their interests and capability to contribute to the projects (Raymond,
2000). In addition, the developers of OSS projects
are distributed all around the world. They communicate and collaborate with each other through
the Internet, using e-mails or discussion boards.
Therefore, effective and efficient communication
and collaboration are critical to OSS success.
However, little empirical research has been
conducted to study the underlying interaction
pattern of OSS teams, especially the dynamics
of the social network structures in OSS development teams. To fill this gap, this study examines
the evolvement of social structure in OSS teams.
The study contributes to the enhancement of the
understanding of OSS development, and provides
foundation for future studies to analyze the antecedents and consequences of social networks
in the OSS context.
The remainder of the paper is structured as
follows. First, prior studies on social network
structures in OSS teams are reviewed. Second,
theories related to social structure and social
network theory are discussed. Third, the research
methodology is presented, and the research results
are reported. Next, discussions of the results, the
limitations, and the implications are provided.
The paper concludes with suggestions for future
research.

LITERATURE REVIEW
The phenomenon of OSS development has attracted considerable attention from both practitioners
and researchers in diverse fields, such as computer
science, social psychology, organization, and
management. Because of the multifaceted nature
of OSS, researchers have investigated OSS phenomenon from varied perspectives. For example,
focusing on technical perspective, researchers
studied issues such as OSS development method-

646

ology (e.g., Jrgensen, 2001) and coding quality

(e.g., Stamelos, Angelis, Oikonomu, & Bleris,
2002). Based on social psychology, researchers
investigated individual motivation (e.g., Hann,
Robert, & Slaughter, 2004), new developers (Von
Krogh, Spaeth, & Lakhani 2003), the social network (e.g., Madey, Freeh, & Tynan, 2002), and
the social structure (e.g., Crowston & Howison,
2005). In terms of organizational and managerial
perspective, researchers examined knowledge
innovation (e.g., Hemetsberger 2004; Lee & Cole
2003, Von Hippel & von Krogh, 2003) and the
governance mechanism (e.g., Sagers 2004).
An OSS development team is essentially a
virtual organization in which participants interact and collaborate with each other through the
Internet. Compared to conventional organizations,
the structure of virtual organizations is decentralized, flat, and nonhierarchical (Ahuja & Carley
1999). However, some researchers challenge the
belief (e.g., Crowston & Howison 2005; Gacek
& Arief, 2004;; Mockus, Fielding, & Herbsleb,
2000; Mockus, Fielding, & Herbsleb, 2002; Moon
& Sproull, 2000). They argue that the social
structure of OSS projects is hierarchical rather
than flat, like a tree (Gacek & Arief, 2004) or an
onion (Crowston & Howison, 2005). The social
structure of OSS teams directly influences the
collaboration and the decision-making process
and further affects the overall performance of the
teams as well as individuals perception of belonging and satisfaction. Therefore, one wonders what
form of social structure might be present in the
OSS development and what type of structure will
emergecentralized or decentralized, hierarchical or nonhierarchical, onion-like or tree-like, or
a combination of the above depending on certain
specific situations?
A social network, as stated by Krebs and
Holley (2004), is generally built in four phases,
each with its own distinct topology (as shown in
Figure 1).

Social Network Structures in Open Source Software Development Teams

Figure 1. Four phases of social structures (from Krebs and Holley 2004)

Stage 1. Scattered Clusters

Stage 3. Multi-Hub Small World Network

1.
2.
3.
4.

scattered clusters,
single hub-and-spoke,
multihub small-world network, and
core/periphery.

Most organizations start from isolated and

distributed clusters (Krebs & Holley, 2004). Then
an active leader emerges and takes responsibility
for building a network that will connect the separate clusters. However, this single-hub topology is
fragile. With more participants entering the group,
the leader changes his/her role to a facilitator and
helps to build multiple hubs, which is stage three.
The core/periphery model, the last stage, is the
most stable structure. In the core/periphery model,
the network core encompasses key group members
who are strongly connected to each other, while
the periphery contains members who are usually
weakly connected to each other as well as to the

Stage 2. Single Hub-and-Spoke

Stage 4. Core/Periphery Network

core members. With the general network building

phases in mind, one can argue that OSS projects
may follow the same four stages (i.e., scattered
clusters, single hub-and-spoke, multihub smallworld network, and core/periphery model). But is
that true for OSS projects? How does the social
structure of OSS teams evolve over time?
Our research addresses the following two
questions:
1.
2.

What is the social structure of OSS

teams?
How does the social structure evolve over
time?

647

Social Network Structures in Open Source Software Development Teams

THEORETICAL FOUNDATION
Social Structure and Social
Interaction
Social structure, as suggested by Schaefer and
Lamm (1998), refers to the way in which society
is organized into predictable relationships. Social
structure can be considered in terms of three
aspectsactors, their actions, and their interactions. The social actor is a relatively static concept
addressing issues such as roles, positions, and
statuses. Individual actors are embedded in the
social environment and, therefore, their actions
are largely influenced by the connections between
each other. Social interaction is generally regarded
as the way in which people respond to one another.
These interaction patterns are to some extent independent of individuals. They exert a force that
shapes both behavior (i.e., actions) and identity
(i.e., actors) (Schaefer & Lamm, 1998).
Research on social interaction focuses on how
individuals actually communicate with each other
in group settings. These studies address issues
such as the interaction patterns, the underlying
rules guiding interaction, the reasons accounting
for the way people interact, and the impacts of
the interaction patterns on the individual behavior
and the group performance. These issues begin by
questioning what might be the interaction pattern
in a specific social setting and that addresses our
research questionunderstanding social interaction of OSS project teams.

Social Network Theory

Social network theory focuses on studying actors as well as their relationships in specific
social settings. Network theory is analogous to
systems theory and complexity theory. It is an
interdisciplinary theory stemming from multiple
traditional fields, including psychology, which
addresses individuals perception of social

648

structure; anthropology, which emphasizes social

relationships; and mathematics, which provides
algorithms (Scott, 2000).
Based on the view of social network, the world
is composed of actors (also called nodes) and ties
between them. The ties can represent either a specific relationship (such as friendship and kinship)
between a pair of actors or define a particular action which an actor performs. Different kinds of
ties specify different networks and are typically
assumed to function differently. For example,
the ties in a family network are distinctive from
those in a working network, and the centrality
in the who loves whom network obviously has
different meaning than the centrality in the who
hates whom network.
Social network theory is based on the intuitive notion that the social interaction patterns are
essential to the individuals who reflect them.
Network theorists believe that how individuals
behave largely depends on how they interact with
others and how they are tied to a social network.
Furthermore, besides individual behavior, network theorists believe that the success or failure of
societies and organizations often depends on the
internal interaction pattern (Freeman, 2004).
Besides the theoretical essence, social network theory is also characterized as a distinctive
methodology encompassing techniques for data
collection, statistical analysis, and visual representation. This approach is usually called social
network analysis and will be discussed in the
research methodology section. This paper draws
on the social network theory to study the interaction pattern of OSS development project.

RESEARCH METHODOLOGY
Social Network Analysis
Social network analysis is used in our study to
investigate the interaction pattern of the OSS
development process. Social network analysis

Social Network Structures in Open Source Software Development Teams

focuses on uncovering the interaction pattern

of interdependent individuals (Freeman, 2004).
Through a structural analysis of a social network
diagram, a map depicting actors as well as the
ties that connect them, social network analysis
can reveal the patterns of relationships and the
relative position of individuals in a specific social setting. This approach has been effectively
used in organizational research, social support,
mental health, and the diffusion of information
(Freeman, 2004).
Social network analysis is used in our study
for two primary reasons. First, the purpose of
social network analysis fits our research objective. Social network analysis aims to analyze the
relationship among a set of actors instead of their
internal attributes. Our research aims to reveal
the interaction pattern of OSS project teams.
Therefore, social network analysis is helpful in
answering our research questions.
Second, the rich interactive data extracted
from OSS projects presents a gold mine for
social network analysis. Social network analysis
is grounded in the systematic analysis of empirical
data. However, there is usually a lack of convenient and objective resources from which to draw
the links (i.e., relationships) among actors. Most
OSS projects have online mailing lists, forums,
and tracking systems that are open to public, thus
providing a rich set of longitudinal data. Based on
these public data, researchers are able to capture
input data sets for social network analysis.

Longitudinal Data
Because we are interested in studying how the
interaction pattern of OSS projects evolves over
time, cross-sectional observations of interaction networks are not sufficient. Cross-sectional
observations of social networks are snapshots of
interactions at a point in time and cannot provide
traceable history, thus limiting the usefulness of
the results. On the other hand, longitudinal observations offer more promise for understanding

the social network structure and its evolvement.

In this study, we extracted longitudinal data on
OSS projects.

Case Selection
OSS projects were selected from the SourceForge1,
which is the worlds largest Web site hosting OSS
projects. SourceForge provides free tools and services to facilitate OSS development. At the time of
the study, it hosted a total of 99,730 OSS projects
and involved 1,066,589 registered users (This data
was retrieved on May 4, 2005). Although a few
big OSS projects have their own Web sites (such
as Linux), SourceForge serves as the most popular
data resource for OSS researchers.
Following the idea of theoretical sampling
(Glaser & Strauss, 1967), three OSS projects
were selected from SourceForge in terms of their
similarities and differences. Theoretical sampling
requires theoretical relevance and purposes
(Orlikowski, 1993). In terms of relevance, the
selection ensures that the interaction pattern of
OSS projects over time is kept similar. Therefore,
the projects that are selected have to satisfy two
requirements. First, the projects must have considerable interaction among members during the
development process. All three projects had more
than 10 developers, and the number of bugs reported was more than 1,000. Second, since we are
interested in the interaction over time, the projects
must have a relatively long life. In our case, all
three projects were at least three years old.
In addition to similarities, differences are
sought among cases because the study aims
to study interaction patterns of various OSS
projects. Therefore, the three projects differ on
several project characteristics, such as project
size, project type, and intended audience. These
differences enable us to make useful contrasts
during data analysis.
The Table 1 summarizes the three projects
with a brief description.

649

Social Network Structures in Open Source Software Development Teams

Table 1. Summary of three projects

Net-SNMP
Net-SNMP allows
system and network
managers to monitor
and manage hosts and
network devices.

Description

Similarities

Differences

3.

650

Compiere is a smart
ERP+CRM solution covering all major business
areasespecially for
small-medium enterprises.

JBoss is a leading open

source Java application
server. After Linux and
Apache, it is the third
major open source
project to receive
widespread adoption
by corporate IT.

1,361

1,695

2,296

Development
duration (more
than 3 years)

55 months (registered
on 10/2000)

47 months (registered on
6/2001)

50 months (registered
on 3/2001)

Software type

Internet, network
management

Enterprise: ERP+CRM

J2EE-based middleware

Group size
(number of
developers)

Small (14)

Median (44)

Large (75)

Intended audience

Developers, system
administrators

Business

Developers, system
administrators

Social network analysis can be divided into the

following three stages (Borgatti, 2002).

2.

J-boss

Bug reports
(more than
1,000 bugs)

Data Collection and Analysis

1.

Compiere ERP + CRM

Data collection. In this stage, researchers

collect data, using surveys and questionnaires, or from documents and other data
resources, and generate input data sets for
social network analysis.
Statistical analysis. Based on mathematics
algorithms, this stage generates network
indices concerning group structure (such
as centralization and density) as well as
individual cohesion (such as centrality and
bridges).
Visual representation. This stage employs
network diagrams to show the interaction

structure as well as the position of specific

actors.
First is the data collection. The data were collected in April 2005 from SourceForge.net. Data
were extracted from the bug tracking system of
each project. We chose the bug tracking system
as the primary data resource for three reasons.
First, open source software is characterized as
peer review of open codes. Raymond (1998)
proposed the Linus law in his well-known
essay The Cathedral and the Bazaar Given
enough eyeballs, all bugs are shallow. Therefore,
the bug system can be viewed as the representative of open source spirit. Second, compared to
other development activities, such as patch posting and feature request, the bug-fixing process is
the most active procedure to illustrate the close

Social Network Structures in Open Source Software Development Teams

Figure 2. An example of the social matrix for an OSS project

The total nuber of actors
dl N=11 format=fullmatrix
SF IDs of all the actors

Lables:
cmsavage dteixeira jcbowman jsber-bnl m-a rapr rtprince sf-robot svenn xbursam ydirson

Data:
0 0
0 0
5 1
0 0
0 0
0 0
0 1
0 0
0 0
0 0
0 0

0
0
0
3
0
1
0
1
0
1
3

0
0
2
0
0
0
0
0
0
0
0

0
0
1
0
0
0
0
0
0
0
0

0
0
0
0
0
0
0
0
0
0
0

0
0
0
0
0
1
0
0
0
0
0

0
0
0
0
0
0
0
0
0
0
2

0
0
1
0
0
0
0
0
0
0
0

collaboration between developers and users as

well as among developers themselves. Finally,
the bug tracking system provides rich data that
record the interactive process.
A Web spider program, which is based on
the work of Crowston and Howison (2005) with
necessary revision, was used to download the bug
tracking Web pages from the project Web site.
After that, a Web parsing program was developed
to analyze the Web pages. The interaction data
was extracted from the bug tracking Web pages
month-by-month, starting from the date the project
was registered until the date the data was downloaded for this study. The output of this stage is
a social matrix describing the interaction among
users. Figure 2 shows an example of such a social
matrix for an OSS project. In the matrix, each row
or column represents a distinctive participant,
which is identified by a unique SourceForge user
identity. The values of cells indicate the degree of
the interaction between each pair of participants,

0
0
1
0
0
0
0
0
0
0
0

0
0
6
0
0
0
0
0
0
0
0

Jcbowman replied to
ydirson 6 times

which is counted by the amounts of messages that

participant A (i.e., row A) replied to participant
B (i.e., column B).
Second is the statistical analysis. Our study focuses on two important and distinctive properties
of network structuregroup centralization and
core/periphery fitness. Ucinet, which was developed by Borgatti, Everett, and Freeman (2002),
was used to calculate these two properties.
Group centralization, as suggested by Wasserman and Faust (1994), refers to the extent to
which a network revolves around a single center.
A typical case of centralized structure is a star
network. Group centralization can be viewed as
a rough measure of inequity between individual
actors, and the variability and dispersion of the
interaction pattern.
The other property is core/periphery fitness. It
measures the extent to which the network is close
to a perfect core/periphery structure. The core/periphery structure depicts a dense, connected group

651

Social Network Structures in Open Source Software Development Teams

Table 2. Three snapshots for each project

Group centralization
(%)

Core/periphery
fitness

Density

Net-SNMP

Compiere

JBoss

st

1.

9.420

15.624

4.931

2nd.

3.071

2.294

4.45

3rd.

2.316

1.288

4.12

1st.

0.674

0.774

0.485

2 .

0.654

0.796

0.477

nd

3 .

0.651

0.765

0.501

1st.

0.0235

0.0584

0.0073

2nd.

0.0109

0.0610

0.0039

rd

3 .

0.0072

0.0571

0.0026

1st.

2.546

2.711

3.438

2nd.

2.794

2.302

3.281

rd

Average distance

Distance-based
cohesion

rd

3 .

2.917

2.278

3.239

1st.

0.181

0.198

0.118

2nd.

0.143

0.253

0.147

3rd.

0.141

0.279

0.136

surrounded by a sparse, unconnected periphery.

The opposite structure is clique, which represents
a structure of multiple subgroups, each with its
own core and peripheries (Borgatti, 2002).
Finally is the visual representation. We used
Ucinet (Borgatti et al., 2002) to draw the interaction networks for each of the three projects.

RESEARCH RESULTS
Snapshots of the Three Projects
Monthly data were extracted from the bug tracking system of each project. To illustrate the trend
of interaction pattern, we provide three snapshots
for each project (see Figures 3-5)2.
Table 2 summarizes the relevant network
characteristics of each project. In addition to the

652

group centralization and core/periphery fitness,

we also report other network characteristics, such
as density, average distance, and distance-based
cohesion. Density depicts how close the network
looks, and it is a recommended measure of group
cohesion (Blau, 1977; Wasserman & Faust 1994).
The value of density ranges from 0 to 1. Average
distance refers to average distance between all
pairs of nodes (Borgatti, 2002). Distance-based
cohesion takes on values between 0 and 1the
larger the values, the greater the cohesiveness.
Looking at the statistical results and the network plots, we can observe the following.
First, the evolvement of interaction patterns
of the three projects reveals a general trend. As
shown in the network plots (i.e., Figures 3-5), the
interaction pattern develops from a centralized
one with a single (sometimes dual) hub with
several distributed nodes, to a core/periphery

Social Network Structures in Open Source Software Development Teams

Figure 3. Interaction patterns of Net-SNMP

April 2002

October 2003

April 2005

Figure 4. Interaction patterns of compiere CRM+ERP

Sep. 2002

Jan. 2004

April 2005

structure that has a core (a group of core developers) together with several hangers-on (periphery).
Intense interactions exist within the core (among
several core developers) and between each core
member and his/her periphery. However, only

loose relationships exist among peripheries. This

pattern suggests a layer structure (i.e., core with
its periphery) instead of a complete flat one with
equal positions across all the members.

653

Social Network Structures in Open Source Software Development Teams

Figure 5. Interaction patterns of JBoss

July 2002

November 2003

April 2005

Second, although the interaction patterns of

the three projects share some commonalities, their
exact shapes are different. The shape of Net-SNMP
(as shown in Figure 3) is more like a typical core/
periphery compared to the other two. Compiere
(as shown in Figure 4) keeps two cores, and the
shape looks like a dumbbell. Jboss (as shown in
Figure 5), which is the largest project among the
three, maintains a more complex structure that
shows multiple layers instead of just one core with
the rest as peripheries (e.g., Net-SNMP)
Third, as time goes by, the group centralization decreases across the three projects, showing
a trend that moves from a centralized structure to
a decentralized structure irrespective of project
sizes (The three projects with different project
sizes are shown in Table 1), project types, and
intended audience.
Fourth, the indices of core/periphery fitness
of each project fluctuate slightly but maintain a

654

relatively high value (larger than 0.5 on average).

However, no observable trend was found across
projects.
Fifth, since each project has a relatively large
group (i.e., more than 100 actors including all the
registered users), the values of density are relatively low with little variation. Therefore, density
is not appropriate for comparing the projects.
From the snapshots, we observed the following trend. First, the OSS interaction network
evolves into a core/periphery structure. Second,
group centralization decreases over time. Third,
core/periphery fitness stays relatively stable. To
verify our observations, we used longitudinal
data generated from the bug tracking systems
to analyze the evolvement of interaction pattern
(discussed in the following section).

Social Network Structures in Open Source Software Development Teams

Table 3. Group centralization and core/periphery fitness based on longitudinal data

Group centralization

Core/periphery fitness

NetSNMP

Compiere

JBoss

Group Centralization and

Core/Periphery Fitness
Table 3 shows the values of both group centralization and core/periphery fitness over time based on
the monthly interaction data. For each figure, the
Y-axis indicates the social structure indices (i.e.,
group centralization or core/periphery fitness), and
the X-axis reflects the time dimension.
Two primary observations can be made based
on the statistical analysis.
First, the group centralization shows a decreasing trend across the three projects. This observation indicates that as OSS projects progress, the
social network structure evolves from centralized
to decentralized and then stabilizes. Also, the
three figures suggest no substantial differences
in the trend among the three projects.

Second, the core/periphery index is maintained

at a relatively stable level for each project over
time. In addition, the average fitness value stays
relatively high for each project (larger than 0.5),
indicating a closeness to a perfect core/periphery
structure.
Besides a holistic view of network structure
for OSS projects, the results also reveal other
interesting findings. For example, by examining
the core members over time, we found a relatively
stable core for each project. The cores are usually project developers and administrators. This
observation further demonstrates the existence of
strong and stable core as well as loose hangers-on
in OSS projects.

655

Social Network Structures in Open Source Software Development Teams

DISCUSSION
This research uses the longitudinal data of three
OSS projects selected from SourceForge to study
the social network structures of OSS teams. The
purpose of this study is to investigate the evolvement of interaction patterns of OSS project teams.
The research results suggest a decrease of group
centralization over time and a tendency of core/
periphery structure in OSS project teams.
The network plots (as shown in Figures 3-5)
indicate a layer structure instead of a flat one as
suggested by earlier literature. The interaction
pattern evolves from a single hub to a core/periphery structure. As the number of participants
increases, a core with only one person (who may
be the starter/initiator of the project) cannot satisfy
the increasing requirements of development and
communication. Therefore, other developers or
active users join the core to serve as key members of the project. This results in a more stable
structure, and the project is less dependent on a
single leader.
With the growth of a software project, more
people are attracted to the project. The original
leader may not be able to solve all the technical
problems encountered in the development process. Each key member has his/her specialty, is
responsible for solving relevant problems, and
has his/her own periphery in the network plot.
Although there are multiple peripheries in the
project, collaboration among key members in the
project is vital. This phenomenon of distribution
and collaboration can be viewed as a critical
success factor of OSS development. And the
evolvement is vividly demonstrated in our social
network analysis.
In a way, the social structure of OSS projects is
both centralized and decentralized. On one hand,
it is centralized in the sense that there is a core that
consists of key members. These key members are
responsible for various issues encountered during
the development process. On the other hand, it
is decentralized in the sense that the decision or

656

communication core is not concentrated on one

or two members but a group of key members.
Like any other research, this research has its
share of limitations. First, the cases were only
selected from SourceForge.net. Although SourceForge is the worlds largest Web site hosting open
source software, there are also some other similar
Web sites. Therefore, the total number of OSS
projects in SourceForge cannot be viewed as the
whole population. However, as we argued before,
SourceForge is probably the best data collection
site for this research.
Second, the bug tracking system was chosen as
our data resource. The selection of bug tracking
system as our research setting and data resource
may have had an effect on the outcome and results.
Besides the bug tracking forum, there are other
forums that also provide space for participants
to communicate with one another, such as mailing lists and feature requests. However, as we
highlighted earlier, the bug systems are the most
active forum, providing rich interaction data. The
bug tracking systems also represent the spirit of
open source software development. Examining
the interaction data from other forums can be one
of our research extensions in the future.
Third, because our research objective is to
investigate interaction pattern, we chose projects
that have a relatively large number of developers,
a large number of bug reports, and relatively
long history. Although we tried to involve different types of projects (i.e., different project
sizes, project types, and intended audience),
these three cases may not be representatives of
OSS projects, for example, small projects with
only one or two developers and few interactions.
Increasing the sample size and including various
types of OSS projects is one of our future research
directions.

Social Network Structures in Open Source Software Development Teams

IMPLICATIONS AND
CONCLUSION
This paper examines the interaction patterns of
OSS teams. The research findings suggest that
the interaction structure starts from a single hub
and evolves to a core/periphery model. We argue
that the social structure of OSS teams is both
centralized and decentralized. It is centralized in
the sense that there exists a relatively stable core
that consists of a group of key developers. It is
also decentralized because of distributed decision
making among key developers and the broad collaboration between developers and users as well
as among developers themselves.
The paper presents the evolvement of the social
structure of OSS projects from a longitudinal
perspective. It also provides empirical evidence
of the change of interaction patterns from a
single hub to a core/periphery model. Moreover,
the paper utilizes social network analysis as the
research method. This approach has been shown
in this research as an effective tool in analyzing
the social structure in OSS teams.
Social structure is an important variable for
understanding social phenomenon. Open source
software, with its open and unique nature, attracts researchers to ask a series of questions.
For example, how do participants of OSS projects
interact and collaborate with each other? What
factors facilitate the interaction and the collaboration? And further, how does the collaboration
affect project performance of OSS teams? Social
network analysis is a good approach to investigate these questions. This research represents a
pioneering effort in this direction.

References
Ahuja, M., & Carley, K. (1999). Network structure
in virtual organizations. Organization Science,
10(6), 741-747.

Blau, P. M. (1977). Inequity and heterogeneity.

New York: Free Press.
Borgatti, S. (2002). Basic social network concepts.
Retrieved from http://www.analytictech.com/networks/basic%20concepts%202002.pdf
Borgatti, S. P., Everett, M. G., & Freeman, L.
C. (2002). Ucinet for Windows: Software for
social network analysis. Harvard, MA: Analytic
Technologies.
Crowston, K., & Howison, J. (2005). The social
structure of free and open source software development. First Monday, 10(2).
Erickson, J., Lyytinen, K., & Siau, K. (2005).
Agile
modeling, agile software development, and extreme programming: The state of research. Journal of Database Management, 16(4), 88-100.
Freeman, L. C. (2004). The development of social
network analysis: A study in the sociology of science. Vancouver, Canada: Empirical Press.
Gacek, C., & Arief, B. (2004). The many meanings
of open source. IEEE Software, 21(1), 34-40.
Glaser, B. G., & Strauss, Anselm L., (1967). The
Discovery of Grounded Theory: Strategies for
Qualitative Research, Chicago, Aldine Publishing Company
Hann, H., Robert, J., & Slaughter, S. (2004). Why
developers participate in open source software
projects: An empirical investigation. In TwentyFifth International Conference on Information
Systems, (pp. 821-830). Washington, DC: .
Hemetsberger, A. (2004). Sharing and creating
knowledge in open-source communities: The
case of KDE. The Fifth European Conference
on Organizational Knowledge, Learning, and
Capabilities in Innsbruck, Austria.
Jrgensen, N. (2001). Putting it all in a trunk:
Incremental software development in the freeBSD open source project. Information Systems
Journal, 11, 321-336.

657

Social Network Structures in Open Source Software Development Teams

Krebs, V., & Holley, J. (2004). Building sustainable communities through network building.
Retrieved from http://www.orgnet.com/BuildingNetworks.pdf
Lee, G. K., & Cole, R. E. (2003). From a firm-based
to a community-based model of knowledge creation: The case of the Linux kernel development.
Organization Science, 14(6), 633-649.
Madey, G., Freeh, V., & Tynan R. (2002). The
open source software development phenomenon:
An analysis based on social network theory (AMCIS2002). Dallas, TX.
Mockus, A., Fielding, R. T., & Herbsleb, J. D.
(2000). A case study of open source software
development: The Apache server. ICSE 2000.
Mockus, A., Fielding, R. T., & Herbsleb, J. D.
(2002). Two case studies of open source software
development: Apache and Mozilla. ACM Transactions on Software Engineering and Methodology,
11(3), 309346.
Moon, J. Y., & Sproull, L. (2000). Essence of
distributed work: The case of Linux kernel. First
Monday, 5(11).
Orlikowski, W. J. (1993). CASE tools as organizational change: investigating incremental and
radical changes in systems development. MIS
Quarterly, 17(3), 309-340.
Raymond, E. S. (1998). The cathedral and the
bazaar. First Monday, 3(3), Retrieved January ,
2005, from http://www.catb.org/~esr/writings/cathedral-bazaar/cathedral-bazaar/
Sagers, G. W. (2004). The influence of network
governance factors on success in open source
software development projects. In Twenty-Fifth
International Conference on Information Systems
(pp. 427-438). Washington, DC:
Schaefer, R. T., & Lamm, R. P. (1998). Sociology
(6th ed.). McGraw-Hill.

Scott, J. (2000). Social network analysis. A handbook (2nd ed.). London: SAGE Publications.
Siau, K., & Cao, Q. (2001). Unified modeling
languageA complexity analysis. Journal of
Database Management, 12(1), 26-34.
Siau, K., Erickson, J., & Lee, L. (2005). Theoretical versus practical complexity: The case of
UML. Journal of Database Management, 16(3),
40-57.
Siau, K., & Loo, P. (2006). Identifying difficulties
in learning UML. Information Systems Management, 23(3), 43-51.
Stamelos, I., Angelis, L., Oikonomu, A., & Bleris,
G. L. (2002). Code quality analysis in open-source
software development. Information Systems
Journal, 12(1), 43-60.
Von Hippel, E., & Von Krogh, G. (2003). Open
source software and the private-collective innovation model: Issues for organization science.
Organization Science, 14, 209-223.
Von Krogh, G., Spaeth, S., & Lakhani, K. R.
(2003). Community, joining, and specialization
in open source software innovation: A case study.
Research Policy, 32(7), 1217-1241.
Wasserman, S., & Faust, K., (1994). Social
Network Analysis: Methods and Applications.
New York: Cambridge University Press.

Endnote

The Web address for SourceForge is www.

sourceforge.net.
The three time stamps for Net-SNMP are
4/2002, 10/2003 and 4/2005; for Compiere
are 9/2002, 1/2004 and 4/2005; and for Jboss
are 7/2002, 11/2003 and 4/2005.

This work was previously published in Journal of Database Management 18(2), edited by K. Siau, pp. 25-40, copyright 2007
by IGI Publishing, formerly known as Idea Group Publishing (an imprint of IGI Global).
658

659

Chapter 2.18

A Graphical User Interface (GUI)

Testing Methodology
Zafar Singhera
ZAF Consulting, USA
Ellis Horowitz
University of Southern California, USA
Abad Shah
R & D Center of Computer Science, Pakistan

ABSTRACT
Software testing in general and graphical user
interface (GUI) testing in particular is one of the
major challenges in the lifecycle of any software
system. GUI testing is inherently more difficult
than the traditional and command-line interface
testing. Some of the factors that make GUI testing
different from the traditional software testing and
significantly more difficult are: a large number of
objects, different look and feel of objects, many
parameters associated with each object, progressive disclosure, complex inputs from multiple
sources, and graphical outputs. The existing testing techniques for the creation and management of
test suites need to be adapted/enhanced for GUIs,
and new testing techniques are desired to make
the creation and management of test suites more
efficient and effective. In this article, a methodol-

ogy is proposed to create test suites for a GUI.

The proposed methodology organizes the testing
activity into various levels. The tests created
at a particular level can be reused at higher
levels. This methodology extends the notion
of modularity and reusability to the testing
phase. The organization and management of
the created test suites resembles closely to the
structure of the GUI under test.

Introduction
Graphical user interfaces (GUl) are an important
part of any end-user software application today
and can consume significant design, development, and testing activities. As much as half
of the source code of a typical user-interaction
intensive application can be related to user inter-

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

A Graphical User Interface (GUI) Testing Methodology

faces (Harold, Gupta, & Soffa, 1993; Horowitz

& Singhera, 1993). GUIs provide an easier way
of using various functions of the application by
organizing them in a hierarchy of options and
presenting only the options which make sense
in the current working context. GUIs help users
concentrate on the problem instead of putting
efforts in remembering all the options provided
by the software application that is being used to
solve the problem, or searching for the right option from a huge list of options provided by the
application. Graphical user interfaces organize
the standard user actions and working paradigms
into various components that are presented
graphically to the user during various usage and
application contexts. GUIs enhance the usability
of an application significantly. However it also
makes application development, testing and
maintenance significantly more difficult (Myers, 1993; Wittel & Lewis,1991). The nature of
GUI applications, their asynchronous mode of
operation, nontraditional input and output, and
hierarchical structure for user interaction make
their testing significantly different and difficult
from the traditional software testing.
Functional and regression testing of graphical user interfaces is significantly more complex
than testing of traditional non-GUI applications
because of the additional complexities mentioned
in the previous paragraph. A number of commercial tools, like Mercury Interactives WinRunner,
XRunner and Segue Softwares SilkPerformer,
are used in the industry to test graphical user
interfaces. These tools provide capture/replay
capabilities to test a graphical user interface.
Although functionality provided by these tools
is sufficient for typical recored/replay scenarios
but they lack an underlying model that can provide more information about the test coverage
or to determine the quality of the user interface
from a particular functional or implementation
perspective. These tools also do not provide a
framework that assists in organized and modular

660

testing. The methodology presented in this article

uses user interface graphs (UIG) as a framework
for organization of test scripts, generation of
modular test suites, and coverage analysis of a
test execution.
In this article, we propose a methodology for
regression testing of graphical user interfaces,
with and without a formal specification of the application under test. The remainder of this article is
organized as follows: Section 2 highlights some of
the best practices and recommendations that help
in testing a GUI application in an organized fashion, improve efficiency and effectiveness of testing, reduces possibility of errors, and minimizes
repeated work. Section 3 describes the major steps
of the proposed methodology. It also introduces
a sample X application, called Xman, which is
used to demonstrate the effectiveness of the suggested strategy. Section 4 demonstrates the testing
methodology when formal specifications of the
application under test are not available. Section 5
illustrates the proposed testing methodology when
the formal specifications of the application under
test are available. This section also describes the
way statistics are collected during a testing activity
and how those can be used to improve the quality
of the testing. Section 6 points out the situations
when a modification to the application under test
might require tuning or recapturing of some of
the test scripts. Section 7 concludes the article
by summarizing our contribution and providing
hints about the future related work.

GUI Testing:
Best Practices and
Recommendations
In this section, we highlight some of the sought
features, well-knows best practices and recommendations for planning a testing activity for a
graphical user interface.

A Graphical User Interface (GUI) Testing Methodology

Every element of the GUI should be considered as an object uniquely identifiable by a

name. The objects should have a well-defined
set of parameters and their response to the
outside events should also be well defined.
The testing activity should be planned
carefully around a formal model of the
application under test. This model should
be powerful enough to provide automatic
test generation and coverage analysis.
Testing of a GUI should be performed in
a layered fashion. The list of objects to be
tested at a particular level, is either built
dynamically while testing at lower levels
or from the specifications of the application
under test. The list of objects for the lowest
level is the basic widget, supported by the
underlying toolkit. While testing the highest
level, it considers the entire application as a
single object. The decision about the number
of testing levels and the qualifying criteria
for a particular testing level must be made
before creating any tests.
The tests should be organized as a hierarchy
of scripts, that is, files containing commands
to simulate user actions and verify results.
This hierarchy should closely correspond to
the object hierarchy of the application under
test. Each directory in the hierarchy holds
scripts that are related to a particular object
and its descendents. The individual scripts
should be as small as possible and should
test one particular feature of each object.
However, if the features are related and
simple, then they can be grouped together
in the same script.
Each script should begin with a cleat and
precise description of the intended purpose
of the script and the state of the application
required for its proper execution. A script
should be divided into three sections. The
first section of the script builds the environ
ment required to test the particular feature of
the application, the script is intended for. The

second section of the script tests the intended

feature of the application being tested by the
script. The third section restores the state
of the AUT and the operating environment,
to a point that existed before entering the
script.
A script should be created in such a way
that some or all the sections of the script
can be executed by calling the script from
another script. It provides reusability feature
in testing also.
Instead of manually capturing or replaying
the test scripts, a tool should be used to
perform these functions automatically and
verify the behavior of the application under
test. The tool should be capable of addressing
an object in the GUI by its symbolic name,
instead of its location, dimensions or any
other contextual information.
The data for the result verification should be
captured in terms of object attributes when
possible and only those attributes should
be captured which are critical to verify the
functions of the application, being tested by
the current script. If image comparisons are
unavoidable, then the images should be captured with reference to the smallest enclosing
object and area of the captured images should
not be more than absolutely required. The
number of verifications should also be kept
to an absolute minimum especially when
image comparisons are involved.
The script commands to simulate user
actions during the replay and the data for
verification of the AUT behavior should be
kept separately. This separation is required
because the verification data might change
depending on the environment while the
script commands should be independent of
the environment and should be valid across
multiple platforms. If script commands and
verification data are stored separately, then
it is easier to port a test suite across multiple platforms. In fact, a good tool should

661

A Graphical User Interface (GUI) Testing Methodology

automatically perform the porting from one

hardware platform to the other.

Proposed Methodology for GUI

Testing
This section proposes a methodology for the
testing of a graphical user interface (GUI). This
proposed methodology is suitable particularly
when one has a tool similar to Xtester (Horowitz
& Singhera, 1993). It follows the recommendations
provided in the previous section. The methodology
works in the following two scenarios:
i.
ii.

Testing without formal specifications/model

of the application under test
Testing with a formal model of the application.

Both these scenarios are described in the following subsections.

Testing without a Formal Model

Creation of formal specifications of a GUI application for its testing purposes is a difficult task and
requires a significant amount of effort. It is also not
feasible to invest resources in creating the formal
specifications of the application; hence the testing
has to be performed without it. The best thing that
can be done in such a situation is to incrementally
build a test hierarchy for the application, by capturing-user sessions in an organized way. Automatic
test generation or coverage analysis is not possible
without formal specifications of the application under
test. The major steps of the proposed methodology
to test an application without a specification are
given below:

Step 1: Initialization
Make basic decisions about the testing activity.
Some of the most important decisions, which must
be taken at this point, are:

Figure 1. Strategy for testing without specifications

Display the_ object on the screen and verify its appearance;
If the object has sub-objects

Add its immediate sub-objects at the top of the list;

If the object qualifies for a higher testing level

Add it to the list of objects for the higher level;
Send expected events to the object and verify its response;
If a new object appears in response to the event
if the new object is not listed as a tested object

Add it to the end of the list;

Pop down the object from the screen;

Mark the object in the list as tested;

662

A Graphical User Interface (GUI) Testing Methodology

The number of testing levels and criteria to

build list of objects for a particular level.
Initialize a list of objects for each level that
holds the names and some information
about the objects of the user interface. The
information includes the way the object can
be mapped on the screen, the mechanism to
unmap it from the screen, and if it has been
tested or not.
The location of the test suite and its organization.
The application resources that will be
used during this testing activity should be
listed.

Step 2: Building the Initial Object List

Go through the documentation of the application
under test and find all the top-level windows,
which might appear as starting windows of the
application. These windows and, their related
information is added to the list of objects for the
first testing level and marked as tested.

Step 3: Building Test Suite

Take the first object from the top of the object list,
which has not been tested, and create test scripts
for it. The procedure for creating test scripts for a
particular object is given in Figure 1. The sub-objects
of the object under test are added to the list of objects
by scanning the object from left to right and top to
bottom. Keep on taking objects from the top of the
object list and testing them until all the objects in
the list are marked as tested. When the list associated with a particular level has no untested object,
start testing objects from the list, associated with
the next higher level. This process continues until
all the levels are tested.

Step 4: Creating Script Drivers

Write higher level scripts for all the top level
windows for any other complex objects, each
of the testing levels and for the entire test suite.
These scripts will replay all the scripts related to
the object and its descendents. The highest-level
script driver should replay each and every script
in the suite.

Step 5: Testing the Test Suite

Make sure that all the scripts and script drivers
work properly and cover all the features of the
application, which needs to be tested. One cannot
do much automatically to determine the quality
of a test suite, in the absence of a formal model
of the application under test. After the creation of
the test suite, run the highest-level script driver
to verify that all the scripts in the suite are capable of properly replaying and trying to match
the features covered by the test suite with those
included in the test requirements or application
documentation

Testing with a Formal Model

The strategy to build test scripts without a formal
specification, discussed in the previous subsection,
puts a lot of responsibility on the person creating
those scripts. The strategy also requires that the
application under test should be running reliably
before the capturing of script is even started. The
scripts created without any formal specification
are also vulnerable to any modification in the
application, which affects its window hierarchy.
It requires that after making any changes to the
application under test, the affected scripts should
be located manually and recaptured or tuned to
offset the modification in the application. It is also
not possible to create test scripts automatically or
to get a coverage measure after running a set of
test suites. To overcome these drawbacks and get
access to advanced features like automatic test
generation and coverage analysis, one has to invest

663

A Graphical User Interface (GUI) Testing Methodology

some effort to formally specify the application

under test. This section provides a methodology
to test an application when its formal specification is provided or resources are available to
build such a specification. The following are the
major steps of the testing methodology when a
formal specification of an application under test
is available.

Step1: Building the Model

Step 3: Build Object Lists

Build a list of objects for each testing level. After
building the formal model, it is possible to build
the object lists for all testing levels. The procedure
for building those lists is to start from the root
of the UIG and perform a post-order walk of the
object hierarchy. Add each visited node to the
object lists associated with the levels, for which
it qualifies.

Build a user interface graph of an application

under test. When resources permit, the very first
step in testing the application should be to build
a formal model of the application under test.
XTester provides such a formal model, called user
interface graph (UIG). UIG provides information
about the object hierarchy of the application. It
also provides information about the nature of a
particular object and the effects of an event in an
object to the other objects in the user interface.
A UIG can be built manually by creating a user
interface description language (UIDL) file, or it
can be created semi-automatically by steering
through the application under test and filling in
the missing information about the objects (see
Horowitz & Singhera, 1993) for more details on
syntax of UIDL and UIG).

Step 4: Building Test Suite

Step 2: Initialization

Write higher level scripts for all the top level

windows or any other complex objects, each of
the testing levels and for the entire test suite.
These scripts will replay all the scripts related
to the object and its descendents. A highest-level
script driver should replay each and every script
in the suite. These scripts can also be created
automatically.

Make basic decisions about the testing activity.

Some of the most important decisions, which must
be taken at this point, are listed as follows:

664

The number of testing levels and qualifying

criteria for each testing level.
The location of the test suite and its organization.
The application resources that will be used
during this testing activity.

The strategy for capturing scripts without any

formal specification, which has been discussed
in the previous section, can also be used for
capturing scripts when the application has been
specified formally. However, capturing scripts
with formal specifications provides us some
additional advantages over the scripts which
have been captured without any specification.
These advantages include an overall picture of
the application under test, and hence a more
efficient test suite, a test suite which is less
affected by the changes in the application, auto
matic test generation, and coverage analysis.

Step 5: Creating Script Drivers

Step 6: Coverage Analysis

Once a test suite has been created; it should be
replayed in its entirety to determine the coverage
provided by the test suite. This coverage should

A Graphical User Interface (GUI) Testing Methodology

be performed at each level, that is, the coverage

criteria for level-I should be the verification of all
the objects in the application that have been created, mapped, unmapped and destroyed, at least
once and every event expected by an object, has
been exercised on it at least once. The criteria for
higher levels is to make sure that all the interactions and side effects among objects which make
a composite object at the corresponding level, has
been verified.

An Introduction to Xman
In this section, we introduce the application
Xman, which is used to demonstrate the effectiveness of the methodology, discussed in
the previous section. Xman is a small application which is distributed with the standard X
release. It provides a graphical user interface
to the UNIX man utility. It has been developed
using the Athena widget set and some of its
windows are shown in Figure 1. The following
paragraphs briefly describe the functionality
of Xman.
When Xman is started, it displays its main
window, called Xman, by default. This main
window contains three buttons: Help, Manual
Page and Quit. Clicking on the manual page button, it displays a window, called manual page. A
Help window is displayed when the help button
is clicked. The quit button is used to exit from
Xman.
The manual page window is organized into
various sections. A breathe top of the window
contains two menu buttons, options and sections,
in the left half and a message area in the right. The
rest of the area below the bar, called text area, is
used to display the names of the available manual
pages in the currently selected section, and the
contents of the currently selected manual page.
Both the names and contents portions of the text
area are resizable and have vertical scrollbars
on the left.

The Options menu contains the following

entries:

Display Directory: It displays names of

manual pages in the entire text area.
Display Manual Page: It displays contents
of the currently selected manual page in the
entire text area.
Help: It displays a help window.
Search: It displays a dialog box to enter the
name of a manual page to search for.
To show both screens, the area is vertically
divided into two halves, with the upper half
showing the directory contents of currently
selected man page section and the lower
half showing the contents of the currently
selected manual page. This option toggles
to Show One Screen and also disables menu
entries Display Directory and Display
Manual Page.
Remove This Manpage: It removes the
Manual Page window from the screen.
Open New Manpage: It creates another
Manual Page window.
Show Version: It displays the current version ofXman in the Message area
Quit: exits from the Xman application.

The Sections menu contains one option for

each manual page section available on the system.
The standard options are User Commands, System
Calls, Subroutines, Devices, File Format, Games,
Miscellaneous, and System Administration.
The Help window displays a limited version
of the Manual Page window. The window has
exactly the same structure as the Manual Page
window but the Sections menu button and the first
five options in the Opt ions menu are disabled.
It displays man page for Xman itself. No more
than one Help window can exist at a time while
an arbitrary number of Manual Page windows
can be created.
The testing of Xman has been organized in
three layers. The first layer verifies the behavior of

665

A Graphical User Interface (GUI) Testing Methodology

individual GUI objects. The second layer verifies

the inter-object effects among objects belonging
to the same top level window. The third layer
verifies the inter-object effects among objects
belonging to different top level windows. The
following sections provide details of this testing
activity.

Testing without Formal

Specifications
This section provides a demonstration for the testing of Xman, when no formal model is available
for it. The following subsections demonstrate

Figure 2. Main windows of Xman

666

each step of the methodology, described in Section 2.

Initialization

Number of testing levels: As Xman is a

fairly small and simple application, so the
testing activity is organized in three (3)
levels. The first level tests the individual
objects in Xman. The second and third levels
verify the interactions and side effects of the
objects which belong to the same top level
window and different top level windows,
respectively.

A Graphical User Interface (GUI) Testing Methodology

Figure 3: Test Suite Hierarchy of Xman

XmanTest
All.scr
Level-1
All.scr
Xman

Help

Search

All.scr

DspXmanWin.scr

XmanMain.scr

RemoveXmanWin.scr

HelpButton.scr

QuitButton.scr

MainPageButton.scr

Level-2
All.scr

Xman

Help

Serach

All.scr

MainWindow.scr

XmanOrder.scr

Level-3
All.scr

Xman

Help

Search

All.sc

Help.scr

ManualPage.scr

Quit.scr

667

A Graphical User Interface (GUI) Testing Methodology

Location of the test suite: Suppose that the

root directory for the test suite being captured, is Xman Test. This directory contains
resource file(s) used during testing, and the
result files are created in the directory, by
default. It also has three subdirectories, that
is, Leve1-1, Leve1-2 and Leve1-3, one for
each testing level.
Object list: A list of objects, called Obj List, is
initialized. This list contains information about
the objects and is initialized to be empty.
Application resources: ~/XmanTest/
Xman.Defau1ts is the file which contains
the default application resources of Xman
for this testing activity and these resources
always remains the same.

Building the Initial Object List

By default, Xman starts with a top box, called
Main Window in Figure 2. However, a command line option, -notopbox, is available which
can be used to bypass the Main Window and
display the Manual Page window directly. As
Main Window and Manual Page window are the
only windows, which can appear when Xman
is started, so the initial object list contains only
these two objects.

Building the Test Suite

This section provides the details on building
all the three testing levels for Xman. To make
things simple, we only discuss the creation of test
suites related to the Main Window of Xman. We
ignore any keyboard accelerators to which Xman
responds. The test suites for other windows can
be created in a similar way. Figure 3 provides the
hierarchy of scripts related to the Main Window
for all the three testing levels.

668

First Level
Suppose that the initial object list is built in such
a way so that Xman Main Window is on the
top of the list. We select it as the object under
test and create a new directory, called ~/Xman
Test/Leve1-1/Xman to build test scripts related to
the first level testing. The scripts related to XMan
Main window object display itself the window
on the screen, exercise all the window manager
operations on it, and then finally pop-down the
window. DspXmanWin.scr script pop-up the
Xman window and verifies that looks right RemoveXman Win.scr pops down the Xman window.
The script XmanMain.scr ruses DspXmanWin.scr
in its entering section to display the Xman window.
It verifies the window manager operations in its
core section and then uses RemoveHe1p Win.
scr script in its leaving section to pop-down the
Xman Window. As soon as the Xman window
pops up on the screen, we see that it contains four
objects, i.e., Manual

Browser Label, Help

There are buttons which are Quit and Manual
Page. Manual Browser label is a static piece
of text and does not respond to any user actions,
so we do not need a script for it. The other three
(3) objects are active objects and respond to user
events so we create one script for each of them.
The entering section of each one of these scripts
calls DspXmanWin.scr to display the Xman
Main Window on the screen. The ending section
of each of these scripts call RemoveXmanwin.
scr to remove the Xman Main Window from
the screen. The core section of Help Button.scr
script verifies the behavior of Help button in
Xman Main Window when it is clicked on by a
mouse button. The core sections of QuitButton.
scr and ManPageBut ton.scr scripts verify the
same thing for Qui t and Manual Page buttons in
Xman Main Window.

A Graphical User Interface (GUI) Testing Methodology

Second Level
The object list of the second level contains all the
top level windows of Xman. As we are considering
the Main Window only in this discussion so we
assume that it is at the top of the list and is selected
for the testing. There is not much interaction going
on in the objects which belong to the Xman Main
Window. The only interaction is the disappearance of Main Window, in response to click on
the Quit button. So there will be only one script
related to the Main Window which will verify
that a click on the Quit button actually destroys
the Main Window of Xman. This script is called
MainWindow.scr and is located in ~/XmanTest/
Level2/. This script is also used DspXmanWin.
scr and RemoveXman.scr script to display and
remove the Main Window from the screen. Another potential script, let us call it XmanOrder.
scr, related to the Main Window verifies that the
order in which Help or Manual Page buttons are
pressed is insignificant. No matter the Help button
is pressed before or after the Manual Page button,
it displays the Help window properly. The same is
also true for the Manual Page button.

Third Level
The object list of the third level includes the root
object only, and tests any interactions among the
top level windows of Xman can be done. Such
interactions which involve the Main Window of
Xman include display of the Help window and
the Manual Page window in response to mouse
clicks on the Help and the Manual page buttons,
respectively. Similarly, it also includes disappearance of all the windows related to Xman in
response to a click on the Quit button. The three
scripts provided at this level, that is, Help.scr,
ManualPage.scr and Quit.scr, verify the behavior,
related to the corresponding button, mentioned
above. This level might also include scripts which
verify application behavior, like multiple clicks

on the Help button and do not create more than

one Help windows while each click on the Manual
Page button create a new Manual Page window.

Creating Script Drivers

Once all the scripts for Xman has been captured,
we need driver scripts so that all the scripts in
the entire suite, all the scripts in a particular testing level or all the scripts related to a particular
object can be executed automatically in the desired sequence. For example, we create a script
driver, at each testing level, which executes all
the scripts created for testing Xman Main Window
and its descendents, at that particular level. These
scripts are ~/XmanTest/Level-l/Xman/All.scr,-/XmanTest/
Level-2/Xman/All. scr, and ~/XmanTest/Level-3/Xman/All.
scr, respectively. The script ~/ XmanTest/Level-l/All.scr
drive all the scripts created for the first testing level
and similarly the other two drivers execute scripts
related to the other two levels. The script ~/XmanTest/All.scr drives all the scripts in all the three levels
of the test suite.

Testing the Test Suite

After the creation of the test suite, it is necessary
to replay all the scripts in the suite and verify if
they work properly and also to make sure that
they cover all the features which need to be tested.
Although, without a formal specification, it is
impossible to perform any reasonable automatic
coverage analysis but at least the replayed events
and the objects which appear during the replay,
can be matched against application documentation
to determine if any object or event has not been
covered by the generated test suite.

Testing with Formal

Specifications
This section demonstrates the testing of Xman
when we have enough resources to build a formal

669

A Graphical User Interface (GUI) Testing Methodology

Figure 4. User interface graph built for Xman

Root

MainWin

Form

StandBy

LikeToSav

Help

Message

Form

Form

Manual Page

Search

Form

MainPage

MainPage

Quit

Help

Label

Section

Option

ManuBar

Cancel

SCWin

Apropos

Text

Label

Form
Option

Section

Message

Scrollbar

ManuBar

Option

Section

TopSW

Message

MainPage

Scrollbar

TopForm

BtmSW

TopSB

Legend
Parent/Child
Mouse Click:
Text Entry:

model for Xman. The following subsection illustrates each step of the methodology, described
in Section 2.2.

Building the Model

When the resources permit, the first step for testing
is building a formal model for the application under
test. Figure 4 displays a User Interface Graph built
for Xman. The Root node of the graph represents
the root window of the screen. The children of
the Root node represent the six top level windows
of Xman. The nodes at lower levels in the graph
represent the descendents of the top level windows.
Let us take the main window of Xman as an example. It is represented as MainWin node in the User

670

Interface Graph. The child of the MainWin node is

the Form node which acts as a container widget
for the buttons and the label in the main window.
The ManPage, Quit and Help nodes represent the
Manual. Page, Quit and Help command buttons
in the main window, respectively. The label node
represents the Manual Browser label in the main
window. The dark black arc from the Manpage
noqe to the Manual Page node represents the fact
that clicking a mouse button over the Manpage
button in the main window affects the top level
window, called Manual Page. The arc from the Quit
node to the Root node represents that a click on
the Quit button affects all the top level windows
of Xman. The type of event represented by an arc
is reflected by the drawing pattern of the arc. The

A Graphical User Interface (GUI) Testing Methodology

two arcs mentioned above have the same pattern

and represent button clicks. An arc with a different
pattern is the arc from the Text node to the search
node. This pattern represents text entry and the
arc represents that entering text in the Text node
affect the search dialog box.

Initialization
All the decision and actions taken at the initialization step, that is, the number of testing levels
and their organization, the location of the test
suite and the application default resources, is
kept the same as for testing without a formal
specification, described in Section 4.1.

Building Object Lists

After building the user interface graph for
Xman, it is possible to build object lists for
all levels of testing. This can be done either
manually or automatically by scanning each
and every node in the User Interface Graph
and verifying if it qualifies to be tested on a
particular testing level. All the objects in Xman
qualify for the first testing level and hence are
placed on the list associated with it. The qualifying criterion for the second level is that the
object must be a top level window, that is, its
corresponding node must be a child of the root
of the user interface graph. Some of the objects
which qualify for the second testing level are
Xman Main Window, Manual Page window,
Help window, Search dialog box and so forth.
The third level treats the entire Xman application as a single object, and its corresponding
node, the root, is the only object which qualifies for the third level of testing.

Building Test Suite

The strategy for capturing scripts without any formal
specifications, discussed in Section 4.3, can be used
for capturing scripts when the application has been

specified formally. However, capturing scripts with

formal specifications also provides us the capability
to write the test scripts manually or generate test
scripts automatically by using the formal model and
the test requirements. All of these three techniques
for building a test suite are explained in the following section.

Capturing Scripts
A tool can be used for capturing user sessions for
building a test suite, in exactly the same way as for
capturing without a formal model, as mentioned
in Section 4.3. However, the presence of a formal
model makes the generated scripts more robust
and easy to follow and debug. A formal model
also provides the flexibility that the captured
scripts can be executed in any order without any
conflict in window names. It becomes easier to
modify the test suite in case of a modification to
the application under test. In fact in most cases
the modification can be reflected in the specification file and the test suite remains valid without
making any changes.

Writing Scripts Manually

The formal specifications of the application, under
test, also allows the user to write scripts manually,
even for the first level of testing. This feature is
particularly helpful when the test scripts need to be
created in parallel with the application development
in order to reduce the total development time. The
way it can be organized is to formally specify the
graphical user interface of the application once the
design is complete. The developers and testers agree
on this formal specification and any future changes
are properly communicated between the two groups.
Having the formal specification of the graphical user
interface at hand, the testers can develop the scripts
manually in parallel with the application development. Once the object hierarchy and the behavior
of the objects is known, the manual writing of test
scripts is as easy as writing a UNIX shell script.

671

A Graphical User Interface (GUI) Testing Methodology

are displayed on the screen at least once. Another

tool might generate scripts to individually select
all the options of each menu in the application
and verify that the application responds in the
expected manner. There can be another tool that
will create scripts for the selection of multiple
menu options and/or command buttons in different sequences to verify the application response.
All the mentioned tools will only create scripts
and validation data that have to be captured by
replaying these scripts and by using caprep
(Horowitz & Singhera, 1993) or a similar tool to
replay these scripts in Update mode.
The logic behind these automatic test generation tools is the same as it is used in Section 5

Automatic Test Generation

The formal specifications of an application under test also provide capabilities to generate test
scripts automatically. A tool can be developed,
which will read the specifications of an application and create test scripts for a particular level of
testing or the entire test suite. Similar tools can be
developed and used to generate test scripts to test
all the objects of a particular type or all the user
actions of a particular type. For example, such a
tool can generate test scripts to exercise clicks on
each command button in an entire application. A
similar tool can generate a suite of test scripts to
verify that all windows in an application under test

Figure 5. Pseudo Code for Automatic Test Generator

Build the User Interface Graph of the application under
test;

Build an object list by a pre-order traversal of the User

Interface Graph using parent-child relationship arcs.
for each "element on the list

do

If the element is a top level window

Create a new directory and change to the

directory.
fi

Display the element on the screen.

if the element accepts any user events

Create a script for the element

for each kind of user event accepted by the

element
do
Add commands in script to

Generate the event on the element;

Verify the effect of that event;
Undo the effect of the event;

fi
done

672

done

A Graphical User Interface (GUI) Testing Methodology

for manual creation of test suites. The tool starts

at the root of the UIG and builds a list of GUI
elements by performing a pre-order walk of the
UIG. During this walk only the arcs that represent
parent-child relationships are considered. After

building this list, its entries are taken one by

one to create test scripts for them. If the current
GUI element, taken from the list, is a top-level
window, then a separate directory is created for
it and the scripts for the element and all of its

Figure 6. Statistics file created by XTester

Legend:
CW=Create Window DW=Destroy Window MW=Map Window
UMW=Unmap Window BP=Button Press BR=Button Release
KP=Key Press KR=Key Release
Object Name
CW DW MW UMWBP BR
Xman :
1
0
2
1 0
Xman*Form :
1
0
2
1 0
Xman*ManualPage :
1
0
2
1 50
Xman*Quit :
1
0
2
1 0
Xman*Help :
1
0
2
1 10
Xman*Label :
1
0
2
1 0
StandBy :
28 23 2
1 0
StandBy*Message :
28 23 2
1 0
LikeToSave :
1
0
0
0 0
LikeToSave*Form :
1
0
0
0 0
LikeToSave*Message :
1
0
0
0
Q
LikeToSave*Yes :
1
0
0
0 0
LikeToSave*No :
1
0
0
0 0
Help :
1
0
7
7 0
Help*Form :
1
0
7
7 0
Help*MenuBar :
1
0
7
7 0
Help*Options :
1
0
7
7 20
Help*Seetions :
1
0
7
7 0
Help*Message :
1
0
7
7 0
Help*TextArea :
1
0
7
7 0
Help*Serollbar :
1
0
7
7 10
Help. Form. 3 :
1
0
7
7 0
Manpage :
27 23 27 1 0
ManPage*Form :
27 23 27 1 0
ManPage*MenuBar :
27 23 27 1 0
ManPage*Options :
27 23 27 1 92
ManPage*Seetions :
27 23 27 1 18
Manpage*Message :
27 23 27 1 0
ManPage*TextArea :
27 23 27 1 0
ManPage*Serollbar :
27 23 27 1 8
ManPage.Form.3 :
27 23 27 1 0
ManPage*DirArea :
27 23 27 1 0
ManPage*DirList :
27 23 27 1 0
27 23 27 1 0
ManPage*List :

0
0
50
0
10
0
0
0
0
0
0
0
0
0
0
0
2
0
0
0
10
0
0
0
0
6
2
0
0
8
0
0
0
0

KP
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0

KR
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0

673

A Graphical User Interface (GUI) Testing Methodology

Figure 7. Pseudo code for quality analyzer

Read in the required criteria for analysis;
Build the User Interface Graph of the application under test;

Read in the specified statistics file(s);

for each element in the User Interface Graph
do

If the element qualifies for the given criteria

Display the required information in proper format;

done

descendents are created in that directory. If the

currently selected element belongs to the category
for which a script is required, then following the
arcs that represent user actions on the element
creates one. Figure 4 shows the pseudo code for
such an automatic test generator.

Coverage Analysis
No testing activity is useful unless it provides
some coverage measures/analysis. This coverage measure reflects the quality of the testing
activity. The UIG provides us with a framework
to determine such a coverage. During capture
or replay of scripts, XTester keeps track of the
user actions and their effects on individual objects. This information is available in a .stt file
at the end of the testing activity. Currently, the
information captured in .stt about a particular
object includes the number of times it was created, mapped, unmapped, and destroyed. It also
accumulates the number of times a mouse button
or keyboard key was pressed or released over the
object. This information helps the user to locate
any particular objects in the application which have
not been created, destroyed, mapped, unmapped,
or received a user event. These statistics can also

674

be used for improving the efficiency of the test

suite by removing the repetitive testing of the same
characteristics, whenever possible. Figure 6 shows
a file created by XTester after replaying a certain
test suite for Xman. The legend is displayed at the
top of the file to describe acronyms for various
actions. The next line after the legend provides
the heading for the table. Each line in the table
provides statistics about a particular object. For
example, the very first line of the table provides
statistics about the Main Window of Xman, named
Xman. This particular line shows that the object
named Xman was created once, never destroyed,
mapped twice on the screen and unmapped once,
by the corresponding test script. It also shows
that the corresponding test suite never exercised
a button press/release or key press/release events
on the Xman object.
Tools can be developed to extract information from
a .stt file. Analysis tools are developed that take
a .uidl file to build a UIG of the application under
test, and a .stt file to get statistics collected from
a particular test run. The tool maps the collected
statistics on to the object hierarchy and produces
an annotated graph. Queries can be run on this
annotated graph to determine the effectiveness
of the test suite. For example, one can see how

A Graphical User Interface (GUI) Testing Methodology

many nodes and arcs in the object hierarchy have

not been exercised by this particular test suite, or
see the objects and inter-object arcs to determine
objects that did not receive an expected user action. Similarly, one can filter out the objects that
did not receive an expected user event to create
an annotated graph that satisfies a particular
criterion.

Invalidation of Test Data

XTester captures information as two entities,
script commands and verification data and saves
them in different files, .scr and .img, respectively.
This section describes the scenario in which
created test script(s) might fail and has to be recaptured or re-written. The following scenarios
invalidate some of the captured scripts:

An application is modified in such a way that

WM _NAME property of a top-level win
dow is changed. This modification will only
affect the scripts that have been captured
without a specification and are related to the
window whose WM _NAME property was
changed. The scripts captured with a formal
specification remain unaffected, provided
the relevant specifications are also modified
to reflect the change in the application.
The application is changed so that the order
or number of children of a particular node
is changed. The scripts that are captured
without a specification and address objects
in the modified hierarchy, are affected and
have to be recaptured or tuned. However,
the scripts captured with a formal specification remain unaffected provided the
specification is also modified accordingly.
XTester provides option to build either full
or short object names. If the application is
modified so that the depth of a hierarchy is
changed, then all the fully qualified names
belonging to that hierarchy will no longer

be valid names and has to be modified to

reflect the change. However, short names
that are relative to their immediate parents
will still remain valid.
If object information has been captured as
an image, then trying to replay the scripts
on another workstation that is incompatible
with the workstation on which image was
captured will give false alarms. The scripts
work fine across multiple platforms, however, verification data is the platform specific
in case of images. An easier way of creating
verification data for a new hardware platform
will be to replay all the scripts in Update
mode that replaces the current verification
data with the newly available one.

Conclusion and Future

Directions
In this article, we have suggested guidelines that are
useful in planning a testing activity for a graphical
user interface (GUI). We have also presented a methodology for testing a GUI, both when specifications
of an application under test is not available, and
when such specifications are provided or resource
code is available to build such a specification. This
article also demonstrates the use of the proposed
methodology to test a sample X application, Xman,
with or without specifications. It also illustrates how
the model is helpful in automatic test generation
and coverage analysis. In the end, we describe the
situations in which the scripts captured by XTester
become invalid.
The methodology and the underlying UIG
framework discussed in this article can be very
effectively used to model and test web sites and
web applications. We are actively working to
extend this proposed methodology to be used
for testing web-based applications and semantic
web applications.

675

A Graphical User Interface (GUI) Testing Methodology

References
Berstel, J., Reghizzi, S. C., Roussel, G., & San
Pietro, P. (2001). A scalable formal method for
design and automatic checking of user interfaces.
In Proceedings of the 23rd International Conference on Software Engineering, (pp. 453-462).
Campos J., & Harrison, M. (2001). Model checking interactor specifications. Automated Software
Engineering, 3(8), 275-310.
Gamma, E., Helm, R., Johnson, R., & Vlissides,
J. (1995). Design patterns. Addison Wesley
Publishers.
Horowitz, E.& Singhera, Z. (1993). Graphical
user interface testing. In proceedings of the Eleventh Annual Pacific Northwest Software Quality
Conference.
Horowitz, E.& Singhera, Z. (1993). XTester
A System for Testing X Applications. Technical
Report No. USC-CS-93-549, Department of Computer Science, University of Southern California,
Los Angeles, CA.
Horowitz, E. & Singhera, Z. (1993). A Graphical
User Interface Testing Methodology. Technical
Report No. USC-CS-93-550, Department of Computer Science, University of Southern California,
Los Angeles, CA.

Harold, M. J., Gupta, R., & Soffa, M. L. (1993).

A methodology for controlling the size of a test
suite. ACM Transactions on Software Engineering and Methodology, 2(3), 270-285.
Mercury Interactive, Mountain View, CA., http://
www.mercury.com
Myers, B. A. (1993). Why are human-computer
interfaces difficult to design and implement?
Technical Report CS-93-183, Carnegie Mellon
University, School of Computer Science.
Myers, B. A., Olsen, D. R., Jr., & Bonar, J. G.
(1993). User interface tools. In proceedings of
ACM INTERCHI93 Conference on Human Factors in Computing Systems, Adjunct Proceedings,
Tutorials, (p. 239).
Sommervill, I. (2001). Software engineering (6th
ed.) Addison Wesley Publishers.
Segue Software Inc., Newton, MA, http://www.
segue.com
Wittel, W. I., Jr. & Lewis, T. G. (1991). Integrating the mvc paradigm into an object-oriented
framework to accelerate gui application development. Technical Report 91-60-D6, Department of
Computer Science, Oregon State University.

This work was previously published in International Journal of Information Technology and Web Engineering 3(2), edited by G. Alkhatib and
D. Rine, pp. 1-18, copyright 2008 by IGI Publishing, formerly known as Idea Group Publishing (an imprint of IGI Global).

676

677

Chapter 2.19

Socio-Cognitive Engineering
Mike Sharples
University of Nottingham, Jubilee Campus, UK

Introduction
Socio-cognitive engineering is a framework for
the systematic design of socio-technical systems
(people and their interaction with technology),
based on the study and analysis of how people
think, learn, perceive, work, and interact. The
framework has been applied to the design of a
broad range of human-centered technologies,
including a Writers Assistant (Sharples, Goodlet,
& Pemberton, 1992), a training system for neuroradiologists (Sharples et al., 2000), and a mobile
learning device for children (Sharples, Corlett, &
Westmancott, 2002). It has been adopted by the
European MOBIlearn project (www.mobilearn.
org) to develop mobile technology for learning.
It also has been taught to undergraduate and
postgraduate students to guide their interactive
systems projects. An overview of the framework
can be found in Sharples et al. (2002).

Background
The approach of socio-cognitive engineering
is similar to user-centered design (Norman &
Draper, 1986) in that it builds on studies of po-

tential users of the technology and involves them

in the design process. But users are not always
reliable informants. They may idealize their
methods, describing ways in which they would
like to or have been told to work, rather than their
actual practices. Although users may be able to
describe their own styles and strategies of working, they may not be aware of how other people
can perform a task differently and possibly more
effectively. Surveys of user preferences can result
in new technology that is simply an accumulation
of features rather than an integrated system.
Thus, socio-cognitive engineering is critical
for the reliability for user reports. It extends beyond individual users to form a composite picture
of the human knowledge and activity, including
cognitive processes and social interactions, styles
and strategies of working, and language and patterns of communication. The term actor is used
rather than user to indicate that the design may
involve people who are stakeholders in the new
technology but are not direct users of it.
The framework extends previous work in soft
systems (Checkland & Scholes, 1990), sociotechnical and cooperative design (Greenbaum &
Kyng, 1991; Mumford, 1995; Sachs, 1995), and
the application of ethnography to system design

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Socio-Cognitive Engineering

(see Rogers & Bellotti [1997] for a review). It

incorporates existing methods of knowledge engineering, task analysis, and object-oriented design,
but integrates them into a coherent methodology
that places equal emphasis on software, task,
knowledge, and organizational engineering.
The framework also clearly distinguishes
studying everyday activities using existing technology from studying how the activity changes
with proposed technology. It emphasizes the dialectic between people and artefacts; using artefacts
changes peoples activities, which, in turn, leads
to new needs and opportunities for design.

Framework
Figure 1 gives a picture of the flow and main
products of the design process. It is in two main
parts: a phase of activity analysis to interpret
how people work and interact with their current
tools and technologies, and a phase of systems
design to build and implement new interactive
technology. The bridge between the two is the
relationship between the Task Model and the
Design Concept. Each phase comprises stages of
analysis and design that are implemented through

specific methods. The framework does not prescribe which methods to use; the choice depends
on the type and scale of the project.
It is important to note that the process is not a
simple sequence but involves a dialogue between
the stages. Earlier decisions and outcomes may
need to be revised in order to take account of later
findings. When the system is deployed, it will enable and support new activities, requiring another
cycle of analysis, revision of the Task Model, and
further opportunities for design.
The elements of socio-cognitive engineering
are as follows:

Project: The diagram shows the process of

design, implementation, and deployment for
a single project.
Actors: Different types of people may be
involved in or affected by the design and
deployment, including (depending on the
scale of the project) design, marketing and
technical support teams, direct users of the
system, and other people affected by it (e.g.,
administrative staff).
Roles: The actors take on roles (e.g., team
leader), which may change during the project.

Figure 1. Overview of the flow and main products of the design process

678

Socio-Cognitive Engineering

Stage: Each box represents one stage of the

project.
Methods: Each stage can be carried out by
one or more methods of analysis and design,
which need to be specified before starting
the stage.
Tools: Each method has associated tools
(for activity analysis, software specification,
systems design, and evaluation) in order to
carry out the method.
Outcomes: Each stage has outcomes that
must be documented, and these are used to
inform and validate the system design.
Measures: Each design decision must be
validated by reference to outcomes from
one of the stages.

The general sequence for socio-cognitive

engineering is as follows:
1.
2.
3.
4.
5.
6.
7.

Form a project team.

Produce General Requirements for the
project.
Decide which methods and tools will be
used for each stage of the project.
Decide how the process and outcomes will
be documented.
Decide how the project will be evaluated.
Carry out each stage of the project, ensuring
that the requirements match the design.
Carry out a continuous process of documentation and evaluation.

The process starts by specifying the General

Requirements for the system to be designed. These
provide broad yet precise initial requirements and
constraints for the proposed system in language
that designers and customers can understand.
They are used to guide the design and to provide
a reference for validation of the system. The requirements normally should indicate:

The scope of the project;

The main actors involved in designing, deploying, using, and maintaining the system;

The market need and business case; and

General attributes and constraints of the
proposed system (i.e., whether it aims to
support individual or collaborative working).

The requirements will be extended and made

more precise as the project progresses.
This leads to two parallel studies: a theorybased study of the underlying cognitive processes
and social activities, and an investigation into how
everyday activities are performed in their normal
contexts. The Theory of Use involves an analysis
of relevant literature from cognitive psychology,
social sciences, and business management to
form a rich picture of the human knowledge and
activity. It is essential that this should offer a clear
guide to system design. Thus, it must be relevant
to the intended use of the system and extend the
requirements in a form that can be interpreted by
software designers and engineers.
The aim of carrying out Field Studies is to
uncover how people interact with current technology in their normal contexts. The role of the
fieldworker is both to interpret activity and to assist technology design and organizational change.
This addresses the widely recognized problem of
ethnographic approaches that, while they can provide an understanding of current work practices,
are not intended to explore the consequences of
socio-technical change.
Table 1 shows a multi-level structure for field
studies, with level 1 consisting of a survey of the
existing organizational structures and schedules,
levels 2 and 3 providing an analysis of situated
practices and interactions of those for whom the
technology is intended, and level 4 offering a
synthesis of the findings in terms of designs for
new socio-technical systems. The four levels give
an overview of activity, leading to a more detailed
investigation of particular problem areas, with
each level illuminating the situated practices and
also providing a set of issues to be addressed for the
next level. These piece together into a composite
picture of how people interact with technology
679

Socio-Cognitive Engineering

Table 1. Multi-level structure for field studies

Level 1

Activity structures and schedules

Activity:

Study work plans, organizational structures, syllabuses, resources.

Purpose:

To discover how the activities are supposed to be conducted.

Outcome:

Description of the existing organizational and workplace structures; identification of significant events.

Level 2

Significant events

Activity:

Observe representative formal and informal meetings and forms of communication.

Purpose:

To discover how activities, communication, and social interaction are conducted in practice.

Outcome:

A description and analysis of events that might be important to system design; identification of
mismatches between how activity has been scheduled and how it is has been observed to happen.

Level 3

Conceptions and conflicts

Activity:

Conduct interviews with participants to discuss areas of activity needing support, breakdowns, issues,
differences in conception.

Purpose:

To determine peoples differing conceptions of their activity; uncover issues of concern in relation to
new technology; explore mismatches between what is perceived to happen and what has been observed.

Outcome:

Issues in everyday life and interactions with existing technology that could be addressed by new
technology and working practices.

Level 4

Determining designs

Activity:

Elicitation of requirements; design space mapping; formative evaluation of prototypes.

Purpose:

To develop new system designs.

Outcome:

Prototype technologies and recommendations for deployment.

in their everyday lives, the limitations of existing practices, and ways in which they could be
improved by new technology.
The outcomes of these two studies are synthesized into a Task Model. This is a synthesis of
theory and practice related to how people perform
relevant activities with their existing technologies.
It is the least intuitive aspect of socio-cognitive
engineering; it is tempting to reduce it to a set of
bullet-point issues, yet it provides a foundation
for the systems design. It could indicate:

680

The main actors and their activity systems;

How the actors employ tools and resources to
mediate their interaction and to externalize
cognition;
How the actors represent knowledge to
themselves and to others;

The methods and techniques that the actors

employ, including differences in approach
and strategy;
The contexts in which the activities occur;
The implicit conventions and constraints
that influence the activity; and
The actors conceptions of their work, including sources of difficulty and breakdown
in activity and their attitudes toward the
introduction of new technology.

The Design Concept needs to be developed

in relation to the Task Model. It should indicate
how the activities identified by the Task Model
could be transformed or enhanced with the new
technology. It should:

Indicate how limitations from the Task

Model will be addressed by new technology;

Socio-Cognitive Engineering

Table 2. A building-block framework for socio-cognitive system design

Software Engineering

Task Engineering

Knowledge Engineering

Organizational
Engineering

Maintain

Installed system

New task structure

Augmented knowledge

Evaluate

Debugging

Usability

Conceptual change, skill

New organizational
structure
Organizational change

development
Integrate
Implement
Design

Prototype System
Prototypes,

Interfaces, Cognitive

Knowledge

Communications,

Documentation

tools

representation

Network resources

Algorithms and

Human-computer

Domain map, user

Socio-technical system

heuristics

interaction

model

Interpret

Task Model

Analyze

Requirements

Survey

Existing systems

Tasks: goals, objects,

Knowledge: concepts,

Workplace: practices,

methods

skills

interactions

Conventional task

Domain knowledge

Organizational

structures and processes

Propose

Outline a system image (Norman, 1986) for

the new technology;
Show the look and feel of the proposed
technology;
Indicate the contexts of use of the enhanced
activity and technology; and
Propose any further requirements that have
been produced as a result of constructing
the Design Concept.

The Design Concept should result in a set of

detailed design requirements and options that can
be explored through the design space.
The relationship between the Task Model and
Design Concept provides the bridge to a cycle of
iterative design that includes:

Generating a space of possible system designs, systematically exploring design option

and justifying design decisions;
Specifying the functional and non-functional
aspects of the system;

structures and schedules

General Requirements

Implementing the system; and

Deploying and maintaining the system.

Although these stages are based on a conventional process of interactive systems design (see
Preece, Rogers, & Sharp [2002] for an overview),
they give equal emphasis to cognitive and organizational factors as well as to task and software
specifications. The stages shown in Figure 1 are
an aid to project planning but are not sufficiently
detailed to show all the design activities. Nor does
the figure make clear that to construct a successful integrated system requires the designers to
integrate software engineering with design for
human cognition, social interaction, and organizational management. The building-block
diagram in Table 2 gives a more detailed picture
of the systems design process.
The four pillars indicate the main processes
of software, task, knowledge, and organizational
engineering. Each brick in the diagram shows

681

Socio-Cognitive Engineering

one outcome of a design stage, but it is not

necessary to build systematically from the bottom up. A design team may work on one pillar
(e.g., knowledge engineering) up to the stage of
system requirements, or ot may develop an early
prototype based on a detailed task analysis but
without a systematic approach to software engineering. How each activity is carried out depends
on the particular application domain, actors, and
contexts of use.
The design activities are modular, allowing
the designer to select one or more methods of
conducting the activity according to the problem
and domain. For example, the usability evaluation
could include an appropriate selection of general
methods for assessing usability, or it could include an evaluation designed for the particular
domain.
It should be emphasized that the blocks are
not fixed entities. As each level of the system is
developed and deployed, it will affect the levels
that follow (e.g., building a prototype system
may lead to revising the documentation or reevaluating the human-computer interaction;
deploying the system will create new activities).
These changes need to be analyzed and supported
through a combination of new technology and
new work practices. Thus, the building blocks
must be revisited both individually to analyze
and update the technology in use, and through a
larger process of iterative redesign.
Although Table 1 shows system evaluation
as a distinct phase, there also will be a continual
process of testing to verify and validate the design,
as shown in Figure 1. Testing is an integral part
of the entire design process, and it is important
to see it as a lifecycle process (Meek & Sharples,
2001) with the results of testing early designs and
prototypes being passed forward to provide an
understanding of how to deploy and implement
the system, and the outcomes of user trials being
fed back to assist in fixing bugs and improving
the design choices.

682

The result of the socio-cognitive engineering

process is a new socio-technical system consisting
of new technology, its associated documentation,
and proposed methods of use. When this is deployed in the workplace, home, or other location,
it not only should produce bugs and limitations
that need to be addressed but also engender new
patterns of work, social, and organizational structures that become contexts for further analysis
and design.

Future Trends
The computer and communications industries are
starting to recognize the importance of adopting
a human-centered approach to the design of new
socio-technical systems. They are merging their
existing engineering, business, industrial design,
and marketing methods into an integrated process,
underpinned by rigorous techniques to capture
requirements, define goals, predict costs, plan
activities, specify designs, and evaluate outcomes.
IBM, for example, has developed the method of
User Engineering to design for the total user experience (IBM, 2004). As Web-based technology
becomes embedded into everyday life, it increasingly will be important to understand and design
distributed systems for which there are no clear
boundaries between people and technology.

Conclusion
Socio-cognitive engineering forms part of an
historic progression from user-centered design and
soft systems analysis toward a comprehensive and
rigorous process of socio-technical systems design
and evaluation. It has been applied through a broad
range of projects for innovative human technology
and is still being developed, most recently as part
of the European MOBIlearn project.

Socio-Cognitive Engineering

References
Checkland, P., & Scholes, J. (1990). Soft systems
methodology in action. Chichester, UK: John
Wiley & Sons.
Greenbaum, J., & Kyng, M. (Eds.). (1991). Design
at work: Cooperative design of computer systems.
Hillsdale, NJ: Lawrence Erlbaum Associates.
IBM. (2004). User engineering. Retrieved August
12, 2004, from http://www-306.ibm.com/ibm/
easy/eou_ext.nsf/publish/1996
Meek, J., & Sharples, M. (2001). A lifecycle
approach to the evaluation of learning technology. Proceedings of the CAL 2001 Conference,
Warwick, UK.
Mumford, E. (1995). Effective systems design and
requirements analysis: The ETHICS approach.
Basingstoke, Hampshire, UK: Macmillan.
Norman, D.A. (1986). Cognitive engineering.
In D.A. Norman, & S.W. Draper (Eds.), User
centred system design. Hillsdale, NJ: Lawrence
Erlbaum.

Sharples, M., Goodlet, J., & Pemberton, L. (1992).

Developing a writers assistant. In J. Hartley
(Ed.), Technology and writing: Readings in the
psychology of written communication (pp. 209220). London: Jessica Kingsley.
Sharples, M., et al. (2000). Structured computer-based training and decision support in
the interpretation of neuroradiological images.
International Journal of Medical Informatics,
60(30), 263-228.
Sharples, M., et al. (2002). Socio-cognitive
engineering: A methodology for the design of
human-centred technology. European Journal of
Operational Research, 132(2), 310-323.

KEY TERMS
Activity System: The assembly and interaction of people and artefacts considered as a holistic
system that performs purposeful activities. See
http://www.edu.helsinki.fi/activity/pages/chatanddwr/activitysystem/

Norman, D.A., & Draper, S. (1986). User centered

system design: New perspectives on humancomputer interaction. Hillsdale, NJ: Lawrence
Erlbaum Associates.

Human-Centred Design: The process of

designing socio-technical systems (people in
interaction with technology) based on an analysis
of how people think, learn, perceive, work, and
interact.

Preece, J., Rogers, Y., & Sharp, H. (2002). Interaction design: Beyond human-computer interaction.
New York: John Wiley & Sons.

Socio-Technical System: A system comprising people and their interactions with technology
(e.g., the World Wide Web).

Rogers, Y., & Bellotti, V. (1997). Grounding

blue-sky research: How can ethnography help?
Interactions, 4(3), 58-63.

Soft Systems Methodology: An approach

developed by Peter Checkland to analyze complex
problem situations containing social, organizational, and political activities.

Sachs, P. (1995). Transforming work: Collaboration, learning and design. Communications of the
ACM, 38(9), 36-44.
Sharples, M., Corlett, D., & Westmancott, O.
(2002). The design and implementation of a mobile learning resource. Personal and Ubiquitous
Computing, 6, 220-234.

System Image: A term coined by Don Norman (1986) to describe the guiding metaphor or
model of the system that a designer presents to
users (e.g., the desktop metaphor or the telephone
as a speaking tube). The designer should aim
to create a system image that is consistent and

683

Socio-Cognitive Engineering

familiar (where possible) and enables the user to

make productive analogies.
Task Analysis: An analysis of the actions and/
or knowledge and thinking that a user performs
to achieve a task. See http://www.usabilitynet.
org/tools/taskanalysis.htm

User-Centered Design: A well-established

process of designing technology that meets users expectations or that involves potential users
in the design process.
User Engineering: A phrase used by IBM
to describe an integrated process of developing
products that satisfy and delight users.

This work was previously published in the Encyclopedia of Human Computer Interaction, edited by C. Ghaoui, pp. 542-547,
copyright 2006 by Information Science Reference, formerly known as Idea Group Reference (an imprint of IGI Global).

684

685

Chapter 2.20

On the Cognitive Processes of

Human Perception with
Emotions, Motivations,
and Attitudes
Yingxu Wang
University of Calgary, Canada

Abstract
An interactive motivation-attitude theory is developed based on the Layered Reference Model
of the Brain (LRMB) and the object-attributerelation (OAR) model. This paper presents a
rigorous model of human perceptual processes
such as emotions, motivations, and attitudes. A
set of mathematical models and formal cognitive
processes of perception is developed. Interactions
and relationships between motivation and attitude
are formally described in real-time process algebra (RTPA). Applications of the mathematical
models of motivations and attitudes in software
engineering are demonstrated. This work is a
part of the formalization of LRMB, which provides a comprehensive model for explaining the
fundamental cognitive processes of the brain and
their interactions. This work demonstrates that
the complicated human emotional and perceptual phenomena can be rigorously modeled and

formally treated based on cognitive informatics

theories and denotational mathematics.

INTRODUCTION
A variety of life functions and cognitive processes has been identified in cognitive informatics
(Wang, 2002a, 2003a, 2003b, 2007b) and cognitive
psychology (Payne & Wenger, 1998; Pinel, 1997;
Smith, 1993; Westen, 1999; Wilson & Keil, 1999).
In order to formally and rigorously describe a
comprehensive and coherent set of mental processes and their relationships, an LRMB has been
developed (Wang & Wang, 2006; Wang, Wang,
Patel, & Patel, 2006) that explains the functional
mechanisms and cognitive processes of the brain
and the natural intelligence. LRMB encompasses
39 cognitive processes at six layers known as the
sensation, memory, perception, action, meta and
higher cognitive layers from the bottom up.

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

On the Cognitive Processes of Human Perception with Emotions, Motivations, and Attitudes

Definition 1: Perception is a set of internal sensational cognitive processes of the brain at the
subconscious cognitive function layer that detects,
relates, interprets, and searches internal cognitive
information in the mind.
Perception may be considered as the sixth
sense of human beings since almost all cognitive life functions rely on it. Perception is also an
important cognitive function at the subconscious
layers that determines personality. In other words,
personality is a faculty of all subconscious life
functions and experience cumulated via conscious
life functions. It is recognized that a crucial
component of the future generation computers
known as the cognitive computers is the perceptual engine that mimic the natural intelligence
(Wang, 2006a, 2007c).
The main cognitive processes at the perception layer of LRMB are emotion, motivation, and
attitude (Wang et al., 2006). This article presents
a formal treatment of the three perceptual processes, their interrelationships, and interactions.
It demonstrates that complicated psychological
and cognitive mental processes may be formally
modeled and rigorously described. Mathematical
models of the psychological and cognitive processes of emotions, motivations, and attitudes are
developed in the following three sections. Then,
interactions and relationships between emotions,
motivations, and attitudes are analyzed. Based
on the integrated models of the three perception
processes, the formal description of the cognitive processes of motivations and attitudes will
be presented using RTPA (Wang, 2002b, 2003c,

2006b, 2007a). Applications of the formal models of emotions, motivations, and attitudes will
be demonstrated in a case study on maximizing
strengths of individual motivations in software
engineering.

THE HIERARCHICAL MODEL OF

EMOTIONS
Emotions are a set of states or results of perception
that interprets the feelings of human beings on
external stimuli or events in the binary categories
of pleasant or unpleasant.
Definition 2: An emotion is a personal feeling
derived from ones current internal status, mood,
circumstances, historical context, and external
stimuli.
Emotions are closely related to desires and
willingness. A desire is a personal feeling or
willingness to possess an object, to conduct an
interaction with the external world, or to prepare
for an event to happen. A willingness is the faculty
of conscious, deliberate, and voluntary choice of
actions.
According to the study by Fischer, Shaver, and
Carnochan (1990) and Wilson and Keil (1999), the
taxonomy of emotions can be described at three
levels known as the sub-category, basic, and super
levels as shown in Table 1.
It is interesting that human emotions at the
perceptual layer may be classified into only two
opposite categories: pleasant and unpleasant.

Table 1. Taxonomy of emotions

Level
Positive (pleasant)

Basic level

Joy

Love

Anger

Sadness

Fear

Bliss, pride,
contentment

Fondness,
infatuation

Annoyance, hostility,
contempt, jealousy

Agony, grief, guilt,

loneliness

Horror, worry

Sub-category level

686

Description

Super level

Negative (unpleasant)

On the Cognitive Processes of Human Perception with Emotions, Motivations, and Attitudes

Various emotions in the two categories can be

classified at five levels according to its strengths
of subjective feelings as shown in Table 2 (Wang,
2005), where each level encompasses a pair of positive/negative or pleasant/unpleasant emotions.
Definition 3: The strength of emotion |Em| is a
normalized measure of how strong a persons
emotion on a five-level scale identified from 0
through 4, that is:
0 |Em| 4

(1)

where |Em| represents the absolute strength of an

emotion regardless whether it is positive (pleasant)
or negative (unpleasant), and the scope of |Em| is
corresponding to the definitions of Table 2.
It is observed that an organ known as hypothalamus in the brain is supposed to interpret
the properties or types of emotions in terms of
pleasant or unpleasant (Payne & Wenger, 1998;
Pinel, 1997; Smith, 1993; Wang et al., 2006;
Westen, 1999).
Definition 4: Let Te be a type of emotion, ES the
external stimulus, IS the internal perceptual status, and BL the Boolean values true or false. The
perceptual mechanism of the hypothalamus can
be described as a function, that is:

Te : ES IS BL

(2)

It is interesting that the same event or stimulus

ES may be explained in different types, in terms
of pleasant or unpleasant, due to the difference of
the real-time context of the perceptual status IS
of the brain. For instance, walking from home to
the office may be interpreted as a pleasant activity
for one who likes physical exercise, but the same
walk due to car breakdown will be interpreted as
unpleasant. This observation and the taxonomy
provided in Tables 1 and 2 leads to the following
Theorem.
Theorem 1: The human emotional system is a
binary system that interprets or perceives an external stimulus and/or internal status as pleasant
or unpleasant.
Although there are various emotional categories in different levels, the binary emotional system
of the brain provides a set of pairwise universal
solutions to express human feelings. For example,
angry may be explained as a default solution or
generic reaction for an emotional event when there
is no better solution available; otherwise, delight
will be the default emotional reaction.

Table 2. The hierarchy of emotions

Level
(Positive/Negative)
0

No emotion

Weak emotion

Moderate emotion

Strong emotion

Strongest emotion

Description
Comfort

Safeness, contentment, fulfillment, trust

Fear

Worry, horror, jealousy, frightening, threatening

Joy

Delight, fun, interest, pride

Sadness

Anxiety, loneliness, regret, guilt, grief, sorrow, agony

Pleasure

Happiness, bliss, excitement, ecstasy

Anger

Annoyance, hostility, contempt, infuriated, enraged

Love

Intimacy, passion, amorousness, fondness, infatuation

Hate

Disgust, detestation, abhorrence, bitterness

687

On the Cognitive Processes of Human Perception with Emotions, Motivations, and Attitudes

THE MATHEMATICAL MODEL OF

MOTIVATIONS
Motivation is an innate potential power of human
beings that energizes behavior. It is motivation
that triggers the transformation from thought
(information) into action (energy). In other words,
human behaviors are the embodiment of motivations. Therefore, any cognitive behavior is driven
by an individual motivation.
Definition 5: A motivation is a willingness or desire triggered by an emotion or external stimulus
to pursue a goal or a reason for triggering an
action.
As described in LRMB (Wang et al., 2006),
motivation is a cognitive process of the brain at
the perception layer that explains the initiation,
persistence, and intensity of personal emotions
and desires, which are the faculty of conscious,
deliberate, and voluntary choices of actions.
Motivation is a psychological and social
modulating and coordinating influence on the
direction, vigor, and composition of behavior.
This influence arises from a wide variety of
internal, environmental, and social sources, and
is manifested at many levels of behavioral and
neural organizations.
The taxonomy of motives can be classified into
two categories known as learned and unlearned
(Wittig, 2001). The latter is the primary motives
such as the survival motives (hunger, thirst, breath,
shelter, sleep, and eliminating pain). The former
are the secondary motives such as the need for
achievement, friendship, affiliation, dominance
of power, and relief anxiety, which are acquired
and extended based on the primary motives.
Definition 6: The strength of motivation M is a
normalized measure of how strong a persons
motivation on a scale of 0 through 100, that is:
0 M 100

688

(3)

where M = 100 is the strongest motivation and M

= 0 is the weakest motivation.
It is observed that the strength of a motivation
is determined by multiple factors (Westen, 1999;
Wilson & Keil, 1999) such as:
The absolute motivation |Em|: The strength
of the emotion.
b. The relative motivation E - S: A relative
difference or inequity between the expectancy of a person E for an object or an action
towards a certain goal and the current status
S of the person.
c. The cost to fulfill the motivation C: A subjective assessment of the effort needed to
accomplish the expected goal.
a.

Therefore, the strength of a motivation can

be quantitatively analyzed and estimated by the
subjective and objective motivations and their cost
as described in the following theorem.
Theorem 2: The strength of a motivation M is
proportional to both the strength of emotion |Em|
and the difference between the expectancy of desire E and the current status S, of a person, and is
inversely proportional to the cost to accomplish
the expected motivation C, that is:
M

2.5

| Em | (E -S )
C

(4)

where 0 |Em| 4, 0 (E,S) 10, 1 C 10, and

the coefficient 2.5 makes the value of M normalized in the scope of [0 .. 100].
In Theorem 2, the strength of a motivation is
measured in the scope 0 M 100. When M >
1, the motivation is considered being a desired
motivation, because it indicates both an existing
emotion and a positive expectancy. The higher the
value of M, the stronger the motivation.
According to Theorem 2, in a software engineering context, the rational action of a manager
of a group is to encourage individual emotional
desire, and the expectancy of each software en-

On the Cognitive Processes of Human Perception with Emotions, Motivations, and Attitudes

gineer and to decrease the required effort for the

employees by providing additional resources or
adopting certain tools.

Corollary 1: There are super strong motivations
toward a resolute goal by a determined expectancy
of a person at any cost.

to people, objects, events, and other stimuli. Attitudes may be formally defined as follows.

It is noteworthy that a motivation is only a

potential mental power of human beings, and a
strong motivation will not necessarily result in a
behavior or action. The condition for transforming a motivation into a real behavior or action is
dependent on multiple factors, such as values,
social norms, expected difficulties, availability
of resources, and the existence of alternative
goals.
The motivation of a person is constrained by
the attitude and decision-making strategies of the
person. The former is the internal (subjective)
judgment of the feasibility of the motivation, and
the latter is the external (social) judgment of the
feasibility of the motivation. Attitude and decision-making mechanisms will be analyzed in the
following subsections.

The modes of attitudes can be positive or

negative, which can be quantitatively analyzed
using the following model.

THE MATHEMATICAL MODEL OF

ATTITUDES
As described in the previous section, motivation
is the potential power that may trigger an observable behavior or action. Before the behavior is
performed, it is judged by an internal regulation
system known as the attitude.
Psychologists perceive attitude in various
ways. Fazio (1986) describes an attitude as an
association between an act or object and an evaluation. Eagly and Chaiken (1992) define that an
attitude is a tendency of a human to evaluate a
person, concept, or group positively or negatively
in a given context. More recently, Wittig (2001)
describes attitude as a learned evaluative reaction

Definition 7: An attitude is a subjective tendency

towards a motivation, an object, a goal, or an
action based on an intuitive evaluation of its
feasibility.

Definition 8. The mode of an attitude A is determined by both an objective judgment of its conformance to the social norm N and a subjective
judgment of its empirical feasibility F, that is:
A

1, N

0, N

(5)

where A = 1 indicates a positive attitude; otherwise,

it indicates a negative attitude.

INTERACTIONS BETWEEN
MOTIVATION AND ATTITUDE
This section discusses the relationship between
the set of interlinked perceptual psychological
processes such as emotions, motivations, attitudes,
decisions, and behaviors as formally modeled in
the preceeding sections. A motivation/attitudedriven behavioral model will be developed for
formally describing the cognitive processes of
motivations and attitudes.
It is observed that motivation and attitude have
considerable impact on behavior and influence the
ways a person thinks and feels (Westen, 1999). A
reasoned action model is proposed by Fishbein
and Ajzen (1975) that suggests human behavior
is directly generated by behavioral intensions,
which are controlled by the attitude and social
norms. An initial motivation before the judgment
by an attitude is only a temporal idea; with the

689

On the Cognitive Processes of Human Perception with Emotions, Motivations, and Attitudes

judgment of the attitude, it becomes a rational

motivation (Wang et al., 2006), also known as the
behavioral intention.
The relationship between an emotion, motivation, attitude, and behavior can be formally and
quantitatively described by the Motivation/Attitude-Driven Behavior (MADB) model as illustrated in Figure 1. In the MADB model, motivation
and attitude have been defined in Equations 4 and
5. The rational motivation, decision, and behavior
can be quantitatively analyzed according to the
following definitions. It is noteworthy that, as
shown in Figure 1, a motivation is triggered by
an emotion or desire.
Definition 9: A rational motivation Mr is a motivation regulated by an attitude A with a positive
or negative judgment, that is:
Mr

M A
2.5

| Em | (E -S )
A
C

(6)

Definition 10: A decision D for confirming an

attitude for executing a motivated behavior is a
binary choice on the basis of the availability of
time T, resources R, and energy P, that is:

1, T

P =

0, T

P =

(7)

Definition 11. A behavior B driven by a motivation

Mr and an attitude is a realized action initiated
by a motivation M and supported by a positive
attitude A and a positive decision D toward the
action, that is:
B

T,

Mr D

F,

otherwise

2.5 | Em | (E -S )
AD
C

(8)

FORMAL DESCRIPTION OF
COGNITIVE PROCESSES OF
MOTIVATION AND ATTITUDE
The formal models of emotion, motivation, and
attitude have been developed in previous sections.
This section extends the models and their relationship into detailed cognitive processes based on
the OAR model (Wang, 2007d) and using RTPA
(Wang, 2002b, 2003c, 2006b, 2007a), which enable
more rigorous treatment and computer simulations
of the MADB model.

Figure 1. The motivation/attitude-driven behavior (MADB) model

Satisfy/dissatisfy

Motivation

Rational
motivation

Stimuli

Emotion

Strengthen/weaken

Attitude
(Perceptual
feasibility)
N
Values/
social
norms

Experience

Internal process

690

Mr

Behavior

Outcome

D
Decision
(physical
feasibility)

T/R/P
Availability
of time,
resources,
and energy
External process

On the Cognitive Processes of Human Perception with Emotions, Motivations, and Attitudes

The Cognitive Process of

Motivations
The mathematical model of rational motivation
is described in Equation 6. Based on Equation 6,
the cognitive process of motivation is presented
in Figure 2. The motivation process is divided
into four major sub-processes known as: (1) to
form motivation goal; (2) to estimate strength of
motivation; (3) to form rational motivation; and
(4) to stimulate behavior for the motivation.
The MADB model provides a formal explanation of the mechanism and relationship between
motivation, attitude, and behavior. The model can
be used to describe how the motivation process
drives human behaviors and actions, and how the
attitude as well as the decision-making process
help to regulate the motivation and determines
whether the motivation should be implemented.

The Cognitive Process of Attitudes

The mathematical model of attitude has been
described in Equation 5. Based on Equation 5,
the cognitive process of attitude is presented in
Figure 3. The attitude process is divided into
three major sub-processes known as: (1) to check
the mode of attitude; (2) to determine physical
availability; and (3) to stimulate behavior for the
motivation.

The Integrated Process of Motivation

and Attitudes
According to the MADB model and the formal description of the motivation and attitude
processes as shown in Figures 1 through 3, the
cognitive processes of motivation and attitude
are interleaved. An integrated process that combines both motivation and attitude is given in
Figure 4 via the following sub-processes: (1) to
form motivation goals; (2) to estimate strength
of motivation; (3) to check the mode of attitude;

(4) to form rational motivation; (5) to determine

physical availability; and (6) to stimulate behavior
for the rational motivation.

MAXIMIZING STRENGTHS OF
INDIVIDUAL MOTIVATIONS
Studies in sociopsychology provide a rich theoretical basis for perceiving new insights into
the organization of software engineering. It is
noteworthy that in a software organization, according to Theorem 2, the strength of a motivation of individuals M is proportional to both the
strength of emotion and the difference between
the expectancy and the current status of a person.
At the same time, it is inversely proportional to
the cost to accomplish the expected motivation C.
The job of management at different levels of an
organization tree is to encourage and improve Em
and E, and to help employees to reduce C.
Example 1: In software engineering project organization, the manager and programmers may be
motivated to the improvement of software quality
to a different extent. Assume the following factors
as shown in Table 3 are collected from a project
on the strengths of motivations to improve the
quality of a software system, analyze how the
factors influence the strengths of motivations of
the manager and the programmers.
According to Theorem 2, the strengths of
motivations of the manager M1 and the programmers M2 can be estimated using Equation 4,
respectively:
M 1 (manager )

and
M 2 (programer )

2.5 | Em | (E -S )
C
2.5 4 (8 - 5)
3
10.0

2.5 3.6 (8 - 6)
8
2.3
691

On the Cognitive Processes of Human Perception with Emotions, Motivations, and Attitudes

Figure 2. The cognitive process of motivation

The Motivation Process
Motivation (I:: oS; O:: OAR ST)
{ I. Form motivation goal(s)
ObjectIdentification (o, A , R )
II. Estimate strength of motivation M(o)N
Quantify (Em(o)N)
// The strength of emotion
Quantify (S(o)N)

// The current status

Quantify (E(o)N)

// The expectancy of desire

Quantify (C(o)N)

// The cost to accomplish

M (o)N

2.5 Em (o)N (E (o)N -NS (o) )

C (o)N

M(o) N > 1

M(o)BL = T

// Positive motivation

M(o)BL = F

// Negative motivation

)
III. Check the mode of attitude A(o)N
// Refer to the Attitude process
IV. Form rational motivation Mr(o)
Mr(o)N := M(o)N A(o)N
(
Mr(o)N > 1
|

Mr(o)BL = T

// Rational motivation

Mr(o)BL = F

// Irrational motivation

)
V. Determine physical availability D(o)N
// Refer to the Attitude process
VI. Stimulate behavior for Mr(o)
(
D(o)N = 1

// Implement motivation o

GenerateAction (Mr(o))
ExecuteAction (Mr(o))
R := R <o, Mr(o)>
|

// Give up motivation o

~
D(o)N := 0
o :=
R :=

)
OAR ST = <O o, A A , R R > // Form new OAR model
Memorization (OAR ST)
}

692

On the Cognitive Processes of Human Perception with Emotions, Motivations, and Attitudes

Figure 3. The cognitive process of attitude

The Attitude Process
Attitude (I:: oS; O:: OAR ST)
{ I. Form motivation goal(s)
ObjectIdentification (o, A , R )
II. Estimate strength of motivation M(o)N
// Refer to the Motivation process
III. Check the mode of attitude A(o)N
// Perceptual feasibility
Qualify (N(o)BL)
// The social norm
Qualify (F(o)BL)

// The subjective feasibility

N(o)BL F(o)BL = T

A(o)N := 1
A(o)N := 0

IV. Form rational motivation Mr(o)

// Refer to the Motivation process
V. Determine physical availability D(o)N
Qualify (T(o)BL)
// The time availability
Qualify (R(o)BL)

// The resource availability

Qualify (P(o)BL)

// The energy availability

T(o)BL R(o)BL P(o)BL = T

D(o)N := 1

// Confirmed motivation

D(o)N := 0

// Infeasible motivation

VI. Stimulate behavior for Mr(o)

(
D(o)N = 1

// Implement motivation o

GenerateAction (Mr(o))
ExecuteAction (Mr(o))
R := R <o, Mr(o)>
|

// Give up motivation o
D(o)N := 0
o :=
R :=

)
OAR ST = <O o, A A , R R > // Form new OAR model
Memorization (OAR ST)
}

Table 3. Motivation factors of a project

Role
The manager
Programmers

Em

3.6

693

On the Cognitive Processes of Human Perception with Emotions, Motivations, and Attitudes

Figure 4. The integrated process of motivation and attitude

The Motivation and Attitude Process
Motivation-Attitude (I:: oS; O:: OARST)
{ I. Form motivation goal(s)
ObjectIdentification (o, A, R)

II. Estimate strength of motivation M(o)N

// The strength of emotion
Quantify (Em(o)N)

Quantify (S(o)N)

// The current status

Quantify (E(o)N)

// The expectancy of desire

Quantify (C(o)N)

// The cost to accomplish

M (o) N

2.5 E m (o) N (E (o) N - S (o) N )

C (o) N

M(o)N > 1

M(o)BL = T

// Positive motivation

M(o)BL = F

// Negative motivation

)
III. Check the mode of attitude A(o)N
// Perceptual feasibility
Qualify (N(o)BL)
// The social norm

Qualify (F(o)BL)

// The subjective feasibility

N(o)BL F(o)BL = T
A(o)N := 1
~
A(o)N := 0

IV. Form rational motivation Mr(o)

Mr(o)N := M(o)N A(o)N
Mr(o)N > 1
(

Mr(o)BL = T

// Rational motivation

Mr(o)BL = F

// Irrational motivation

)
V. Determine physical availability D(o)N
Qualify (T(o)BL)
// The time availability

Qualify (R(o)BL)

// The resource availability

Qualify (P(o)BL)

// The energy availability

T(o)BL R(o)BL P(o)BL = T

D(o)N := 1

// Confirmed motivation

D(o)N := 0

// Infeasible motivation

VI. Stimulate behavior for Mr(o)

D(o)N = 1
(

// Implement motivation o

GenerateAction (Mr(o))

ExecuteAction (Mr(o))

R := R <o, Mr(o)>
|

// Give up motivation o

D(o)N := 0
o :=
R :=

)
OARST = <O o, A A, R R> // Form new OAR model
Memorization (OARST)

694

On the Cognitive Processes of Human Perception with Emotions, Motivations, and Attitudes

Figure 5 The chain of motivation in a software organization

Basic
human
needs
of
individuals

Motivation

Behavior

Productivity

Organizational
objectives
Attitude

Quality

The social environment of software engineering

The results show that the manager has much

stronger motivation to improve the quality of the
software system than that of the programmers in
the given project. Therefore, the rational action
for the manager is to encourage the expectancy
of the programmers or to decrease the required
effort for the programmers by providing additional
resources or adopting certain tools.
According to sociopsychology (Wiggins,
Eiggins, & Zanden, 1994), social environment,
such as culture, ethical norms, and attitude
greatly influences peoples motivation, behavior,
productivity, and quality towards collaborative
work. The chain of individual motivation in a
software organization can be illustrated as shown
in Figure 5.
Cultures and values of a software development
organization helps to establish a set of ethical
principles or standards shared by individuals of
the organization for judging and normalizing
social behaviors. The identification of a larger set
of values and organizational policy towards social
relations may be helpful to normalize individual
and collective behaviors in a software development
organization that produces information products
for a global market.
Another condition for supporting creative
work of individuals in a software development
organization is to encourage diversity in both ways
of thinking and work allocation. It is observed in
social ecology that a great diversity of species and

a complex and intricate pattern of interactions

among the populations of a community may confer
greater stability on an ecosystem.
Definition 12: Diversity refers to the social and
technical differences of people in working organizations.
Diversity includes a wide range of differences
between people such as those of race, ethnicity,
age, gender, disability, skills, educations, experience, values, native language, and culture.
Theorem 3: The diversity principle states that
the more diversity of the workforce in an organization, the higher the opportunity to form new
relations and connections that leads to the gain
of the system fusion effect.
Theorem 3 is particularly useful for software
development organizations where creative work
products are engineered. System theory indicates
that if the number of components of a system
reaches a certain levelthe critical massthen
the functionality of the system may be dramatically increased (Wang, 2007a). That is, the increase of diversity in a system is the condition
to realize the system fusion effect, which results
in a more powerful system with newly created
relations and behaviors that only belong to the
system as a whole.

695

On the Cognitive Processes of Human Perception with Emotions, Motivations, and Attitudes

CONCLUSION

REFERENCES

This article has described the perceptual processes

of emotions, motivations, and attitudes based on
which complicated psychological and mental processes may be formally modeled and rigorously
explained. Relationships and interactions between
motivation and attitude have been formally described in RTPA. It has been recognized that the
human emotional system is a binary system that
interprets or perceives an external stimulus and/or
internal status as in the categories of pleasant or
unpleasant. It has revealed that the strength of a
motivation is proportional to both the strength
of the emotion and the difference between the
expectancy of desire and the current status of a
person and is inversely proportional to the cost to
accomplish the expected motivation. Case studies
on applications of the interactive motivation-attitude theory and cognitive processes of motivations and attitudes in software engineering have
been presented.
This work has demonstrated that the complicated human emotional and perceptual phenomena, as well as their natural drives and constraints,
can be rigorously modeled in denotational mathematics and be formally treated and described.
This work has been based on two fundamental
cognitive informatics models: the LRMB and
the OAR model. The former has provided a
blueprint to exploring the natural intelligence
and its mechanisms. The latter has established a
contextual foundation to reveal the logical representation of information, knowledge, and skills
in the perceptual space of the brain.

Eagly, A. H., & Chaiken, S. (1992). The psychology

of attitudes. San Diego, CA: Harcourt Brace.

ACKNOWLEDGMENT
The author would like to acknowledge the Natural Science and Engineering Council of Canada
(NSERC) for its support to this work. We would
like to thank the anonymous reviewers for their
valuable comments and suggestions.

696

Fazio, R. H. (1986). How do attitudes guide

behavior? In R. M. Sorrentino & E. T. Higgins
(Eds.), The handbook of motivation and cognition: Foundations of social behavior. New York:
Guilford Press.
Fischer, K. W., Shaver, P. R., & Carnochan, P.
(1990). How emotions develop and how they
organize development. Cognition and Emotion,
4, 81-127.
Fishbein, M., & Ajzen, I. (1975). Belief, attitude,
intention, and behavior: An introduction to theory
and research. Reading, MA: Addison-Wesley.
Payne, D. G., & Wenger, M. J. (1998). Cognitive
psychology. New York: Houghton Mifflin.
Pinel, J. P. J. (1997). Biopsychology (3rd ed.).
Needham Heights, MA: Allyn and Bacon.
Smith, R. E. (1993). Psychology. St. Paul, MN:
West Publishing Co.
Wang, Y. (2002a). On cognitive informatics
(Keynote Lecture). In Proceedings of the 1st IEEE
International Conference on Cognitive Informatics (ICCI02) (pp. 34-42). IEEE CS Press.
Wang, Y. (2002b). The real-time process algebra
(RTPA). The International Journal of Annals of
Software Engineering, 14, 235-274.
Wang, Y. (2003a).
Cognitive informatics: A new
transdisciplinary research field. Brain and Mind:
A Transdisciplinary Journal of Neuroscience and
115-127.
Neurophilosophy, 4(2),
Wang, Y. (2003b). On

cognitive informatics. Brain

and Mind: A Transdisciplinary Journal of Neuroscience and Neurophilosophy, 4(2), 151-167.
Wang, Y. (2003c). Using process algebra to
describe human and software behaviors. Brain

On the Cognitive Processes of Human Perception with Emotions, Motivations, and Attitudes

and Mind: A Transdisciplinary Journal of Neuroscience and Neurophilosophy, 4(2), 199-213.

Wang, Y. (2005). On the cognitive processes of
human perceptions. In Proceedings of the 4th
IEEE International Conference on Cognitive
Informatics (ICCI05) (pp. 203-211). IEEE CS
Press.
Wang, Y. (2006a). Cognitive informaticsTowards the future generation computers that think
and feel (Keynote speech). In Proceedings of the
5th IEEE International Conference on Cognitive Informatics (ICCI06) (pp. 3-7). IEEE CS
Press.
Wang, Y. (2006b). On

the informatics laws and

deductive semantics of software. IEEE Transactions on Systems, Man, and Cybernetics, 36(2),
161-171.
Wang, Y. (2006c). On abstract systems and system
algebra. In Proceedings of the 5th IEEE International Conference on Cognitive Informatics
(ICCI06) (pp. 332-343). IEEE CS Press.
Wang, Y. (2007a). Software engineering foundations: A software science perspective. CRC Software Engineering Series, 2(1), 580.
Wang, Y. (2007b). The

theoretical framework of
cognitive informatics. The International Journal
of Cognitive Informatics and Natural Intelligence,
-27.
1(1), 1

Wang, Y. (2007c). Towards theoretical foundations of autonomic computing. The International

Journal of Cognitive Informatics and Natural
Intelligence, 1(3), 1-15.
Wang, Y. (2007d). The OAR model of neural
informatics for internal knowledge pepresentation in the brain. The International Journal of
Cognitive Informatics and Natural Intelligence,
1(3), 64-75.
Wang, Y., & Wang, Y. (2006). Cognitive informatics models of the brain. IEEE Transactions on
Systems, Man, and Cybernetics, 36(2), 203-207.
Wang, Y., Wang, Y., Patel, S., & Patel, D. (2006).
A layered reference model of the brain (LRMB).
IEEE Transactions on Systems, Man, and Cybernetics, 36(2), 124-133.
Westen, D. (1999). Psychology: Mind, brain,
and culture (2nd ed.). New York: John Wiley &
Sons.
Wiggins, J. A., Eiggins, B. B., & Zanden, J. V.
(1994). Social psychology (5th ed.). New York:
McGraw-Hill.
Wilson, R. A., & Keil, F. C. (Eds.). (1999). The
MIT encyclopedia of the cognitive sciences.
Cambridge, MA: The MIT Press.
Wittig, A. F. (2001). Schaums outlines of theory
and problems of introduction to psychology (2nd
ed.). New York: McGraw-Hill.

This work was previously published in the International Journal of Cognitive Informatics and Natural Intelligence, Vol. 1,
Issue 4, edited by Y. Wang, pp. 1-13, copyright 2007 by IGI Publishing, formerly known as Idea Group Publishing (an imprint
of IGI Global).

697

698

Chapter 2.21

Sociotechnical System Design

for Learning:
Bridging the Digital Divide
with CompILE
Benjamin E. Erlandson
Arizona State University, USA

Abstract
CompILE is a sociotechnical comprehensive interactive learning environment system for personal
knowledge management and visualization that represents the growing collective knowledge an individual
gathers throughout his or her lifespan. A network of intelligent agents connects the user and his or her
inhabited knowledge space to external information sources and a multitude of fellow users. Following a
brief perspective on educational technology, concepts of human-computer interaction, and a description
of CompILE, this chapter will introduce CompILE as a sociotechnical system supported by an enriched
design process. From an educational perspective, CompILE can bridge the digital divide by creating
community, embracing culture, and promoting a learning society.

Introduction
This chapter begins with a brief perspective on
educational technology, concepts of humancomputer interaction, and a description of the
Comprehensive Interactive Learning Environment (CompILE) as a knowledge management
system controlled using a network of intelligent
software agents. Strategies for bridging the digital

divide using CompILE will be presented using

the four-part framework of the Wheel of Policy
Instruments constructed by van Dijk (2005).
The effect of CompILE on society will be approached from the standpoint of lifelong learning
and usability engineering. CompILE will create
a sense of community via participatory research,
scenario-based design, and heuristic evaluation.
In terms of culture, CompILE will rely heavily

Copyright 2008, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Sociotechnical System Design for Learning

upon ethnography to refine its ability to facilitate

cross-cultural communication, and a distinct
effort to achieve functional beauty as an interactive system should help its participants achieve a
higher level of appreciation for the aesthetics of
culture from a philosophical perspective. Globalization can be addressed by CompILE in terms
of ubiquitous computing, the mobile user, visual
literacy, and just-in-time translation. The chapter
will conclude with a description of CompILE as
a hybrid entity, and the essence of CompILE as
a sociotechnical system.
This chapter has two specific objectives:

Introducing CompILE as a sociotechnical

system supported by an enriched design
process
Using CompILE to bridge the digital divide
by creating community, embracing culture,
and promoting a learning society.

Background
One trait of humans that makes us advanced
social creatures is our ability to create artificial
devices or artifacts that expand our capabilities
(Norman, 1993). Throughout history, many of
these artifacts have been perceived as the final
piece to the ultimate puzzle of technological advancement; the last thing humans would create
to meet all their future needs. The automobile,
the airplane, the telephone, the television, the
microwave, the Internet; the list goes on. While
no single invention can actually be that final
puzzle piece, major technological advancements
cannot be denied as a driving force of society.
Norman (1993) agrees, noting the essential nature
of technology for growth of human knowledge
and mental capabilities.
An important consideration to be made when
designing any new interface is the anticipated
impact it will have on human activity. In fact,
most significant consequences of design are

based on impact to human activity. In an effort

to reduce occurrence of these consequences, the
design process should include a well-managed
team of participants with a variety of knowledge
and skills, willing to completely deconstruct any
particular problem that may arise (Carroll, 1991).
Why are these consequences so important? These
consequences have the power to make technology completely redundant, existing merely for
entertainment.
This breakpoint of redundancy is paramount in
education, which, upon clarification, is essentially
a purposed form of communication between one
or more individuals, intended to facilitate learning
and knowledge construction. Using a learningcentric approach, the application of technology
to the processes of teaching and learning can
be seen as an amplification of the channels of
communication. There is no point in seeking to
improve the educational communication process
with technology if all efforts result in glamorous
toys with no effect beyond distraction. Computers
cannot be placed in schools for an administration
to show that it is on the forefront with cuttingedge educational technology. Technology must
be carefully integrated with curricula to be used
as tools, or artifacts, fundamental to the learning process.
This concept extends beyond the classroom
geographically and temporally. Students using
the technology must be wholly engaged in the
content of the curriculum, using the technology as a mechanism of that engagement. Kling
(2000) reinforces this notion of engagement when
describing such an application of technology as
the development of a sociotechnical system,
or an information technology system that cannot
separate its existence from the social context of its
users. Once this point of engagement is reached,
the technology available today and tomorrow
can create amazing opportunities for experience
and growth within and beyond the classroom. To
reach this point, a sociotechnical system must
be designed and developed, with an eye toward

699

Sociotechnical System Design for Learning

the inseparable bond between human, machine,

and use.

Human-computer interaction
There are four main sources of knowledge underlying information systems design: (1) systematic
empirical evidence of user-interface relations,
(2) models and theories of human cognition and
artificial systems, (3) field studies of how humans
cope with tasks in real-world settings, and (4)
specific work domain knowledge derived from
textbooks, interviews with experts, and analytical methods such as task analysis (Rasmussen,
Anderson, & Bernsen, 1991). The diversity of
these sources underscores the interdisciplinary
nature of the studies and practices involved with
human-computer interaction (HCI), including but
not limited to the natural sciences, engineering,
the social sciences, and humanities.
One might assume that the foundations of HCI
are purely objective, due to the binary nature of
information systems. Computers are objective
machines based on logic, but humans are interacting with them; this completely unbalances the
equation. HCI research and development must take
into account human knowledge, decision-making
and planning, current mental representations of
the user based on causal and intentional relationships, and the fact that interactive behavior is
influenced by the environment in which it takes
place (Bechtel, 1997; Gifford, 1997; Sommer,
1969).
It is increasingly inadequate for system designers to consider the modules of the information
system in isolation from each other (Rasmussen
et al, 1991). Increasingly numerous applications
working in seamless integration on the same
computer (and across networks) becomes more
feasible with each new cycle in computer processing technology. Modules can no longer be
designed around the requirements of a narrow
group of users (Rasmussen et al, 1991). As greater

700

numbers of users become cognizant of the true

potential of personal computing, the narrow
user groups formerly associated with modules
and applications in information systems are
expanding rapidly. The expansion presents an
enormous challenge to designers; accounting for
so much more variety when building interfaces
for their products. Accounting for variety involves
the careful specification of user communities in
tandem with a thorough identification of tasks.
With novice users, the most success with this
dual process can be found through reliance on
several measurable criteria: time taken to learn
specific functions, speed of task performance,
rate of human error, subjective user satisfaction,
and human retention of functions over time (Shneiderman, 1993).
The less time users must spend learning a
system, the more time they can spend being productive. In man-made environments (interfaces),
there are two phases of activity: adaptation and
adapted. During adaptation, a users activity
is being guided by exploratory behavior. Once
adapted, a users behavioral patterns have reached
a state of equilibrium, or a satisfactory state of
transparency between user input and system
output (Rasmussen et al., 1991). The characteristics of any particular workplace and the type
of work being done determine the priority of the
two phases.

Comprehensive interactive
learning environments
The workplace of most interest to those designing interfaces for educational technology would
be the transitioning traditional classroom, which
can be propelled beyond the 21st century with
CompILE. CompILE relies heavily upon provision of access to information to all students at
all stages of the educational process, as well as
continual facilitation of mediated communication
between humans and machines. This is accom-

Sociotechnical System Design for Learning

plished through a network of intelligent agents

whose existence is focused entirely upon two
processes: (1) facilitation of learning for the user
and (2) constantly learning about the user. These
two processes are inseparable, as an integral part
of the agents facilitation of learning processes for
each user is learning the quirks of each users
processes in situations of learning (both formal
and informal). To create a personal knowledge
management system, agents must know the personality of the user almost better than the user
knows himself or herself.
As a system for personal knowledge management and visualization, CompILE can be
considered an inhabited information space, or
IIS (Snowdon, Churchill, & Frcon, 2004), that
represents the growing collective knowledge an
individual gathers throughout his or her lifespan.
To be more specific, CompILE connects the user
and his or her inhabited knowledge space (IKS).
This IKS can be navigated individually by the user
or collaboratively with additional individuals. Any
number of interfaces could be used to access this
IKS. Most of these interfaces would function as
intermediaries between the internal knowledge
architecture, external information sources, the
user(s), and associated intelligent agents.
These agents can permanently link education
and technology into a symbiotic relationship.
Agents have become fairly reliable forms of autonomous software, intertwined more actively
with the social lives of humans than any technology ever before, playing an increasingly critical
role in the organization, in particular the selforganization, of social life (Brown & Duguid,
2000). Education is a highly socialized activity,
with constant interaction between people a fundamental part of learning. A positive attribute of
software agents is their ability to deal with the
increasingly intimidating amounts of information
that people are forced to deal with on a daily basis
(Brown & Duguid, 2000). This ability would enable agents to assist students with keeping track
of much larger amounts of information.

Agents that exist today (mainly for consumer

purposes) can be broken down into three categories: information brokering, product brokering,
and merchant brokering. Information brokers
are agents that gather information from a wide
variety of resources, and perform operations
with the gathered information. Product broker
agents are designed to track their owners history
(usually shopping habits) and alert that person to
the availability of particular products in various
venues. Agents that perform merchant brokering
are essentially price comparison agents, trying to
find the best possible buy for their owner. No
successful negotiation agents (a fourth category)
have ever been created (Brown & Duguid, 2000).
Humans negotiate through nonverbal behavior in
order to accommodate themselves to one another;
constructing and adjusting the social fabric, but
software agents have yet to be built with this
capacity.
Information brokering agents would have
the most direct crossover into education, being
resourceful for students, teachers, and administrators; taking the form of research agents. Product
brokering agents would translate to learning
agents, tracking learning styles and suggesting
individual supplements and alternative approaches
to the basic curriculum for each student. Merchant
brokering agents would not necessarily be feasible
in an educational environment on a regular basis,
but a similar concept can be realized in terms of
brokering a users time. Time management agents
can assist the user by calculating approximate durations for activities, with a focus on multitasking
and efficiency. Negotiation agents are an integral
part of the CompILE system, facilitating intercommunication amongst all types of participants, with
the agents dutifully negotiating with each other
on the users behaviors as the users themselves
go about their daily routines. As a networked
system of these four types of agents, CompILE
could prove to be a vital instrument to help our
society bridge the digital divide.

701

Sociotechnical System Design for Learning

The digital divide: Wheel of

policy instruments
Van Dijk (2005) addresses the digital divide with
a four-part strategy called the Wheel of Policy
Instruments, which denotes four perspectives
for access to information and communication
technology (ICT): motivational, material, skills,
and usage. Motivational access features increasing surplus values and user-friendliness of ICTs,
organizing user trust by regulation, promoting
useful applications of ICTs, and providing specific services for underserved groups. In terms
of surplus value and user-friendliness, every use
of CompILE must be beneficial for the user, with
little or no margin for user errors that do nothing but frustrate the user, decreasing user confidence in the reliability of the CompILE system.
Additionally, certain agents can focus upon the
add-on value of a users current actions within
CompILE, giving suggestions for efficiency that
the user may or may not have noticed; thus, increasing the surplus value of the system.
Regulation and promotion of ICTs can be
seen as a symbiotic relationship. Government
and corporate institutions can work together
to promote useful applications of CompILE by
upholding regulations in an open, trustworthy
manner that promotes the benefits of CompILE
(and other applications and services) to the user.
A democratic, user-centric approach to legislation
of regulations can serve as a valuable promotional
tool as well. CompILE can help eliminate underserved groups using special agents designed to
find information and services specifically targeted
to the individual demographics of the agents
owner/user, such as cultural groups, gender- and
age-specific information, or sociopolitical, health,
and spiritual or secular information.
Material access features universal service, access to basic provisions, promotion of broadband
access, support of competition, interconnectivity
amongst media, conditional subsidies to lagging

702

groups, and creation of public access points. As

insinuated, much of materials access improvement
involves government and corporate institutions
cooperating to provide equal access to ICT services for all citizens. A good start would be the
deprivatization of broadband Internet access into
a public utility, much like public water and sewer
service or electricity co-ops. Furthermore, public
access points for broadband Internet connectivity
(both wired and wireless) would be a fundamental
element for the complete operation of CompILE.
A combination of passive RFID tags and agents
would allow for user identity and seamless integration from one access point to another, with
just-in-time delivery of information to the user.
This should all be available free of charge as part
of a public broadband infrastructure.
Skills access features connecting all schools,
libraries, hospitals, and cultural centers, adapting
curricula for operational, instrumental, and strategic skills, creating better educational software
and better digital teaching skills, teaching both
basic and advanced skills to everyone, extending
adult education to learning digital skills, and an
emphasis upon distance education at home and
in local classes or meetings. Adaptation of curricula for operational, instrumental, and strategic
ICT skills would help users meet advancing
technology at a middle ground of usability
and user-friendliness. This should also lead to
an increased knowledge base and skill sets for
those who will be maintaining and supporting
the CompILE system on a regular basis.
CompILE can reduce the need for purposed
standalone educational software applications by
providing such a dynamic, comprehensive digital
teaching and learning environment. Additionally,
CompILE can improve digital teaching skills with
the assistance of special agents for preservice
teachers while they are still in training. These
agents can be carried over with these teachers
into the in-service environment for professional
development to expand technological skills and
knowledge.

Sociotechnical System Design for Learning

As opposed to a somewhat narrowed emphasis on distance education at home and in

local classes, an emphasis should be made on a
broader inclusion of self-paced, asynchronous
collateral/supplemental learning and inquiry.
The path of learning should extend throughout
the waking hours of each individuals day. The
formation of networked learning groups could
supplement the current formal education system,
leading to a cottage industry of sorts in the
field of education and training. CompILE could
facilitate these encounters both synchronously
and asynchronously, both in-person and across
distances via ICTs.
Usage access features support for learning on
the job and at home, more hardware and software
designed for underserved groups, special content
for minorities and socially deprived groups, open
access to all public and scientific information and
major cultural events, support for competition in
software and services, promotion of open-source
software, lifelong learning and circulation of
applications in jobs and education, and the full
integration of ICTs into social and user environments. The success of CompILE depends (perhaps
most heavily) upon open access to all public and
scientific information. This information must be
available to all who wish to learn, reflecting back
upon the users trust of the systemCompILEs
agents must be able to deliver all the pertinent
information related to a subject of inquiry. Also,
upon completion of an initial performance or
showing, major cultural events should be released into the public domain. The option should
be available for immediate, free access to any
cultural event, in appropriate digital format via
the Internet.
The CompILE application will be open source
software. With the source code available for public development, advanced users would be able
to modify CompILE with add-ons or plug-ins
that can increase the benefits of CompILE use.
Similar to other open source software projects,
these add-ons will be shared with other users

across the Internet. In addition to extending the

basic internal functionality of the main CompILE
application, the add-ons would allow CompILE
to communicate with an increasing number of
current (and future) software applications. This
would allow for a much fuller integration of ICTs
into social and user environments, based on the
intended widespread use of CompILE as a tool
set for learning. The increased user-friendliness
and surplus value realized with better communication between CompILE, and other applications
would promote lifelong learning and the circulation of useful software applications in jobs and
education.

CompILE and society

The concept of lifelong learning has its official
origination in the 1973 UNESCO Faure Report
(Faure, Herrera, Kaddoura, Lopos, Petrovsky,
Rahnema, & Ward, 1972), which indicates the
worlds transition to a learning society and a
knowledge economy. Since then, many definitions of lifelong learning have surfaced. Smith and
Spurling (1999) approach lifelong learning from
a holistic perspective, with empirical and moral
elements. Concerned with the scope of lifelong
learning, the empirical element includes continuity, intention, and unfolding strategy as keys to the
learning process. The moral element deals with the
character of the learner, in terms of personal and
social commitment, respect of others, and respect
of the truth. Most importantly, Smith and Spurling
give a wide berth to the activities that fall within
the realm of lifelong learning, including formal,
informal, and self-directed learning.
Crick (2005) hints toward the autonomy of the
learner, noting that learning and metacognition are
activities that we must do for ourselves, leading to
a stronger self-awareness of and responsibility for
learning. Fischer (2001) maintains that the arena
for lifelong learning is highly contextualized and
driven by the specific yet flexible demands of

703

Sociotechnical System Design for Learning

these situations of context. Considering the wide

range of learning activities, learner autonomy, and
highly contextual situations of learning, its no
surprise that humans must embrace the concept
of a learning society.
One approach to understanding this concept of
a learning society is in terms of a social learning system, which can be understood as a constellation of communities of practice (Wenger,
2000). However, from a sociopolitical perspective,
Lambeir (2005) suggests that lifelong learning
has become a new kind of power mechanism,
highlighting two important trends: (1) a shift from
a knowledge-based society to an information
society, with learning having an emphasis upon
the retrieval, dissemination, and evaluation of
information; and (2) the increasing autonomy of
the learner. With or without consideration of the
politics involved, how can technology be applied
to such a socially expansive concept?
The Open University Knowledge Network is
under development as a software environment
for lifelong learners community knowledge network, based on Wengers community of practice
framework (McAndrew, Clow, Taylor, & Aczel,
2004). The designers chose to use evolutionary
design methods, building only essential elements
into the software, to be improved with user
feedback. From their experiences with the OU
Knowledge Network, they conclude that knowledge management software is an excellent tool
for the collection of resources and experiences
within a community, only if great care is taken
when introducing the system and in maintaining
existing work practices.
Koper and Tattersall (2004) note four main
characteristics of lifelong learning to be considered when developing ICT support networks:

704

Self-direction of learners
Mobile access to distributed information
Heterogeneity of learners
Necessity for the maintenance of individual
records of growth in competence

Their Learning Network (Koper & Sloep,

2003), developed exclusively with open source
software, is currently in the pilot stages, also at the
Open University of the Netherlands. Designed to
support seamless, ubiquitous access to learning
facilities at work, home, and schools, this architecture helps lower the access barriers to lifelong
learning (Koper & Tattersall, 2004).
Two crucial areas of research taking place simultaneously at the Center for Lifelong Learning
and Design (L3D) are knowledge-based systems
(KBS) and human-computer communication
(HCC), with the idea that success cannot fully be
realized in either area without substantial progress
in the other (Fischer, 2001). All of these projects
surrounding the application of technology in support of lifelong learning have similar indications,
working with information, autonomy and heterogeneity of the learner, improvement through user
feedback, maintaining existing work practices,
and the inseparability of KBS and HCC, lead to
a necessary focus upon the absolute importance
of the usability of applied technology.

Usability
There are five distinct attributes of usability:
learnability, efficiency, memorability, errors,
and satisfaction. An interface should be simple
enough to learn that users can quickly become
productive with the application. Efficiency is also
essential, as a user should be able to maintain
high levels of productivity with the application
upon learning the interface. The interface should
be easy to remember, so those users that are not
constantly in contact with the application can
easily reorient themselves when necessary. User
error should be kept to a minimum, and errors
that are unavoidable should have easy avenues of
recovery. The interface should be enjoyable for
the user, creating a sense of satisfaction with the
application (Nielsen, 1993).

Sociotechnical System Design for Learning

The common sense foundation of usability

makes it quite easy for a designer to forget the
actual level of importance that usability has in
the design process. The concepts become second
nature, and because of this may become secondrate. Also, it may seem as though an interface
designer for a fairly complex system would have
to choose between ease of use (learnability) and
productivity (efficiency). This choice can be
avoided by providing multiple interaction styles
along with user action accelerators (Nielsen,
1993). Through a purposed focus upon multiple
interaction styles, accelerators, or both, CompILE
must compensate completely for learnability and
efficiency. Not only should this focus increase
endearment of the user population to CompILE,
it should create a general increase in the appreciation of the usability design process amongst
members of this user community.

CompILE and community

The discovery-based design process inherent
in sociotechnical systems effectively integrates
the designers as participants within the system
itself. This integration indicates a need for change
within educational research as well. Geelan (2006)
would agree, noting that the time has come for
drastic changes in the methods of educational
research, with the need for a highly participatory
approach. This can be accomplished with an appropriate combination of scenario-based design,
heuristic evaluation, and a better understanding
of individual human participatory roles within
CompILE.

Scenario-based design
While most engineering methods seek to control
the complexity and fluidity of design using filtering and decomposition techniques, scenario-based
design techniques seek to exploit the complexity

and fluidity of design by learning more about the

structure and dynamics of the problem domain.
This process of exploitation promotes diverse
views of the situation at hand, as well as intimate
interaction with the concrete elements of that
problem domain (Carroll, 2000). To exemplify
this scenario-based process, consider an ambitious
educational administration that wishes to implement CompILE across its entire school system.
The problem domain of CompILE consists of all
the individual schools within the system and the
usual issues that these separate schools face:
curricular consistency, communication, funding,
staffing, regulations, parental involvement, and
so forth.
There are two fundamental premises of the
scenario-based approach that make it such an attractive design option. First, descriptions of how
people accomplish tasks are a primary working
design representation. Second, each context in
which humans experience and act provides detailed constraint for the development and application of information technologies (Carroll, 2000).
Considering these two premises, it is easy to see
how a scenario-based approach for designing
CompILE would radically differ from the same
type of approach for designing an accident prevention program for a nuclear power plant.
Scenarios are contexts of interaction that highlight key aspects of a particular problem domain
or system: goals suggested by the appearance
and behavior of the system; what people try to do
with the system; what procedures are adopted or
not adopted, carried out successfully or erroneously; and reflections of system participants. In
addition, these scenarios are comprised of four
characteristic elements: setting, or the place and
its props; agents/actors, or the people operating within the system; goals/objectives, or the
changes an agent/actor wishes to achieve in the
circumstances of the setting; and plot, including
sequences of actions and events that can facilitate,
obstruct, or be irrelevant to the given goals/objectives (Carroll, 2000).

705

Sociotechnical System Design for Learning

Creating and using scenarios pushes designers beyond simple static answers. The emphasis
on raising questions makes it easier for designers to integrate reflection and action into their
own design practice. The process creates constant integration between the designers and the
constituents of the problem domain by evoking
reflection, contrasting the simultaneously concrete and flexible nature of scenarios, promoting
work orientation, and melding abstraction with
categorization. Scenarios also allow analysts
and designers to visually sketch interactions in
order to probe relationships (Carroll, 2000). It is
precisely the nuances revealed through these questions and sketches that hold the keys to progress
within the design of a system like CompILE. In
addition, much of the reflection/action paradox
is resolved by scenarios, as they provide a language for action that invokes reflection (Carroll,
2000). For example, a designer concerned with
the implementation of stand-alone Web-browsing
functionality within the global navigation menu of
the CompILE interface would have a much easier
time making a primary decision if he or she knew
the current classroom dynamic between students,
teachers, and the integration of Web content as a
daily part of the curriculum.
Scenarios of use reconcile concreteness and
flexibility by embodying concrete design actions
that evoke concrete move testing by designers,
and by facilitating open-ended exploration of
design requirements and possibilities (Carroll,
2000). This approach is perfect for a project such
as CompILE, because many decisions would be
cut-and-dry, based on preceding projects within
the same vein, but the specific needs of any school
system would call for the refreshment of openended exploration. Exploration of this nature
would help avoid overapplication of technology,
as would the practical approach to interface use
and analysis; heuristic evaluation.

706

Heuristic evaluation
Heuristic evaluation is the process of a person
viewing an interface and making value judgments
based on the 10 heuristics of usability, using his
or her own common sense or intuition. Nielsen
(1993) provides the following 10 heuristics of
usability:

Simple and natural dialog

Speak the users language
Minimize user memory load
Consistency
Feedback
Clearly marked exits
Shortcuts
Good error messages
Error prevention
Help and documentation

One of the best ways to find mistakes and problems in interfaces is to use the interface and look
for them. Evaluators should work individually,
only communicating after completion, including
written or recorded reports following inspection
(Nielsen, 1993). This lowers the probability of
biased evaluations. For CompILE, individualized
evaluation might not work, since students would
be using the interface in close proximity to each
other within the classroom environment. Even
though the results would likely be biased for these
group evaluation sessions, the bias itself would be
an integral part of the feedback loop in designing
CompILE. Examining off-line communication
between students would be essential for creating seamless computerized intercommunication
functionality.
Observers are a necessity for evaluating CompILE, since the evaluators would include young
children, incapable of successfully completing
comprehensive written evaluations. However, the
observer method would still have its shortcomings, as much of the responsibility for interface
content analysis still falls upon the evaluator. Due

Sociotechnical System Design for Learning

to the inexperience that most children would have

with such an evaluation, the fact that observers
are free to give ample hints for interface operation
would be crucial to successful testing of CompILE.
The short attention spans of most school-aged
children combined with the complex nature of the
CompILE interface would necessitate the series
of short, small-scope evaluation sessions (as opposed to one marathon session). The product of
a heuristic evaluation of any interface is a list of
problems about that interface, directly referencing the 10 heuristics. When designers receive the
report, the dots between problem and solution are
mostly connected.
Unfortunately, many children might have
difficulty remembering the 10 heuristics. There
are at least three solutions to compensate for this
difficulty: (1) place posters or reference cards
of the heuristics near the child evaluator; (2)
provide simple training for any child who would
be heuristically evaluating the interface; or (3)
create a third role within the evaluation process,
a supporter. A supporter would be an observer
with more clout, helping the child with his or
her analysis as opposed to merely recording it.
However, this supporter role does not have to be
entirely fulfilled by humans. Supporters could
manifest as quality feedback agents, similar to
many software applications already in existence.
Instead of the typical reactionary role that most
of these quality feedback agents play, CompILE's
supporter agents would assume a highly proactive
role, providing an almost constant link between
the users (from all roles of participation) and the
processes of scenario-based design and heuristic
evaluation.

Roles of participation
How can these agents specifically benefit the
current and future individuals involved in the
educational system: students, teachers, parents,
administrators, researchers, and designers? Stu-

dents could benefit from research agents, tutoring agents, archival agents, and communication
agents. Research agents could constantly scour
the infinite resources of the World Wide Web,
reporting back to the student (or another agent)
whenever any relative information is discovered.
Tutoring agents would be highly customizable
personal agents that grow with the student. Much
like the aforementioned product brokering agents,
these personal tutoring agents would remember the
learning history of the student, keeping track of
patterns of learning. This would enable the tutor to
create a custom-built learning package that would
cater directly to the childs learning style.
Archival agents could track the student
throughout the day, digesting every idea communicated (and piece of work created) by the
student into a digital archive. In the CompILE
environment, every interaction in the classroom
would be recorded in digital format. These records
would be readily available, produced quickly and
quietly by the archival agents. If a student cannot
remember exactly what a teacher said or wrote
at the beginning of the day, he or she would no
longer need to worry, with everything available
on a permanent basis. Issues of storage would
need to be considered, but a half-sufficient remedy could be the use of a single storage facility
(accessible by all the students in the class) for all
in-class communication. Communication agents
representing each student could queue up for the
teacher, and he or she could answer the questions
in the order in which they were received. This
would work much like the spooling process
for a networked printer. A student could place
the requests and continue working on some other
task while his or her communication agent representative waited to be served.
The teacher could employ several intelligent
teaching assistant agents to deal with the overload
of student questions. Many students would likely
repeat the same question, so one of these assistant
agents could be created specifically to tackle this
issue, essentially replicating the teachers original

707

Sociotechnical System Design for Learning

answer for each of the students. Teachers could

employ grading agents to assist them with grading exams, especially the objective evaluations.
This portion of the grading process could be an
automatic function of CompILE, with the teachers
grading agents merely collecting the data from
the system for the teachers records. Observational agents that track the students progress
and individual learning styles could compare
notes with the teachers lesson plan preparation
agents (primarily used like research agents)
to create customized lessons for the class that
compensated for each students current position
within the learning spectrum. It is even possible
that these agents could continue working while
the lesson is being presented, providing real-time
diagnostics for the teacher, giving him or her the
ability to change the speed or direction of the
lesson on the fly!
As parents would have easy, secure access
to CompILE, teachers could send conference
requests via agents. As opposed to unanswered
e-mails or phone calls that might have arrived at
a time inconvenient to the parent, agents would
be willing to wait indefinitely for a response, dutifully delivering it back to the teacher immediately
upon receipt. Parents could also create agents to
request progress reports on their children, or to
create conferencing agents of their own, if they
wished to speak with the teacher. Messenger
agents could provide a way for parents to alert their
children (and the teachers and administration) to
any sudden change of plans that might arise on
any particular day. This would be particularly
helpful if the family has more than one child in
the same school system, but different facilities
(one in elementary school, one in middle school),
or in entirely different systems. One agent could
carry the message to all the different constituents
involved.
In addition to progress report agents, parents
could create agents that gathered their childrens
homework assignments for the day (or week),
creating a chart for easy time management,

708

helping the children finish their homework efficiently and effectively. The homework agents
could also serve as tutoring assistance agents,
gathering information to assist the parent who
might need a refresher on the fundamentals to
be learned in the homework. These agents could
report back to the teacher on the following day,
giving a brief evaluation of the previous evenings
proceedings.
Administrators would have many uses for
agents. In addition to agents already mentioned,
the administration could employ agents to monitor
teachers, creating periodical evaluations that
indicate the progress of the students; one indication
of the teachers ability to teach. Financial agents
could help the administration cope with more
political issues, such as which departments need
more funding for equipment or textbooks, with
research agents in turn scouring the resources of
the Internet for the best options. Communication
agents could serve the administration by contacting other school systems with similar demographic
issues and comparing notes.
Researchers could employ the assistance of
statistical agents that would perform data collection, sorting, and statistical analysis. Observation
agents could continually observe the actions of
different types of participants, searching for
recognizable patterns of use of the CompILE
system. As with administrators, communication
agents could assist researchers by linking to the
research taking place at other school systems with
similar and different circumstances, performing
trend comparisons and facilitating the collaboration of researchers on pertinent issues.
Design teams could rely heavily upon testing
agents for the initial alpha testing of upgrades
and new modules to be added to the CompILE
application tool set, as the purpose of alpha testing
is to find all possible ways to break software,
and agents are the perfect digital drones for simulating endless combinations of user actions in
an organized fashion. Role-playing agents could
assist designers by demonstrating alternative

Sociotechnical System Design for Learning

solutions in the scenario-based design approach.

Documentation agents could be responsible for
tracking and archiving each iteration within the
design teams process, allowing for appropriate
reflection upon completion of a project.
Special care should be given when CompILE
is implemented to prevent oversaturation of agents
within the cybersphere. Most agents should be
able to multitask. Or it is possible that the better
solution would involve millions of minimalist
agents designed to complete one simple task and
dispose of themselves. In either case, the possibilities abound for agent use in education, and the
main purpose of their existence should be to allow
humans to perform their roles in a manner that
embodies valid participation in the educational
community. Additionally, the persistence of researchers and designers on the scene in schools
should raise the comfort level of other participants
(especially children) due to familiarity. Perhaps
increased exposure to research and design processes for other participant roles would cause an
increased general appreciation of these processes.
More importantly, though, researchers would
likely gain more honest insight into participant
processes, especially on the qualitative level.

CompILE and culture

Ethnography is the process of identifying the customs, habits, and key differences between different
races, cultures, and groups of people. Due to its
small focus and personalized nature, ethnography
is known for its rich descriptions providing historical depth and contextual perspective, tracing
social processes within groups (Howard, 2002).
The rich descriptions and contextual perspective
can provide the foundation for the new form of
education that accommodates comprehensive
interactive learning environments. In terms of
communication, ethnography is concerned with
patterns and functions of communication, the
nature and definition of speech community, the

components of communicative competence, and

linguistic and social universals and inequalities
(Saville-Troike, 2003).
Patterns of communication exist on the societal, group, and individual level. The interconnectedness of these patterns largely affects the
roles of individuals in society in general, as well
as within an educational setting. If a person was
unaware that the patterns differ from one group
of people to the next, it would be easy to misunderstand the implications of a particular attempt
at communication by someone from a different
group or culture. The fact that our schools are
becoming increasingly multicultural opens a
floodgate of potential misconception.
The classroom can be viewed as a model for
societal structure in the sense that it can serve as
a microcosm of the communicative situations that
exist in our society. Considering this, the need for
more attentiveness to the context of the discourse
makes sense, but is not always possible at the time
the discourse takes place. Teachers have too many
other things to think about, and the context of
discourse cannot possibly always be their main
concern. Teachers can have more time to explain
the context of discourse to students, thanks to the
streamlined efficiency of CompILE. Addressing this context in the moment can reinforce
an open-mindedness in students that leads to a
broader appreciation of all cultures.
In addition to a broader appreciation of the
diversity of human cultures, CompILE can also
foster an overall appreciation for culture from a
more philosophical perspective, one that considers
aesthetic beauty. Adler (1981) refers to the subjective side of beauty as enjoyable beauty, and
the objectivity of beauty as admirable beauty.
Gelernter (1998) takes the concept of admirable
beauty and refines it into a more fitting term for
the technological world: machine beauty, which
can be considered a keen symmetry of simplicity
and power. Conceptually, CompILE exemplifies
this marriage of simplicity and power, providing
access to so much information in such an orga-

709

Sociotechnical System Design for Learning

nized, efficient way. However, its development

falls under the shadow of the beauty paradox:
(1) most computer technologists are oblivious to
beauty; (2) the best are obsessed with it; (3) the
public has a love-hate relationship with beauty
in computing; (4) beauty is the most important
quality that exists in the computer world, when
all is said and done (Gelernter, 1998). The subjectivity of the love-hate relationship between users
and computers is the biggest hindrance to the
development of simple yet powerful applications
such as CompILE. An application can be admirably beautiful to all its users, but some will not
experience its enjoyable beauty, at first. A deeper
appreciation for the admirable beauty can lead to
a newfound appreciation and enjoyment.

Future trends: CompILE and

globalization
The globalization of our learning society will
come about with the increased ubiquity of ICTs
(including those yet to be realized), which is
leading to the disappearance of the traditional
brick-and-mortar classroom space. CompILE
must be designed in a way to facilitate participants
responses to this removal in a seamless, beautiful manner. A core of visual communication and
literacy (Dondis, 1973), backed by a network of
translation agents, will advance the intercultural
community that would thrive in this global learning society, based entirely upon open access to
information by all participants.
As mobile inhabitants of a global society,
lifelong learners should not be held stationary
by any sociotechnical system. A learner must
be able to collaborate effectively, sharing the
contents of his or her IKS with others using a
sophisticated visualization interface. As we begin
working across national and cultural boundaries
with higher frequency, verbal communication in
a common language might not resonate clearly,
but recognition of patterns in others knowledge

710

spaces, using basic elements of visual form could

transcend this issue altogether. If we as a society
were to become visually literate, would a common
visual language lead to better understanding,
communication, and cooperative learning? Perhaps this visual communication process could
initially be augmented by translation agents, a
specialized form of negotiation agents that augment the verbal communication between people
who cannot find a common spoken language between them. These agents would be reliant upon
the advent of highly accurate voice recognition
technology, and instantaneous translation would
be paramount to provide seamless assistance during conversation.

Conclusion
As a sociotechnical system, CompILE must
manifest as a hybrid system in at least two ways.
First, its users, in all forms of participation, would
in fact be de facto designers. Second, CompILE
would combine reality and virtual environments
in a manner conducive to constructive learning
practices, taking full advantages of the added
dimensionality provided by this combination.
Imagine a learner interested in Greek civilization,
standing in modern-day Greece, able to augment
his or her current visual field with images of
classic Greek structures overlaid precisely upon
those structures in their current form. Upon the
request of the user, these augmenting images
could animate, showing the degradation of the
structures over time, highlighting access to information about specific historic events that had
direct impact on the process. This is just one
example of how CompILE could embody the
careful integration of ICT as a tool fundamental
to the learning process.
The agents of CompILE can help promote
higher levels of information literacy amongst
its users, which in time should lead to a more
informed consumer society, which will provide

Sociotechnical System Design for Learning

society with a more appropriate mindset to create

community, embrace all cultures, and become a
true learning society. The CompILE system, obviously, is a continuous work in progress, still in its
infancy. Much more discussion must take place
concerning theories and applications as a continuing predevelopment phase. It is precisely such a
discussion, as well as the continued involvement
in its design and development, that will empower
users of CompILE to cross the digital divide.

References
Adler, M. J. (1981). Six great ideas. New York:
Simon & Schuster.
Bechtel, R. B. (1997). Environment & behavior:
An introduction. Thousand Oaks, CA: Sage.
Brown, J. S., & Duguid, P. (2000). The social
life of information. Boston: Harvard Business
School Press.
Carroll, J. M. (1991). Designing interaction: Psychology at the human-computer interface. New
York: Cambridge University Press.
Carroll, J. M. (2000). Making use: Scenariobased design of human-computer interactions.
Cambridge: MIT Press.
Crick, R. D. (2005). Being a learner: A virtue for
the 21st century. British Journal of Educational
Studies, 53(3).
Dondis, D. A. (1973). A primer of visual literacy.
Cambridge: MIT Press.
Faure, E., Herrera, F., Kaddoura, A.-R., Lopos,
H., Petrovsky, A. V., Rahnema, M., & Ward, F.
C. (1972). Learning to be: The world of education
today and tomorrow. Paris: UNESCO.
Fischer, G. (n.d.). Introduction to L3D. Retrieved
December, 2006, from http://l3d.cs.colorado.
edu/introduction.html

Fischer, G. (2001). Lifelong learning and its support with new media. In N. J. Smelser & P. B.
Baltes (Eds), International encyclopedia of social
and behavioral sciences. Amsterdam: Elsevier.
Geelan, D. (2006). Undead theories: Constructivism, eclecticism and research in education.
Rotterdam: Sense Publishers.
Gelernter, D. (1998). Machine beauty: Elegance
and the heart of technology. New York: Basic
Books.
Gifford, R. (1997). Environmental psychology:
Principles and practice (2nd ed.). Boston: Allyn
and Bacon.
Howard, P. N. (2002). Network ethnography and
the hypermedia organization: New media, new
organizations, new methods. New Media and
Society, 4.
Kling, R. (2000). Learning about information
technologies and social change: The contribution
of social informatics. The Information Society,
16(3).
Koper, E. J. R., & Sloep, P. (2003). Learning
networks: Connecting people, organizations,
autonomous agents and learning resources to
establish the emergence of effective lifelong
learning. Heerlen: The Open University of the
Netherlands.
Koper, E. J. R., & Tattersall, C. (2004). New
directions for lifelong learning using network
technologies. British Journal of Educational
Technology, 35(6).
Lambeir, B. (2005). Education as liberation: The
politics and techniques of lifelong learning. Educational Philosophy and Theory, 37(3).
Marks, W., & Dulaney, C. L. (1998). Visual information processing on the World Wide Web.
In C. Forsythe, E. Grose, & J. Ratner (Eds.),
Human factors and Web development. Mahwah,
NJ: Lawrence Erlbaum.

711

Sociotechnical System Design for Learning

McAndrew, P., Clow, D., Taylor, J., & Aczel, J.

(2004). The evolutionary design of a knowledge
network to support knowledge management and
sharing for lifelong learning. British Journal of
Educational Technology, 35(6).
Nielsen, J. (1993). Usability engineering. San
Diego, CA: Morgan Kaufman.
Norman, D. A. (1993). Things that make us smart.
Reading, MA: Addison-Wesley.
Rasmussen, J., Anderson, H. B., & Bernsen, N.
O. (Eds.). (1991). Human-computer interaction.
Hillsdale, NJ: Lawrence Erlbaum.
Saville-Troike, M. (2003). The ethnography of
communication: An introduction. (3rd ed.). Malden,
MA: Blackwell.
Shneiderman, B. (Ed.). (1993). Sparks of innovation in human-computer interaction. Norwood,
NJ: Ablex.

Smith, J., & Spurling, A. (1999). Lifelong learning: Riding the tiger. London: Cassell.
Snowdon, D. N., Churchill, E. F., & Frcon, E.
(2004). Inhabited information spaces: Living with
your data. New York: Springer.
Sommer, R. (1969). Personal spaces: The behavioral basis of design. Englewood Cliffs, NJ:
Prentice Hall.
van Dijk, J. A. G. M., (2005). The deepening divide:
Inequality in the information society. Thousand
Oaks, CA: Sage.
Wenger, E. (2000). Communities of practice:
The key to knowledge strategy. In E. Lesser, M.
Fontaine, & J. Slusher, (Eds.), Knowledge and
communities, (pp. 3-20). Oxford: ButterworthHeinemann.

This work was previously published in Social Information Technology: Connecting Society and Cultural Issues, edited by T.
Kidd and I.L. Chen, pp. 348-362, copyright 2008 by Information Science Publishing (an imprint of IGI Global).

712

713

Chapter 2.22

Problem Frames for

Sociotechnical Systems
Jon G. Hall
The Open University, UK
Lucia Rapanotti
The Open University, UK

Abstract

Introduction

This chapter introduces Problem Frames as a

framework for the analysis of sociotechnical
problems. It summarizes the Problem Frames
approach, its techniques and foundations, and
demonstrates, through theory and examples, how
it can be applied to simple sociotechnical systems.
The chapter continues with the description of an
extended Problem Frame framework that allows
the treatment of more general sociotechnical problems. This extension covers social components of
a system individuals, groups or organisations
bringing them within the remit of the design
activity. The aim of the chapter is to make the
Problem Frames framework more accessible to the
software practitioner, especially those involved in
the analysis of sociotechnical problems, as these
problems have so far received only scant coverage
in the Problem Frames literature.

By sociotechnical system we mean a collection

of interacting components in which some of the
components are people and some are technological. In this chapter we focus on the requirements
analysis of sociotechnical systems in which some
of the technological subsystems are computerbased, these systems forming the largest part of
modern software design problems.
More precisely, there are two (not necessarily
disjoint) sub-classes of sociotechnical systems
that we will treat in this chapter. The first subclass contains those systems in which existing
components or sub-systems (that is, domains)
are to be allowed, through software, to interact.
An example from this first class might be the
problem of designing software for the operator
of heavy machinery. The larger second class
contains those systems for which software, a user
interface, and user instruction are to be designed
to enable a new process or service. An example

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Problem Frames for Sociotechnical Systems

of this second class might be the development of

a new customer call centre.
The use of Problem Frames (PFs) underpins
our requirements analysis process. As described
in Jackson (1998), PFs are a concretization of the
ideas of Michael Jackson and others in the separation of machine and its environments descriptions.
This separation is generally accepted as being a
useful principle for requirements analysis. We will
have cause, later in the chapter, in dealing with a
more general class of sociotechnical problems, to
further detail this separation, but nothing we do
compromises its fundamental status.
The usual representation of the separation of
machine and environment descriptions is as the
two ellipse model, illustrated in Figure 1. In
that figure world knowledge W is a description
of the relevant environment; R is the statement
of requirements; S is the specification that mediates between environment and machine; M
is the description of the machine; and P is the
program that, on machine M, implements the
specification S. The role of W is to bridge the
gap between specification S and requirements R.
More formally (Gunter, Gunter, Jackson, & Zave,
2000; Hall & Rapanotti, 2003; Zave & Jackson,
1997), W,S |-R.
One of the aims of the PF framework is to
identify basic classes of problems that recur
throughout software development. Each such
class should be captured by a problem frame that
provides a characterization for the problem class.
Sociotechnical systems are an important class of
problems and so should be representable within
the PF framework, possibly with their own (collection of) problem frame(s).
In a fundamental sense, of course, the PF
framework already deals with sociotechnical
systems: problem frames are an attempt to allow
customer and developer to come together to match
real-world problems and technological solutions.
There are many examples in Jackson (2001) as
to how this relationship can be facilitated using
problem frames.

714

We observe, however, that the application

of problem frames to particular sociotechnical
problems remains under-explored. Currently some
discussion of HCI appears in Jackson (2001), and
some analysis appears in Jackson (1998), but otherwise there is little in-depth coverage of how to
apply problem frames in this context. In this chapter we show, in some detail, how problem frames
can be applied to sociotechnical systems.
Our development is threefold. We first show
how the problem of representing interaction
with (and not just control of) technology can be
represented within the PF framework. To do this
we introduce two new basic problem frames, the
User Interaction Frame and the User Commanded
Behaviour Frame, each dealing with the class of
user-interaction problems.
Secondly we show how architectural artifacts
can be used to guide the analysis of sociotechnical
problems. To do this we discuss the notion of an
Architectural Frame (AFrame for short), a new
PF artifact that can be used to guide problem
decomposition in the light of particular solution
expertise as might, for instance, exist in a software development company. As an exemplar of
AFrames and their use, we define and apply an
AFrame corresponding to the Model View Controller (MVC) architectural style (Bass, Clements,
& Kazman, 1998).
Lastly we adapt the PF framework to meet
the needs of representing the problems of more

Figure 1. The requirements analysis model

Problem Frames for Sociotechnical Systems

complex sociotechnical systems, including those

in which, as well as user-machine interaction, user
training needs to be addressed. This causes us to
consider a reification of the two-ellipse model into
three ellipses to represent machine, environment,
and user descriptions. Consequently, by interpreting this third ellipse in the PF framework, we
discover a new type of PF domain the knowledge
domain to represent a users knowledge and their
instruction needs, and argue that this leads to
a more general PF framework for sociotechnical
systems.

Chapter Overview

for activating or deactivating the catalyst unit,

and to monitor outputs. Based on the operators
commands, the system should instruct the unit accordingly and regulate the flow of cooling water.
Attached to the system is a gearbox: whenever the
oil level in the gearbox is low, the system should
alert the operator and halt execution.
The chapter is organized as follows. The next
section develops the problem frame representation
of the chemical reactor problem and uses this to
recall the basis of problem representation in the PF
framework. The section on problem classification
provides a small taxonomy of problem classes,
including those of relevance to sociotechnical
systems. The next section addresses problem
decomposition, both in the classical PF framework
and through AFrames. The section on requirements analysis models for sociotechnical systems
details our separation into three of the various
domain descriptions and uses this to motivate a
new type of PF domain, the knowledge domain.
The final section concludes the chapter.

In the major part of this chapter we will use a

well-known chemical reactor problem (Dieste
& Silva, 2000; Leveson, 1986) as an example
for illustration of techniques. Later we briefly
describe the design of a cold-calling system.
The chemical reactor is a sociotechnical system
and is representative of the class of operatorcontrolled safety- (and mission-) critical systems.
A schematic for the chemical reactor hardware
appears in Figure 2.
A statement of the problem is as follows:

Problem Representation

A computer system is required to control the

safe and efficient operation of the catalyst unit and
cooling system of a chemical reactor. The system
should allow an operator to issue commands

We first consider the PF representation of the

chemical reactor problem. Within the PF framework problems are represented through problem
diagrams. A problem diagram defines the shape

Figure 2. The chemical reactor schematic (adapted from Dieste & Silva, 2000)

715

Problem Frames for Sociotechnical Systems

of a problem: it records the characteristic descriptions and interconnections of the parts (or
domains) of the world the problem affects; it
places the requirements in proper relationship
to the problem components; it allows a record of
concerns and difficulties that may arise in finding its solution.
For the chemical reactor, there are a number
of domains, including those that appear in the
schematic of Figure 2. Also the operator will
play an important role in issuing commands to
control the catalyst and cooling systems. Placing
all of these domains in their correct relationship
to each other leads to the problem diagram shown
in Figure3.
The components are:
Operation machine: The machine domain,
that is, the software system and its underlying hardware. The focus of the problem is
to build the Operation machine.

Other boxes (Cooling System, Catalyst,

and so forth): Given domains representing
parts of the world that are relevant to the
problem.
Shared phenomena: The ways that domains
communicate. These can include events,
entities, operations, and state information.
In Figure 3, for example, the connection
between the Operation machine and the
Cooling System is annotated by a set e,
containing the events increase_water and
decrease_water, and a set f, containing the
phenomenon water_level. Phenomena in e
are controlled by the Operation machine;
this is indicated by an abbreviation of the
domain name followed by !, that is, OM!.
Similarly the phenomenon in f is controlled
by the Cooling System, indicated by the
CS!.
The dotted oval Safe and efficient operation:
The requirement, that is, what has to be true

Figure 3. The chemical reactor problem diagram

a: {open_catalyst, close_catalyst}

b: {catalyst_status, water_level}
c: {open_catalyst, close_catalyst}
d: {is_open, is_closed}
e: {increase_water, decrease_water}

716

f: {water_level}
g: {oil_level}

Problem Frames for Sociotechnical Systems

of the world for the (operation) machine to

be a solution to the problem.
In the connections between the requirement
and the domains, a dotted line indicates
that the phenomena are referenced by (that
is, an object of) the requirement, while a
dotted arrow indicates that the phenomena
are constrained (that is, a subject for the
requirement). In Figure 3, for instance,
the oil level in the gear box is referenced
while the cooling systems water level is
constrained.
Phenomena at the requirement interface
(for example, those of sets f or d) can be,
and usually are, distinct from those at the
machine domain interface (for example,
those of sets c and e). The former are called
requirement phenomena; the latter, specification phenomena. The intuition behind
this distinction is that the requirement is
expressed in terms of elements of the problem, while the specification (that is, what
describes a machine domain) is expressed
in terms of elements of the solution.
Other artifacts that are not represented on the
problem diagram but are related to it are domain
and requirement descriptions. Such descriptions
are essential to the analysis of a problem and address relevant characteristics and behaviours of
all given domains, the machine domain, and the
requirement.
An important distinction in the PF framework
is that of separating two types of descriptions:
indicative and optative. Indicative descriptions
are those that describe how things are; optative
descriptions describe how things should be. In
this sense, in a problem diagram, given domain
descriptions are indicative, while requirement and
machine descriptions are optative. In other words
things that have to do with the problem domain
are given, while things that have to do with the
solution domain can be chosen. For instance, in the

chemical reaction problem indicative descriptions

of the Catalyst, Cooling System, and Gear Box
should include characteristics of the domains that
are of interest in the specification of the machine,
say, the mechanics of opening the catalyst, or of
changing the water level in the cooling system, or
the oil level in the gear box. On the other hand the
requirement should express some constraints on
the status and operations of those domains that,
when satisfied, result in their safe and efficient operation. Finally the machine specification should
describe how we would like the control system
to behave and interact with the given domains so
that the requirements are met.
Indeed not all domains share the same characteristics. There is a clear difference between
a cooling system and an operator. In a cooling
system there exist some predictable causal relationships among its phenomena. For instance,
it could be described as a state machine, with
a set of clearly identified states and predictable
transitions between them. A domain with such
characteristics is known as a causal domain. On
the other hand, an operators phenomena lack such
predictable causal relationships. We can describe
the actions an operator should be allowed to do
but cannot guarantee that they will be executed
in any particular order, or not at all, or that some
other (unpredicted) actions will not be executed
instead. A domain with such characteristics is
known as biddable.
The distinction between causal and biddable
domains is an important one, as it has ramifications for the type of descriptions we can provide
and the assumptions we can make in discharging
proof obligations during the analysis of a problem,
as we will see in the following sections.
Of course there exist other types of domains,
each with its own characteristics. An exhaustive
domain classification is beyond the scope of this
chapter and can be found in, for example, Jackson
(1995), Jackson (1998) and Jackson (2001).

717

Problem Frames for Sociotechnical Systems

Problem Classification

Programs

One of the intentions of the PF framework is to

classify problems. An initial problem classification
is given in Jackson (2001). Therein are identified
five basic problem frames. They are basic because
they represent relatively simple, recurrent problems in software development. In describing the
basis of problem frames, the intention is not to
be exhaustive in its classification of problems.
Indeed there are other frames by Jackson (see,
for example, Jackson, 1998) that do not make it
into that classification.
In this section we present a simple taxonomic
development, beginning with the simplest of all
problem frames and adding domain types (modulo
topology). As each type has different characteristics, the resulting problem frames represent
different classes of problems. In doing so we
introduce a new basic problem frame, the User
Interaction Frame, which is novel in the problem
class it represents within the PF framework (but
not, of course, within software engineering). The
taxonomy is summarized in Figure4.

The simplest form of problem representable in

problem frames is that of producing a program
from a given specification. This is illustrated in
Figure5. Although this is a sub-problem of all
software engineering problems, it is not a very
interesting problem class to be analyzed using
problem frames: nothing exists outside the machine. Other techniques, such as JSD (Jackson,
1997) or design patterns (Gamma, Helm, Johnson,
& Vlissides, 1995), are probably more appropriate to tackle this problem class. Indeed the PF
framework does not include a problem frame for
this class of problems.

Figure 4. Simple problem frame taxonomy: adding domain types

Embedded Controllers
For a problem class to be purposefully analyzed
in PFs, some given domain(s) of interest must exist outside the machine. An interesting problem
class is identified in Jackson (2001) by introducing a single causal domain. The problem frame
is known as the Required Behaviour Frame, and
its characterizing problem is that of building a
machine that controls the behaviour of some part
of the physical world so that it satisfies certain
conditions. Some software engineers may find it
easy to identify this problem with that of building
an embedded controller (although it does apply
also to more general software problems).
The Required Behaviour Frame is illustrated in
Figure6. The frame has a topology that is captured
by a frame diagram (that of the illustration).
The frame diagram resembles a problem diagram, but it also includes some further annotation.

Figure 5. Writing programs

718

Problem Frames for Sociotechnical Systems

This provides an indication of the characteristics of

the domains and phenomena involved in problems
of the class. For the Required Behaviour Frame:
The Controlled domain is causal (the C
annotation in the figure). Its phenomena
are also causal (indicated by C on the arcs)
they are directly caused or controlled by
a domain and may cause other phenomena
in turn.
The Control machine has access to causal
phenomena of the Controlled domain (in
C2) and controls another set of phenomena,
which are also shared with the Controlled
domain (in C1). Intuitively phenomena in
C1 are used by the machine to control the
domain, while phenomena in C2 are used
to obtain information and feedback on the
functioning and state of the domain.
The requirement, Required behaviour, is
expressed in terms of a set (C3) of causal
phenomena of the Controlled domain.
When a problem of a particular class is identified, it can be analyzed through the instantiation of the corresponding frame diagram. The

instantiation is a process of matching problem

and frames domains and their types, as well as
problem and frames phenomena types. The result
of the instantiation is a problem diagram, which
has the same topology of the frame diagram but
with domains and phenomena grounded in the
particular problem.
Let us return to the chemical reactor problem
and consider the problem of regulating the water
level in isolation. This can be regarded as a required behaviour problem (Figure7) with some
safety requirement on the water level.
For a problem to be fully analyzed, the instantiation of a problem frame is only the first step
of the process. Suitable domain and requirement
descriptions (see the section on problem representation) need to be provided and the frame
concern needs to be addressed. The frame concern
is an overall correctness argument, common to
all the problems of the class. It is the argument
that must convince you, and your customer, that
the specified machine will produce the required
behaviour once combined with the properties of
the given domains. Each problem frame comes
with a particular concern whose structure depends on the nature of the class problem. For

Figure 6. Required Behaviour Frame

Figure 7. Regulating the water level in the cooling system as a required behaviour problem

e:{increase_water,decrease_water}
f:{water_level}

719

Problem Frames for Sociotechnical Systems

the Required Behaviour Frame, the argument is

outlined in Figure8.

User Interaction
Another interesting class of problems can be
obtained by including a single biddable domain
outside the machine, which represents the user of
the system. We call the resulting problem frame
the User Interaction Frame, and its characterizing problem is that of building a machine that
enforces some rule-based interaction with the
user. The frame diagram for the User Interaction
Frame is given in Figure9.
The Interaction machine is the machine to be
built. The User is a biddable domain representing
the user who wants to interact with the machine.
The requirement gives the rules that establish
legal user/machine interactions.
The manifestation of the user/machine interaction is through exchanges of causal phenomena

(in C1) controlled by the user and symbolic

phenomena (in Y1) controlled by the machine.
Intuitively the user issues commands in C1 and
the machine provides feedback through Y1. The
interaction rules specify the legal correspondence
of user and machine phenomena.
In the chemical reactor problem we can isolate
the operator/machine interaction as a user interaction problem as illustrated in Figure10, where
the requirement establishes some rules on the
relationship between operators commands and
system feedback, say, that an open_catalyst command cannot be issued when the water_level value
is below a set threshold. Note that this would be
the perspective of a User Interface (UI) designer,
whose main concern is the user interacting with
a black box system. Indeed, in the wider problem
analysis of the chemical reactor problem, machine
responses to user commands depend on a faithful
representation of the internal state of, say, the
catalyst or the cooling system.

Figure 8. Frame concern for the Required Behaviour Frame

1
2
3

We will build a machine that behaves like this, so that...

knowing that the controlled domain works like this...
we can be sure that its behaviour will be this.

Figure 9. User Interaction Frame

720

Problem Frames for Sociotechnical Systems

Figure11 illustrates the frame concern for the

User Interaction Frame.

User Commanded Behaviour

The Required Behaviour and the User Interaction
Frames are representative of relatively simple
problems, albeit often recurring in software
development. It is possible, and indeed likely,
that other interesting problem classes could be
identified by considering single given domains
of some other type (see discussion at the end of
the section on problem representation). However
we do not pursue this any further here. We look,
instead, at what happens when there are two given
domains outside the machine.

As for the single domain case, other interesting

classes of problems emerge. In fact the remaining
four basic problem frames introduced in Jackson
(2001) are all of this form.
If we add a biddable domain to the Requirement Behaviour Frame, we obtain a User Commanded Behaviour Frame, which is illustrated
in Figure12. Its characterizing problem is that
of building a machine that will accept the users
commands, impose control on some part of the
physical world accordingly, and provide suitable
feedback to the user. Jackson (2001) introduces a
subclass of this frame, the Commanded Behaviour
Frame, which does not require the user to receive
any feedback.

Figure 10. Operator/system interaction as an instance of the User Interaction Frame

a: {open_catalyst, close_catalyst}

b: {catalyst_status, water_level}

Figure 11. Frame concern for the User Interaction Frame

1
Given this set of machine phenomena, when the user causes this phenomena (it
may or may not be sensible or viable)...
2
if sensible or viable the machine will accept it...
3
resulting in this set of machine phenomena...
4
thus achieving the required interaction in every case.

721

Problem Frames for Sociotechnical Systems

Figure 12. User Commanded Behaviour Frame

Figure 13. Controlling the catalyst as an instance of a user commanded behaviour problem

a: {open_catalyst, close_catalyst}
c: {open_catalyst, close_catalyst}

In the chemical reactor problem, we can apply the User Commanded Behaviour Frame to
analyze how the catalyst is controlled by the
operator. The corresponding problem diagram
is given in Figure13.
A possible description of the interaction rules
could be as follows. The machine shall allow the
user to control the catalyst under the following
constraints:
1. catalyst_status is a faithful representation
of the state of the catalyst
2. the initial state of the catalyst is catalyst_
closed
3. possible user commands are open_catalyst
or close_catalyst
722

b: {catalyst_status}
d: {is_open, is_closed}

4. state transitions are represented in Figure14.

The frame concern for the User Commanded
Behaviour Frame is given in Figure15. From
the figure you will notice that the argument has
two parts: satisfying the required behaviour of
the domain (from 1 to 4) and providing suitable
feedback to the user (5 and 6).

Problem Decomposition
Most real problems are too complex to fit basic
problem frames. They require, rather, the structur-

Problem Frames for Sociotechnical Systems

Figure 14. State machine model for the catalyst

Figure 15. The frame concern for the User Commanded Behaviour Frame

1
Given a choice of commands in the current state, when the user issues this
command (it may or may not be sensible).
2
if sensible or viable, the machine will cause these events...
3
resulting in this state or behaviour...
4
which satisfies the requirement...
5
and which the machine will relate to the user...
6
thus satisfying the requirement in every case.

ing of the problem as a collection of (interacting)

sub-problems. In this section we discuss two
ways of decomposing problems within the PF
framework. The first, classical decomposition,
proceeds through sub-problem identification and
problem frame instantiation. The second, our
novel approach, combines sub-problem identifi-

cation with guided architectural decomposition

using AFrames.

Classical Decomposition
In classical PF decomposition a problem is decomposed into simpler constituent sub-problems that

723

Problem Frames for Sociotechnical Systems

can then be analysed separately. If necessary each

sub-problem can be decomposed further, and so
forth, until only very simple sub-problems remain.
Decomposition proceeds through the identification of sub-problems that fit a recognised problem
class and the instantiation of the corresponding
problem frame. We illustrate the process on the
chemical reactor problem.
There are three sub-problems:
1. a user-commanded behaviour problem, for
the operator to control the catalyst;
2. a required behaviour problem, for regulating
the water flow in the cooling system; and

3. a sub-problem to issue a warning (and halt

the system) when there is an oil leak in the
gearbox.
Addressing sub-problems 1 and 2 means
instantiating the corresponding problem frames
to derive problem diagrams for each sub-problem. These were depicted in Figures13 and 7,
respectively.
The third sub-problem has no standard problem
frame to represent it. The closest fit would be the
Information Display Frame (Jackson, 2001), but
this requires a decision on how the alarm will be
raised. Here we have made an arbitrary choice of
introducing a Bell domain, and assume that it will

Figure 16. Raising the alarm as an instance of the information display problem

h:{ring_bell}

i:{bell_ringing}

g:{oil_level}

Figure 17. Further decomposition of the user Commanded Behaviour sub-problem

a: {open_catalyst, close_catalyst}
c: {open_catalyst, close_catalyst}

724

b: {catalyst_status}
d: {is_open, is_closed}

Problem Frames for Sociotechnical Systems

ring when the oil level in the gearbox is below

a certain threshold. The resulting sub-problem
diagram is shown in Figure16.
Finally we already know from the simple
taxonomy in the third section that sub-problem
one can be decomposed further, resulting in a
required behaviour and a user interaction subproblem. These are shown in Figure17.

AFrames
AFrame decomposition complements classical
decomposition in providing guidance and decomposition and recomposition rules. The rational
behind AFrames is the recognition that solution
structures can be usefully employed to inform
problem analysis.
AFrames characterise the combination of
a problem class and an architectural class. An
AFrame should be regarded as a problem frame for
which a standard sub-problem decomposition
(that implied by an architecture or architectural
style) exists. AFrames are a practical tool for subproblem decomposition that allow the PF practitioner to separate and address, in a systematic
fashion, the concerns arising from the intertwining
of problems and solutions, as has been observed
to take place in industrial software development
(Nuseibeh, 2001). Further motivation for, and other
examples of, AFrames can be found in Rapanotti,
Hall, Jackson, & Nuseibeh (2004).

The MVC (short for Model-View-Controller)

(see, for example, Bass et al., 1998) is a way of
structuring a software solution into three parts
a model, a view, and a controller to separate
and handle concerns related, respectively, to
the modeling of a domain of interest, the visual
feedback to the user, and the user input. The controller interprets user inputs and maps them into
commands to the model to effect the appropriate
change. The model manages one or more data
elements, responds to queries about its state, and
responds to instructions to change state. The view
is responsible for feedback on the models state to
the user. Standard communication patterns (for
example, the Observer pattern (Gamma et al.,
1995)) apply between the MVC parts.
Here we introduce the MVC AFrame as applied to the User Commanded Behaviour Frame.
This represents the class of user commanded
behaviour problems for which an MVC solution
is to be provided.
The intention of using the MVC in the solution space is recorded through an annotation of
the machine as illustrated in Figure18. Guidance
on decomposition is in the form of decomposition
templates, which are applied to obtain sub-problem diagrams. The decomposition templates for
the MVC AFrame are given in Figure19. It can
be seen from the figure that the original problem
is decomposable into two sub-problems, whose

Figure 18. MVC annotation of the User Commanded Behaviour Frame

725

Problem Frames for Sociotechnical Systems

Figure 19. Decomposition templates for the MVC AFrame

machine domains are the View and Controller

machines (in the MVC sense). Also a Model domain is introduced that represents an abstraction
of the real-world domain to be controlled. This is
a designed domain (Jackson, 2001), that is, one
that we have the freedom to design, as it will
reside inside the solution machine. The resulting
sub-problems are then: that of building a View
machine to display the Models representation
of the state of the controlled domain and that of
building a Controller machine that acts on the
Model, which will pass on the commands to the
controlled domain. In PF terms the Model acts as a
connection domain between the real-world domain
and presentation and control subsystems.
The application of the MVC AFrame to the subproblem of controlling the catalyst (see Figure13)
results in the decomposition of Figure20.
We see at least two strengths of AFrames. The
first is that they suggest how a problem would need

726

to be restructured for a particular solution form.

For instance, in the MVC case, that an abstract
model of the catalyst needs to be produced (or,
for that matter, a connection domain between
Operator and Gearbox a Bell would be
needed).
The second is that they help the recomposition of sub-problem solutions into the original
problem. Recomposition is facilitated by the
fact that AFrame decomposition is regularized
through the application of the AFrame templates.
For the MVC this is through the identification of
the links among its architectural elements. The
recomposition diagram for the MVC AFrame is
illustrated in Figure21 and its frame concern in
Figure22. In Figure 21 the recomposition of the
model, view, and controller domains follows the
MVC architectural rules.

Problem Frames for Sociotechnical Systems

Figure 20. MVC decomposition of the user commanded behaviour sub-problem

a: {open_catalyst, close_catalyst}
c: {open_catalyst, close_catalyst}
e: {open, closed}

b: {catalyst_status}
d: {is_open, is_closed}

Figure 21. MVC recomposition

A Requirements Analysis
Model for Sociotechnical
Systems
The consideration of more sophisticated humanmachine relationships is our next concern. To be
specific we now wish to look at users behaviour
as being the subject of requirements statements,

admitting that users are the source of much of the

flexibility in sociotechnical systems. In short we
wish to allow the design of human instruction
to be the subject of the requirements engineering process addressed through problem frames
alongside that of the program.
Paraphrasing this we might say that the human, as well as the machine, is to be the subject of

727

Problem Frames for Sociotechnical Systems

Figure 22. Discharging the correctness argument in MVC recomposition

1
Given a choice of commands in the current state, when the user issues this
command (it may or may not be sen-sible)...
2
if sensible or viable, the machine will cause these events...
3
resulting in this state or behaviour...
4
which satisfies the requirement...
5
and which the machine will relate to the user...
6
thus satisfying the requirement in every case.

optative descriptions. Foundationally this means

the separation of the description of the world from
that of the human who is the subject of the design.
This leads to the reification of the original ellipse
model shown in Figure 23. In it we have three ellipses that for the Human H with knowledge K,
that for the Machine M with program P, and that
for the remaining Environment W with requirements R. Of course, just as machines outside the
design process have indicative descriptions in W,
so do humans.
With the introduction of the human H, we
identify and separate two new areas of interest,
which now form explicit foci for design:
the specification UI, anonymous in the S
region in the original model, which deter-

728

mines the Human-Machine interface; and

the specification I, missing from the original
model, which determines the knowledge and
behaviour that is expected of the human as
a component of the sociotechnical system.
As in the original model the description W
has the role of bridging the gaps between the
requirement R and the specification S, in our
extension W has the role of bridging the gaps
between the requirement R and the instruction I,
human-machine interface UI and specification S
together. More precisely we assert that S, I, UI,
and W must be sufficient to guarantee that the
requirements of the sociotechnical system are
satisfied. More formally:

Problem Frames for Sociotechnical Systems

Figure 23. The extended requirements analysis

model

Figure 24. The general sociotechnical problem

diagram

Figure 25. Outline problem diagram for the coldcalling system

a knowledge domain to represent that domain. In

a problem diagram a knowledge domain should
be represented as a domain box with a double bar
on the right-hand side (to distinguish it from the
machine domain).
The most general form of a sociotechnical
problem, as a problem diagram, is shown in
Figure24. In the figure both Knowledge and
Machine domains are subjects of design, as are
their shared user interface phenomena.
An example of how a real-world sociotechnical problem could be treated in the new model
is that of designing a cold-calling system to
allow an interviewee to be telephoned and for
their responses to certain questions, asked by
an interviewer, to be recorded in a database for
future analysis. The problem is to design both
the technical subsystem (the machine domain)
and the instructions that guide the interaction of
the interviewer (the knowledge domain) with the
interviewee. The interviewee sits firmly in the
environment (as a biddable, indicative domain).
The interaction with the database is a machine
task.
The outcome of the design process, in addition
to the specification for the technical subsystem,
might be a script for the interviewer and the human-machine interface as used by the interviewer.

W,S,I,UI |- R

A Problem Frame Interpretation

In the PFs framework the machine domain represents the machine for which the specification S
must be designed. By analogy a new domain type
will be required to represent the human for which
the instruction I has to be designed. To this end
we introduce into the PF framework the notion of

729

Problem Frames for Sociotechnical Systems

The problem diagram for this example is outlined

in Figure 25.

bring the sociotechnical system problems under

the design remit of the PF framework.

Conclusion

Acknowledgment

In their classical form problem frames happily represent interactions between a user and
a machine, as might be characteristic of simple
sociotechnical systems. In this chapter we have
presented an enrichment of the PF framework
to allow the representation and analysis of more
complex sociotechnical systems. To do this we
have introduced two new problem frames, the
User Interaction and User Commanded Behaviour
Frames. Although not exhaustive in their treatment of socio-technological interaction problems,
they hopefully will provide a sound basis for a
richer taxonomy of user interaction within the
PF framework.
One of the as-yet under-developed areas within
the PF framework is the treatment of problem decomposition, in particular from the perspective of
how to do it in practice. We are currently exploring
the development and use of AFrames. An AFrame
offers guidance in problem decomposition on the
basis of solution space structures. In this chapter
we have shown how basic sociotechnical interaction problems can be decomposed when the target
architecture is to be the MVC style.
Although both these enrichments are new in
the PF framework, they do not move outside of
its original conceptual basis in the two-ellipse
model of requirements analysis. In contrast we
have seen in this chapter that the development of
general sociotechnical systems raises challenges
for the PF framework. We have suggested solutions to these challenges in the reification of the
two-ellipse model to a three-ellipse version, in
which social sub-systems individuals, groups,
and organisations can also be considered as the
focus of the design process. With the introduction of the knowledge domain, the manifestation
of this extension in problem frames, we aim to

We acknowledge the kind support of our colleagues, especially Michael Jackson and Bashar
Nuseibeh in the Department of Computing at the
Open University.

730

References
Bass, L., Clements, P., & Kazman, R. (1998).
Software architecture in practice. SEI Series in
Software Engineering. Addison Wesley.
Dieste, O., & Silva, A. (2000). Requirements:
Closing the gap between domain and computing
knowledge. Proceedings of SCI2000, II.
Gamma, E., Helm, R., Johnson, R., & Vlissides,
J. (1995). Design patterns. Addison-Wesley.
Gunter, C., Gunter, E., Jackson, M., & Zave, P.
(2000). A reference model for requirements and
specifications. IEEE Software, 3(17), 37-43.
Hall, J.G., & Rapanotti, L. (2003). A reference
model for requirements engineering. Proceedings
of the 11th International Conference of Requirements Engineering, 181-187.
Jackson, M. (1995). Software requirements &
specifications: A lexicon of practice, principles,
and prejudices. Addison-Wesley.
Jackson, M. (1997). Principles of program design.
Academic Press.
Jackson, M. (2001). Problem frames. AddisonWesley.
Jackson, M.A. (1998). Problem analysis using
small problem frames [Special issue]. South African Computer Journal, 22, 47-60.

Problem Frames for Sociotechnical Systems

Leveson, N. (1986). Software safety: What, why

and how. ACM Computing Surveys, 18(2), 125163.
Nuseibeh, B. (2001). Weaving together requirements and architectures. IEEE Computer, 34(3),
115-117.

position. Proceedings of the 12th International

Conference of Requirements Engineering.
Zave, P., & Jackson, M. (1997). Four dark corners
of requirements engineering. ACM Transactions
on Software Engineering and Methodology, 6(1),
1-30.

Rapanotti, L., Hall, J.G., Jackson, M., & Nuseibeh,

B. (2004). Architecture-driven problem decom-

This work was previously published in Requirements Engineering for Sociotechnical Systems, edited by J. L. Mate & A. Silva,
pp. 318-339, copyright 2005 by Information Science Publishing (an imprint of IGI Global).

731

732

Chapter 2.23

Integrating Semantic
Knowledge with Web Usage
Mining for Personalization
Honghua Dai
DePaul University, USA
Bamshad Mobasher
DePaul University, USA

Abstract
Web usage mining has been used effectively as
an approach to automatic personalization and as
a way to overcome deficiencies of traditional approaches such as collaborative filtering. Despite
their success, such systems, as in more traditional
ones, do not take into account the semantic knowledge about the underlying domain. Without such
semantic knowledge, personalization systems
cannot recommend different types of complex
objects based on their underlying properties and
attributes. Nor can these systems possess the
ability to automatically explain or reason about
the user models or user recommendations. The
integration of semantic knowledge is, in fact,
the primary challenge for the next generation
of personalization systems. In this chapter we
provide an overview of approaches for incorporating semantic knowledge into Web usage mining
and personalization processes. In particular, we

discuss the issues and requirements for successful

integration of semantic knowledge from different
sources, such as the content and the structure of
Web sites for personalization. Finally, we present
a general framework for fully integrating domain
ontologies with Web usage mining and personalization processes at different stages, including
the preprocessing and pattern discovery phases,
as well as in the final stage where the discovered
patterns are used for personalization.

Introduction
With the continued growth and proliferation of
e-commerce, Web services, and Web-based information systems, personalization has emerged
as a critical application that is essential to the
success of a Website. It is now common for Web
users to encounter sites that provide dynamic
recommendations for products and services, tar-

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Integrating Semantic Knowledge with Web Usage Mining for Personalization

geted banner advertising, and individualized link

selections. Indeed, nowhere is this phenomenon
more apparent as in the business-to-consumer
e-commerce arena. The reason is that, in todays
highly competitive e-commerce environment,
the success of a site often depends on the sites
ability to retain visitors and turn casual browsers
into potential customers. Automatic personalization and recommender system technologies have
become critical tools, precisely because they help
engage visitors at a deeper and more intimate level
by tailoring the sites interaction with a visitor to
her needs and interests.
Web personalization can be defined as any
action that tailors the Web experience to a particular user, or a set of users (Mobasher, Cooley
& Srivastava, 2000a). The experience can be
something as casual as browsing a Website or
as (economically) significant as trading stocks
or purchasing a car. Principal elements of Web
personalization include modeling of Web objects
(pages, etc.) and subjects (users), categorization
of objects and subjects, matching between and
across objects and/or subjects, and determination of the set of actions to be recommended
for personalization. The actions can range from
simply making the presentation more pleasing
to anticipating the needs of a user and providing
customized information.
Traditional approaches to personalization
have included both content-based and userbased techniques. Content-based techniques use
personal profiles of users and recommend other
items or pages based on their content similarity
to the items or pages that are in the users profile.
The underlying mechanism in these systems is
usually the comparison of sets of keywords representing pages or item descriptions. Examples of
such systems include Letizia (Lieberman, 1995)
and WebWatcher (Joachims, Freitag & Mitchell,
1997). While these systems perform well from
the perspective of the end user who is searching
the Web for information, they are less useful in
e-commerce applications, partly due to the lack

of server-side control by site owners, and partly

because techniques based on content similarity
alone may miss other types of semantic relationships among objects (for example, the associations
among products or services that are semantically
different, but are often used together).
User-based techniques for personalization, on
the other hand, primarily focus on the similarities
among users rather than item-based similarities.
The most widely used technology user-based
personalization is collaborative filtering (CF)
(Herlocker, Konstan, Borchers & Riedl, 1999).
Given a target users record of activity or preferences, CF-based techniques compare that record
with the historical records of other users in order
to find the users with similar interests. This is the
so-called neighborhood of the current user. The
mapping of a visitor record to its neighborhood
could be based on similarity in ratings of items,
access to similar content or pages, or purchase of
similar items. The identified neighborhood is then
used to recommend items not already accessed or
purchased by the active user. The advantage of this
approach over purely content-based approaches
that rely on content similarity in item-to-item
comparisons is that it can capture pragmatic
relationships among items based on their intended
use or based on similar tastes of the users.
The CF-based techniques, however, suffer
from some well-known limitations (Sarwar,
Karypis, Konstan & Riedl, 2000). For the most
part these limitations are related to the scalability and efficiency of the underlying algorithms,
which requires real-time computation in both the
neighborhood formation and the recommendation phases. The effectiveness and scalability
of collaborative filtering can be dramatically
enhanced by the application of Web usage mining techniques.
In general, Web mining can be characterized
as the application of data mining to the content,
structure, and usage of Web resources (Cooley,
Mobasher & Srivastava, 1997; Srivastava, Cooley,
Deshpande & Tan, 2000). The goal of Web min-

733

Integrating Semantic Knowledge with Web Usage Mining for Personalization

ing is to automatically discover local as well as

global models and patterns within and between
Web pages or other Web resources. The goal of
Web usage mining, in particular, is to capture
and model Web user behavioral patterns. The
discovery of such patterns from the enormous
amount of data generated by Web and application
servers has found a number of important applications. Among these applications are systems to
evaluate the effectiveness of a site in meeting user
expectations (Spiliopoulou, 2000), techniques
for dynamic load balancing and optimization of
Web servers for better and more efficient user
access (Palpanas & Mendelzon, 1999; Pitkow &
Pirolli, 1999), and applications for dynamically
restructuring or customizing a site based on users predicted needs and interests (Perkowitz &
Etzioni, 1998).
More recently, Web usage mining techniques
have been proposed as another user-based approach to personalization that alleviates some
of the problems associated with collaborative
filtering (Mobasher et al., 2000a). In particular,
Web usage mining has been used to improve the
scalability of personalization systems based on
traditional CF-based techniques (Mobasher, Dai,
Luo & Nakagawa, 2001, 2002).
However, the pure usage-based approach to
personalization has an important drawback: the
recommendation process relies on the existing
user transaction data, and thus items or pages
added to a site recently cannot be recommended.
This is generally referred to as the new item
problem. A common approach to resolving this
problem in collaborative filtering has been to
integrate content characteristics of pages with the
user ratings or judgments (Claypool et al., 1999;
Pazzani, 1999). Generally, in these approaches,
keywords are extracted from the content on the
Website and are used to either index pages by
content or classify pages into various content
categories. In the context of personalization, this
approach would allow the system to recommend
pages to a user, not only based on similar users,

734

but also (or alternatively) based on the content

similarity of these pages to the pages the user
has already visited.
Keyword-based approaches, however, are
incapable of capturing more complex relationships among objects at a deeper semantic level
based on the inherent properties associated with
these objects. For example, potentially valuable
relational structures among objects such as relationships between movies, directors, and actors,
or between students, courses, and instructors, may
be missed if one can only rely on the description
of these entities using sets of keywords. To be able
to recommend different types of complex objects
using their underlying properties and attributes,
the system must be able to rely on the characterization of user segments and objects, not just
based on keywords, but at a deeper semantic level
using the domain ontologies for the objects. For
instance, a traditional personalization system on
a university Website might recommend courses in
Java to a student, simply because that student has
previously taken or shown interest in Java courses.
On the other hand, a system that has knowledge of
the underlying domain ontology might recognize
that the student should first satisfy the prerequisite
requirements for a recommended course, or be
able to recommend the best instructor for a Java
course, and so on.
An ontology provides a set of well-founded
constructs that define significant concepts and
their semantic relationships. An example of an
ontology is a relational schema for a database
involving multiple tables and foreign keys semantically connecting these relations. Such
constructs can be leveraged to build meaningful
higher-level knowledge in a particular domain.
Domain ontologies for a Website usually include
concepts, subsumption relations between concepts (concept hierarchies), and other relations
among concepts that exist in the domain that the
Web site represents. For example, the domain
ontologies of a movie Website usually include
concepts such as movie, actor, director,

Integrating Semantic Knowledge with Web Usage Mining for Personalization

theater, and so forth. The genre hierarchy can

be used to represent different categories of movie
concepts. Typical relations in this domain may
include Starring (between actors and movies),
Directing, Playing (between theaters and
movies), and so forth.
The ontology of a Website can be constructed
by extracting relevant concepts and relations from
the content and structure of the site, through machine learning and Web mining techniques. But,
in addition to concepts and relations that can be
acquired from Web content and structure information, we are also interested in usage-related
concepts and relations in a Website. For instance,
in an e-commerce Website, we may be interested
in the relations between users and objects that
define different types of online activity, such as
browsing, searching, registering, buying, and
bidding. The integration of such usage-based relations with ontological information representing
the underlying concepts and attributes embedded
in a site allows for more effective knowledge
discovery, as well as better characterization and
interpretation of the discovered patterns.
In the context of Web personalization and
recommender systems, the use of semantic knowledge can lead to deeper interaction of the visitors
or customers with the site. Integration of domain
knowledge allows such systems to infer additional
useful recommendations for users based on more
fine grained characteristics of the objects being
recommended, and provides the capability to
explain and reason about user actions.
In this chapter we present an overview of the
issues related to and requirements for successfully
integrating semantic knowledge in the Web usage
mining and personalization processes. We begin
by providing some general background on the use
of semantic knowledge and ontologies in Web
mining, as well as an overview of personalization
based on Web usage mining. We then discuss
how the content and the structure of the site can
be leveraged to transform raw usage data into
semantically-enhanced transactions that can be

used for semantic Web usage mining and personalization. Finally, we present a framework for more
systematically integrating full-fledged domain
ontologies in the personalization process.

Background
Semantic Web Mining
Web mining is the process of discovering and extracting useful knowledge from the content, usage,
and structure of one or more Web sites. Semantic
Web mining (Berendt, Hotho & Stumme, 2002)
involves the integration of domain knowledge
into the Web mining process.
For the most part the research in semantic
Web mining has been focused in application
areas such as Web content and structure mining. In this section, we provide a brief overview
and some examples of related work in this area.
Few studies have focused on the use of domain
knowledge in Web usage mining. Our goal in this
chapter is to provide a road map for the integration
of semantic and ontological knowledge into the
process of Web usage mining, and particularly,
in its application to Web personalization and
recommender systems.
Domain knowledge can be integrated into the
Web mining process in many ways. This includes
leveraging explicit domain ontologies or implicit
domain semantics extracted from the content or
the structure of documents or Website. In general, however, this process may involve one or
more of three critical activities: domain ontology
acquisition, knowledge base construction, and
knowledge-enhanced pattern discovery.

Domain Ontology Acquisition

The process of acquiring, maintaining and enriching the domain ontologies is referred to as
ontology engineering. For small Web sites with
only static Web pages, it is feasible to construct
a domain knowledge base manually or semi735

Integrating Semantic Knowledge with Web Usage Mining for Personalization

manually. In Loh, Wives and de Oliveira (2000),

a semi-manual approach is adopted for defining
each domain concept as a vector of terms with the
help of existing vocabulary and natural language
processing tools.
However, manual construction and maintenance of domain ontologies requires a great deal
of effort on the part of knowledge engineers, particularly for large-scale Websites or Websites with
dynamically generated content. In dynamically
generated Websites, page templates are usually
populated based on structured queries performed
against back-end databases. In such cases, the
database schema can be used directly to acquire
ontological information. Some Web servers send
structured data files (e.g., XML files) to users and
let client-side formatting mechanisms (e.g., CSS
files) work out the final Web representation on client agents. In this case, it is generally possible to
infer the schema from the structured data files.
When there is no direct source for acquiring
domain ontologies, machine learning and text
mining techniques must be employed to extract
domain knowledge from the content or hyperlink
structure of the Web pages. In Clerkin, Cunningham and Hayes (2001), a hierarchical clustering
algorithm is applied to terms in order to create
concept hierarchies. In Stumme, Taouil, Bastide,
Pasquier and Lakhal (2000) a Formal Concept
Analysis framework is proposed to derive a
concept lattice (a variation of association rule
algorithm). The approach proposed in Maedche
and Staab (2000) learns generalized conceptual
relations by applying association rule mining.
All these efforts aim to automatically generate
machine understandable ontologies for Website
domains.
The outcome of this phase is a set of formally
defined domain ontologies that precisely represent the Website. A good representation should
provide machine understandability, the power of
reasoning, and computation efficiency. The choice
of ontology representation language has a direct
effect on the flexibility of the data mining phase.

736

Common representation approaches are vectorspace model (Loh et al., 2000), descriptive logics
(such as DAML+OIL) (Giugno & Lukasiewicz,
2002; Horrocks & Sattler, 2001), first order logic
(Craven et al., 2000), relational models (Dai &
Mobasher, 2002), probabilistic relational models
(Getoor, Friedman, Koller & Taskar, 2001), and
probabilistic Markov models (Anderson, Domingos & Weld, 2002).

Knowledge Base Construction

The first phase generates the formal representation of concepts and relations among them. The
second phase, knowledge base construction, can
be viewed as building mappings between concepts
or relations on the one hand, and objects on the
Web. The goal of this phase is to find the instances
of the concepts and relations from the Websites
domain, so that they can be exploited to perform
further data mining tasks. Learning algorithms
plays an important role in this phase.
In Ghani and Fano (2002), a text classifier is
learned for each semantic feature (somewhat
equivalent to the notion of a concept) based on
a small manually labeled data set. First, Web
pages are extracted from different Websites that
belong to a similar domain, and then the semantic
features are manually labeled. This small labeled
data set is fed into a learning algorithm as the
training data to learn the mappings between
Web objects and the concept labels. In fact, this
approach treats the process of assigning concept
labels as filling missing data. Craven et al.
(2000) adopt a combined approach of statistical
text classification and first order text classification
in recognizing concept instances. In that study,
learning process is based on both page content
and linkage information.

Knowledge-Enhanced Web Data

Mining
Domain knowledge enables analysts to perform
more powerful Web data mining tasks. The ap-

Integrating Semantic Knowledge with Web Usage Mining for Personalization

plications include content mining, information

retrieval and extraction, Web usage mining, and
personalization. On the other hand, data mining
tasks can also help to enhance the process of
domain knowledge discovery.
Domain knowledge can improve the accuracy of document clustering and classification
and induce more powerful content patterns. For
example, in Horrocks (2002), domain ontologies
are employed in selecting textual features. The
selection is based on lexical analysis tools that
map terms into concepts within the ontology. The
approach also aggregates concepts by merging the
concepts that have low support in the documents.
After preprocessing, only necessary concepts are
selected for the content clustering step. In McCallum, Rosenfeld, Mitchell and Ng (1998), a concept
hierarchy is used to improve the accuracy and the
scalability of text classification.
Traditional approaches to content mining and
information retrieval treat every document as a set
or a bag of terms. Without domain semantics, we
would treat human and mankind as different
terms, or, brake and car as unrelated terms.
In Loh et al. (2000), a concept is defined as a
group of terms that are semantically relevant, for
example, as synonyms. With such concept definitions, concept distribution among documents is
analyzed to find interesting concept patterns. For
example, one can discover dominant themes in a
document collection or in a single document; or
find associations among concepts.
Ontologies and domain semantics have been
applied extensively in the context of Web information retrieval and extraction. For example, the
ARCH system (Parent, Mobasher & Lytinen,
2001) adopts concept hierarchies because they
allow users to formulate more expressive and
less ambiguous queries when compared to simple
keyword-based queries. In ARCH, an initial user
query is used to find matching concepts within a
portion of concept hierarchy. The concept hierarchy is stored in an aggregated form with each
node represented as a term vector. The user can

select or unselect nodes in the presented portion

of the hierarchy, and relevance feedback techniques are used to modify the initial query based
on these nodes.
Similarly, domain-specific search and retrieval
applications allow for a more focused and accurate search based on specific relations inherent in
the underlying domain knowledge. The CiteSeer
system (Bollacker, Lawrence & Giles, 1998) is a
Web agent for finding interesting research publications, in which the relation cited by is the
primary relation discovered among objects (i.e.,
among published papers). Thus, CiteSeer allows
for comparison and retrieval of documents, not
only based on similar content, but also based
on inter-citation linkage structure among documents.
CiteSeer is an example of an approach for
integrating semantic knowledge based on Web
structure mining. In general, Web structure mining tasks take as input the hyperlink structure of
Web pages (either belonging to a Website or relative to the whole Web), and output the underlying
patterns (e.g., page authority values, linkage similarity, Web communities, etc.) that can be inferred
from the hypertext co-citations. Another example
of such an approach is the PageRank algorithm,
which is the backbone of the Google search engine.
PageRank uses the in-degree of indexed pages
(i.e., number of pages referring to it) in order to
rank pages based on quality or authoritativeness.
Such algorithms that are based on the analysis of
structural attributes can be further enhanced by
integrating content semantics (Chakrabarti et al.,
1998). Web semantics can also enhance crawling
algorithms by combining content or ontology with
linkage information (Chakrabarti, van den Berg
& Dom, 1999; Maedche, Ehrig, Handschuh, Volz
& Stojanovic, 2002).
The use of domain knowledge can also provide
tremendous advantages in Web usage mining and
personalization. For example, semantic knowledge
may help in interpreting, analyzing, and reasoning
about usage patterns discovered in the mining

737

Integrating Semantic Knowledge with Web Usage Mining for Personalization

phase. Furthermore, it can enhance collaborative

filtering and personalization systems by providing concept-level recommendations (in contrast
to item-based or user-based recommendations).
Another advantage is that user demographic data,
represented as part of a domain ontology, can be
more systematically integrated into collaborative
or usage-based recommendation engines. Several
studies have considered various approaches to
integrate content-based semantic knowledge into
traditional collaborative filtering and personalization frameworks (Anderson et al., 2002; Claypool
et al., 1999; Melville, Mooney & Nagarajan,
2002; Mobasher, Dai, Luo, Sun & Zhu, 2000b;
Pazzani, 1999). Recently, we proposed a formal
framework for integrating full domain ontologies
with the personalization process based on Web
usage mining (Dai & Mobasher, 2002).

Web Usage Mining and

Personalization
The goal of personalization based on Web usage
mining is to recommend a set of objects to the

current (active) user, possibly consisting of links,

ads, text, products, and so forth, tailored to the
users perceived preferences as determined by
the matching usage patterns. This task is accomplished by matching the active user session
(possibly in conjunction with previously stored
profiles for that user) with the usage patterns
discovered through Web usage mining. We call
the usage patterns used in this context aggregate
usage profiles since they provide an aggregate representation of the common activities or interests
of groups of users. This process is performed by
the recommendation engine which is the online
component of the personalization system. If the
data collection procedures in the system include
the capability to track users across visits, then
the recommendations can represent a longer term
view of users potential interests based on the
users activity history within the site. If, on the
other hand, aggregate profiles are derived only
from user sessions (single visits) contained in log
files, then the recommendations provide a shortterm view of users navigational interests. These
recommended objects are added to the last page

Figure 1. General Framework for Web Personalization Based on Web Usage Mining The Offline
Pattern Discovery Component

738

Integrating Semantic Knowledge with Web Usage Mining for Personalization

Figure 2. General Framework for Web Personalization Based on Web Usage Mining The Online
Personalization Component

in the active session accessed by the user before

that page is sent to the browser.
The overall process of Web personalization
based on Web usage mining consists of three
phases: data preparation and transformation, pattern discovery, and recommendation. Of these,
only the latter phase is performed in real-time.
The data preparation phase transforms raw Web
log files into transaction data that can be processed
by data mining tasks. A variety of data mining
techniques can be applied to this transaction data
in the pattern discovery phase, such as clustering,
association rule mining, and sequential pattern
discovery. The results of the mining phase are
transformed into aggregate usage profiles, suitable for use in the recommendation phase. The
recommendation engine considers the active user
session in conjunction with the discovered patterns
to provide personalized content.
The primary data sources used in Web usage
mining are the server log files, which include Web

server access logs and application server logs.

Additional data sources that are also essential
for both data preparation and pattern discovery
include the site files (HTML, XML, etc.) and
meta-data, operational databases, and domain
knowledge. Generally speaking, the data obtained
through these sources can be categorized into four
groups (see also Cooley, Mobasher, & Srivastava,
1999; Srivastava et al., 2000).

Usage data: The log data collected automatically by the Web and application servers represents the fine-grained navigational
behavior of visitors. Depending on the goals
of the analysis, this data needs to be transformed and aggregated at different levels
of abstraction. In Web usage mining, the
most basic level of data abstraction is that
of a pageview. Physically, a pageview is an
aggregate representation of a collection of
Web objects contributing to the display on

739

Integrating Semantic Knowledge with Web Usage Mining for Personalization

740

a users browser resulting from a single user

action (such as a clickthrough). These Web
objects may include multiple pages (such as
in a frame-based site), images, embedded
components, or script and database queries
that populate portions of the displayed page
(in dynamically generated sites). Conceptually, each pageview represents a specific
type of user activity on the site, e.g.,
reading a news article, browsing the results
of a search query, viewing a product page,
adding a product to the shopping cart, and
so on. On the other hand, at the user level,
the most basic level of behavioral abstraction is that of a server session (or simply a
session). A session (also commonly referred
to as a visit) is a sequence of pageviews
by a single user during a single visit. The
notion of a session can be further abstracted
by selecting a subset of pageviews in the
session that are significant or relevant for
the analysis tasks at hand. We shall refer to
such a semantically meaningful subset of
pageviews as a transaction. It is important to
note that a transaction does not refer simply
to product purchases, but it can include a
variety of types of user actions as captured
by different pageviews in a session.
Content data: The content data in a site is
the collection of objects and relationships
that are conveyed to the user. For the most
part, this data is comprised of combinations
of textual material and images. The data
sources used to deliver or generate this data
include static HTML/XML pages, image,
video, and sound files, dynamically generated page segments from scripts or other
applications, and collections of records from
the operational database(s). The site content
data also includes semantic or structural
meta-data embedded within the site or individual pages, such as descriptive keywords,
document attributes, semantic tags, or HTTP
variables. Finally, the underlying domain

ontology for the site is also considered part

of content data. The domain ontology may
be captured implicitly within the site, or it
may exist in some explicit form. The explicit representation of domain ontologies
may include conceptual hierarchies over
page contents, such as product categories,
structural hierarchies represented by the
underlying file and directory structure in
which the site content is stored, explicit
representation of semantic content and relationships via an ontology language such
as RDF, or a database schema over the data
contained in the operational databases.
Structure data: The structure data represents the designers view of the content
organization within the site. This organization is captured via the inter-page linkage
structure among pages, as reflected through
hyperlinks. The structure data also includes
the intra-page structure of the content
represented in the arrangement of HTML
or XML tags within a page. For example,
both HTML and XML documents can be
represented as tree structures over the space
of tags in the page. The structure data for a
site is normally captured by an automatically
generated site map which represents the
hyperlink structure of the site. A site mapping tool must have the capability to capture
and represent the inter- and intra-pageview
relationships. This necessity becomes most
evident in a frame-based site where portions of distinct pageviews may represent
the same physical page. For dynamically
generated pages, the site mapping tools
must either incorporate intrinsic knowledge
of the underlying applications and scripts,
or must have the ability to generate content
segments using a sampling of parameters
passed to such applications or scripts.
User data: The operational database(s) for
the site may include additional user profile
information. Such data may include demo-

Integrating Semantic Knowledge with Web Usage Mining for Personalization

graphic or other identifying information

on registered users, user ratings on various
objects such as pages, products, or movies,
past purchase or visit histories of users, as
well as other explicit or implicit representation of users interests. Obviously, capturing
such data would require explicit interactions
with the users of the site. Some of this data
can be captured anonymously, without any
identifying user information, so long as
there is the ability to distinguish among
different users. For example, anonymous
information contained in client-side cookies
can be considered a part of the users profile
information, and can be used to identify
repeat visitors to a site. Many personalization applications require the storage of
prior user profile information. For example,
collaborative filtering applications, generally, store prior ratings of objects by users,
though, such information can be obtained
anonymously, as well.
For a detailed discussion of preprocessing
issues related to Web usage mining see Cooley
et al. (1999). Usage preprocessing results in a set
of n pageviews, P = {p1 ,p2 ,,pn }, and a set of m
user transactions, T = {t1 ,t2 ,,tm }, where each ti
in T is a subset of P. Pageviews are semantically
meaningful entities to which mining tasks are
applied (such as pages or products). Conceptually,
we view each transaction t as an l-length sequence
of ordered pairs:
t = ( p1t , w( p1t ), ( p2t , w( p2t ), , ( plt , w( plt )

where each pt i = pj for some j in {1,,n}, and w(pt i )

is the weight associated with pageview pt i in transaction t, representing its significance (usually, but
not exclusively, based on time duration).
For many data mining tasks, such as clustering
and association rule discovery, as well as collaborative filtering based on the kNN technique, we
can represent each user transaction as a vector

over the n-dimensional space of pageviews. Given

the transaction t above, the transaction vector

t is given by: t = wtp , wtp , , wtp , where each

wtp = w( p tj ), for some j in {1,,n}, if pj appears in
the transaction t, and wtp = 0, otherwise. Thus,
conceptually, the set of all user transactions can
be viewed as an mn transaction-pageview matrix,
denoted by TP.
Given a set of transactions as described above,
a variety of unsupervised knowledge discovery
techniques can be applied to obtain patterns. These
techniques such as clustering of transactions (or
sessions) can lead to the discovery of important
user or visitor segments. Other techniques such as
item (e.g., pageview) clustering and association or
sequential pattern discovery can be used to find
important relationships among items based on
the navigational patterns of users in the site. In
each case, the discovered patterns can be used in
conjunction with the active user session to provide
personalized content. This task is performed by
a recommendation engine.
1

Requirements for Semantic

Web Usage Mining
In this section, we present and discuss the essential
requirements in the integration of domain knowledge with Web usage data for pattern discovery.
Our focus is on the critical tasks that particularly
play an important role when the discovered patterns are to be used for Web personalization. As
a concrete example, in the last part of this section
we discuss an approach for integrating semantic
features extracted from the content of Web sites
with Web usage data, and how this integrated
data can be used in conjunction with clustering
to perform personalization. In the next section,
we go beyond keyword-based semantics and
present a more formal framework for integrating
full ontologies with the Web usage mining and
personalization processes.

741

Integrating Semantic Knowledge with Web Usage Mining for Personalization

Representation of Domain
Knowledge
Representing Domain Knowledge as
Content Features
One direct source of semantic knowledge that can
be integrated into mining and personalization processes is the textual content of Web site pages. The
semantics of a Web site are, in part, represented
by the content features associated with items or
objects on the Web site. These features include
keywords, phrases, category names, or other
textual content embedded as meta-information.
Content preprocessing involves the extraction of
relevant features from text and meta-data.
During the preprocessing, usually different
weights are associated with features. For features
extracted from meta-data, feature weights are
usually provided as part of the domain knowledge specified by the analyst. Such weights may
reflect the relative importance of certain concepts.
For features extracted from text, weights can
normally be derived automatically, for example
as a function of the term frequency and inverse
document frequency (tf.idf) which is commonly
used in information retrieval.
Further preprocessing on content features can
be performed by applying text mining techniques.
This would provide the ability to filter the input
to, or the output from, other mining algorithms.
For example, classification of content features
based on a concept hierarchy can be used to limit
the discovered patterns from Web usage mining
to those containing pageviews about a certain
subject or class of products. Similarly, performing
learning algorithms such as, clustering, formal
concept analysis, or association rule mining on
the feature space can lead to composite features
representing concept categories or hierarchies
(Clerkin, Cunningham, & Hayes, 2001; Stumme
et al., 2000).
The integration of content features with usage-based personalization is desirable when we

742

are dealing with sites where text descriptions

are dominant and other structural relationships
in the data are not easy to obtain, e.g., news
sites or online help systems, etc. This approach,
however, is incapable of capturing more complex
relations among objects at a deeper semantic level
based on the inherent properties associated with
these objects. To be able to recommend different
types of complex objects using their underlying
properties and attributes, the system must be
able to rely on the characterization of user segments and objects, not just based on keywords,
but at a deeper semantic level using the domain
ontologies for the objects. We will discuss some
examples of how integrated content features and
usage data can be used for personalization later
in this Section.

Representing Domain Knowledge as

Structured Data
In Web usage mining, we are interested in the
semantics underlying a Web transaction or a user
profile which is usually composed of a group of
pageview names and query strings (extracted
from Web server logs). Such features, in isolation,
do not convey the semantics associated with the
underlying application. Thus, it is important to
create a mapping between these features and the
objects, concepts, or events they represent.
Many e-commerce sites generate Web pages by
querying operational databases or semi-structured
data (e.g., XML and DTDs), from which semantic
information can be easily derived. For Web sites
in which such structured data cannot be easily acquired, we can adopt machine learning techniques
to extract semantic information. Furthermore, the
domain knowledge acquired should be machine
understandable in order to allow for further processing or reasoning. Therefore, the extracted
knowledge should be represented in some standard
knowledge representation language.
DAML+OIL (Horrocks & Sattler, 2001) is an
example of an ontology language that combines

Integrating Semantic Knowledge with Web Usage Mining for Personalization

the Web standards from XML and RDF, with the

reasoning capabilities from a description logic
SHIP(DL). The combinations of relational models and probabilistic models is another common
approach to enhance Web personalization with
domain knowledge and reasoning mechanism.
Several approaches to personalization have used
Relational Models such as Relational Markov
Model (Anderson et al., 2002). Both of these approaches provide the ability to represent knowledge at different levels of abstraction, and the
ability to reason about concepts, including about
such relations as subsumption and membership.
In Dai & Mobasher (2002), we adopted the
syntax and semantics of another ontology representation framework, SHOQ(D), to represent
domain ontologies. In SHOQ(D), the notion of
concrete datatype is used to specify literal values
and individuals which represent real objects in
the domain ontology. Moreover, concepts can be
viewed as sets of individuals, and roles are binary
relations between a pair of concepts or between
concepts and data types. The detailed formal
definitions for concepts and roles are given in Horrocks & Sattler (2001) and Giugno & Lukasiewicz
(2002). Because our current work does not focus
on reasoning tasks such as deciding subsumption
and membership, we do not focus our discussion
on these operations. The reasoning apparatus in
SHOQ(D) can be used to provide more intelligent
data mining services.

Building Mappings Between

Usage-Level and Domain-Level
Instances
During usage data preprocessing or post processing, we may want to assign domain semantics
to user navigational patterns by mapping the
pageview names or URLs (or queries) to the instances in the knowledge base. To be more specific,
instead of describing a users navigational path
as: a1 , a2 , ..., an (where ai is a URL pointing to
a Web resource), we need to represent it using

the instances from the knowledge base, such as:

movie(name=Matrix), movie(name=Spiderman),
, movie(name=Xman). With the help of a preacquired concept hierarchy, we may, for example,
be able to infer that the current users interest is
in the category of Action&Sci-Fi. We refer to
this semantic form of usage data as Semantic
User Profiles. These profiles, in turn, can be
used for semantic pattern discovery and online
recommendations. In the context of personalization applications, domain-level (semantic)
instances may also need to be mapped back to
Web resources or pages. For example, a recommendation engine using semantic user profiles
may result in recommendations in the form of a
movie genre. This concept must be mapped back
into specific pages, URLs, or sections of the site
relating to this genre before recommendations
can be relayed to the user.

Using Content and Structural

Characteristics
Classification algorithms utilizing content and
structural features from pages are well-suited for
creating mappings from usage data to domainlevel instances. For example, in Craven et al.
(2000) and Ghani & Fano (2002) classifiers are
trained that exploit content or structural features
(such as terms, linkage information, and term
proximity) of the pageviews. From the pageview
names or URLs we can obtain the corresponding Web content such as meta-data or keywords.
With help from text classification algorithms, it
is possible to efficiently map from keywords to
attribute instances (Ghani & Fano, 2002).
Another good heuristics used in creating
semantic mappings is based on the anchor text
associated with hyperlinks. If we can build the
complete user navigational path, we would be
able to acquire the anchor text for each URL or
pageview name. We can include the anchor text
as part of the content features extracted from
the body of documents or in isolation. However,

743

Integrating Semantic Knowledge with Web Usage Mining for Personalization

whereas the text features in a document represent

the semantics of the document, itself, the anchor
text represents the semantics of the document to
which the associated hyperlink points.

Using Query Strings

So far, we have overlooked the enormous amount
of information stored in databases or semi-structured documents associated with a site. Large
information servers often serve content integrated
from multiple underlying servers and databases
(Berendt, & Spiliopoulou, 2000). The dynamically
generated pages on such servers are based on
queries with multiple parameters attached to the
URL corresponding to the underlying scripts or
applications. Using the Web server query string
recorded in the server log files it is possible to
reconstruct the response pages. For example, the
following are query strings from a hypothetical
online bookseller Web site:
http://www.xyz.com/app.cgi?action=viewitem&
item=1234567&category=1234
http://www.xyz.com/app.cgi?action=search&se
archtype=title&searchstring= web+mining
http://www.xyz.com/app.cgi?action=order&ite
m=1234567&category=1234& couponid=3456
If the background database or semi-structured
documents are available, then we can access the
content of the instances in the response pages
via the name-value pairs from the query strings.
This enriches our knowledge base of user interest. In the above bookseller Web site example, if
we were able to access background database, we
would be able to get the content of item 1234567
in category 1234. In this case, we could have
the book name, price, author information of this
item. We could recommend other books in the
same content category or written by the same
author. More generally, in well-designed sites,

744

there is usually an explicitly available semantic

mapping between query parameters and objects
(such as products and categories), which would
obviate the need to reconstruct the content of
dynamic pages.

Levels of Abstraction
Capturing semantic knowledge at different levels
of abstraction provides more flexibility both in the
mining phase and in the recommendation phase.
For example, focusing on higher-level concepts in
a concept hierarchy would allow certain patterns
to emerge which otherwise may be missed due
to low support. On the other hand, the ability to
drill-down into the discovered patterns based
on finer-grained subconcepts would provide the
ability to give more focused and useful recommendations.
Domain knowledge with attributes and relations requires the management of a great deal more
data than is necessary in traditional approaches
to Web usage mining. Thus, it becomes essential
to prune unnecessary attributes or relations. For
example, it may be possible to examine the number
of distinct values of each attribute and generalize the attributes if there is a concept hierarchy
over the attribute values. In Han & Fu (1995) a
multiple-level association rule mining algorithm
is proposed that utilizes concept hierarchies. For
example, the usage data in our hypothetical movie
site may not provide enough support for an association rule: Spiderman, Xmen Xmen2, but
mining at a higher level may result in obtaining
a rule: Sci-Fi&Action, Xmen Xmen2. In
Anderson et al. (2002) relational Markov models
are built by performing shrinkage (McCallum et
al., 1998) between the estimates of parameters
at all levels of abstractions relative to a concept
hierarchy. If a pre-specified concept hierarchy does
not exist, it is possible to automatically create such
hierarchies through a variety of machine learning
techniques, such as hierarchical agglomerative
clustering (Stumme et al., 2000).

Integrating Semantic Knowledge with Web Usage Mining for Personalization

Integration of Semantics at Different

Stages of Knowledge Discovery
The semantic information stored in the knowledge base can be leveraged at various steps in
the knowledge discovery process, namely in the
preprocessing phase, in the pattern discovery
phase, or during the post-processing of the discovered patterns.

Preprocessing Phase
The main task of data preprocessing is to prune
noisy and irrelevant data, and to reduce data volume for the pattern discovery phase. In Mobasher,
Dai, Luo, & Nakagawa (2002), it was shown
that applying appropriate data preprocessing
techniques on usage data could improve the effectiveness of Web personalization. The concept
level mappings from the pageview-level data to
concepts can also be performed in this phase.
This results in a transformed transaction data
to which various data mining algorithms can
be applied. Specifically, the transaction vector t
given previously can be transformed into a vector
t ' = wot , wot , , wot , where each oj is a semantic
object appearing in one of the pageviews contained
in the transaction, and is a weight associated with
that object in the transaction. These semantic
objects may be concepts appearing in the concept
hierarchy or finer-grained objects representing
instances of these concepts.
1

Pattern Discovery Phase

Successful utilization of domain knowledge in
this phase requires extending basic data mining
algorithms to deal with relational data to concept
hierarchies. As an example, consider a distancebased data mining technique such as clustering.
The clustering of flat single-relation data (such as
Web user transactions) involves the computation
of similarities or distance among transaction vec-

tors. In such cases, normally simple vector-based

operations are used. However, in the presence
of integrated domain knowledge represented
as concept hierarchies or ontologies, the clustering algorithms will have to perform much
more complex similarity computations across
dimensions and attributes. For example, even if
the two user transactions have no pageviews in
common, they may still be considered similar
provided that the items occurring in both transactions are themselves similar based on some of
their attributes or properties. The integration of
domain knowledge will generate semantic usage patterns, introducing great flexibility as well
as challenges. The flexibility lies in the pattern
discovery being independent of item identities.
The challenge is in the development of scalable
and efficient algorithms to perform the underlying
computational tasks such as similarity computations. We discuss this issue further below.

Post-Processing Phase
Exploiting domain knowledge in this phase can
be used to further explain usage patterns or to
filter out irrelevant patterns. One possibility is to
first perform traditional usage mining tasks on
the item-level usage data obtained in the preprocessing phase, and then use domain knowledge to
interpret or transform the item level user profiles
into domain-level usage profiles (Mobasher &
Dai, 2002) involving concepts and relations in
the ontology. The advantage of this approach is
that we can avoid the scalability issues that can
be endemic in the pattern discovery phase. The
disadvantage is that some important structural
relationships may not be used during the mining
phase resulting in lower quality patterns.

Aggregation Methods for Complex

Objects
To characterize patterns discovered through data
mining techniques, it is usually necessary to de-

745

Integrating Semantic Knowledge with Web Usage Mining for Personalization

rive aggregate representation of the patterns. An

example of this situation is clustering applications.
In the context of Web user transactions, clustering
may result in a group of sessions or visitors that
are considered similar because of their common
navigational patterns. The vector representation
of these transactions facilitates the aggregation
tasks: the centroid (mean vector) of the transaction cluster acts as a representative of all of the
transactions in that cluster. However, in the case of
semantically enhanced transactions, the aggregation may have to be performed independently for
each of the attributes associated with the objects
contained in the cluster.
For example, clustering may result in a group
of users who have all visited pages related to several movies. To be able to characterize this group
of users at a deeper semantic level, it would be
necessary to create an aggregate representation of
the collection of movies in which they are interested. This task would require aggregation along
each dimension corresponding to the attributes
of movie instances, such as genre, actors,
directors, etc. Since each of these attributes
require a different type of aggregation function
depending on the data type and the domain, it
may be necessary to associate various aggregation
functions with the specification of the domain
ontology, itself. In the next section we present
one approach for solving this problem.

The computation of vector similarity between

A and B, Sim(A,B), is dependent on the semantic similarities among the component objects,
SemSim(ai ,bj). For instance, one approach might
be to compute the weighted sum or average of the
similarities among object pairs, such as in:

Measuring Semantic Similarities

Sim( A, B) =

Measuring similarities (alternatively, distances)

among objects is a central task in many data
mining algorithms. In the context of Web usage
mining this may involve computing similarity
measures among pageviews, among user transactions, or among users. This also becomes a critical
task in personalization: a current users profile
must be matched with similar aggregate profiles
representing the discovered user patterns or segments. As in the case of the aggregation problem
discussed above, when dealing with semantically

In general, computing the semantic similarity,

SemSim(a,b), is domain dependent and requires
knowledge of the underlying structure of among
objects. If both objects can be represented using
the same vector model (e.g., pages or documents
represented as bags of words), we can compute
their similarity using standard vector operations.
On the other hand, if their representation includes
attributes and relations specified in the domain
ontology, we need to first make sure that the ob-

746

enhanced transactions, measuring similarities

poses additional challenges. This is because the
similarity of two transactions depends on the
similarities of the semantic objects contained
within the transactions.
Let us again consider the static vector model
for representing a Web transaction t (or a user
profile): t = wt p 1 , wt p 2 , ..., wt p n. Computing similarity between two such vectors is straightforward
and can be performed using measures such as
cosine similarity, Euclidean distance, Pearson
correlation (e.g., in case the weights represent
user ratings).
When such vectors are transformed according
to the underlying semantics, however, the computation of similarities will involve the computation
of semantic similarities among the concepts or
objects, possibly using different domain-specific
similarity measures. Let A and B be two transformed transactions, each represented as a set of
semantic objects in a site:
A = {a1 , a2 , ..., am } and B = {b1 , b2 , ..., bl}.

a A

bB

SemSim(a, b)

AB

Integrating Semantic Knowledge with Web Usage Mining for Personalization

jects can be classified under a common ontological

schema and then measure similarities along the
different dimensions corresponding to each attribute. The notion of semantic matching among
objects and classes has been a subject of considerable study recently (Rodriguez & Egenhofer,
2003; Palopoli, Sacca, Terracina, & Ursino, 2003;
Ganesan, Garcia-Molina, & Widom, 2003).
For example, such an approach was used in Jin
& Mobasher (2003) in the context of collaborative
filtering with movies. In this work, association
analysis was first performed on the genre attribute to define a genre hierarchy. Furthermore,
the year attribute was discretized into intervals, while other attributes, such as cast, were
treated as a bag of words. These preprocessing
steps allowed for the definition of appropriate
similarity measures for each attribute. Finally,
the semantic similarity between two movies, i
and j, was defined as a linear combination of attribute-level similarities:
SemSim(i, j ) =

CastSim(i, j ) +

DirectorSim(i, j ) +

GenreSim(i, j ) + ...

where, i are predefined weights for the corresponding attributes.

Example: Using Content Features for

Semantic Web Usage Mining
As an example of integrating semantic knowledge with the Web usage mining process, let us
consider the especial case of using textual features from the content of Web pages to represent
the underlying semantics for the site. As noted
earlier, each pageview p can be represented as a
k-dimensional feature vector, where k is the total
number of extracted features (words or concepts)
from the site in a global dictionary. This vector
can be given by:
p = fw( p, f1 ), fw( p, f 2 ), , fw( p, f k )

where fw(p, fj) is the weight of the jth feature in

pageview p, for 1 j k. For the whole collection of pageviews in the site, we then have an nk
pageview-feature matrix PF = {p1, p2, ..., pn}.
There are now at least two basic choices as to
when content features can be integrated into the
usage-based personalization process: pre-mining
integration or post-mining integration.
The pre-mining integration involves the transformation of user transactions, as described earlier,
into content-enhanced transactions containing
the semantic features of the pageviews. While, in
practice, there are several ways to accomplish this
transformation, the most direct approach involves
mapping each pageview in a transaction to one
or more content features. The range of this mapping can be the full feature space, or feature sets
(composite features) which in turn may represent
concepts and concept categories. Conceptually, the
transformation can be viewed as the multiplication
of the transaction-pageview matrix TP, defined
earlier, with the pageview-feature matrix PF. The
result is a new matrix TF = {t1, t2, ..., tn }, where
each ti is a k-dimensional vector over the feature
space. Thus, a user transaction can be represented
as a content feature vector, reflecting that users
interests in particular concepts or topics.
Various data mining tasks can now be performed on the content-enhanced transaction data.
For instance, if we apply association rule mining
to such data, then we can get a group of association rules on content features. As an example,
consider a site containing information about
movies. This site may contain pages related to the
movies themselves, actors appearing in the movies, directors, and genres. Association rule mining
process could generate a rule such as: {British,
Romance, Comedy Hugh Grant}, suggesting that users who are interested in British
romantic comedies may also like the actor Hugh
Grant (with a certain degree of confidence). During the online recommendation phase, the users
active session (which is also transformed into a
feature vector) is compared with the discovered

747

Integrating Semantic Knowledge with Web Usage Mining for Personalization

rules. Before recommendations are made, the

matching patterns must be mapped back into Web
pages or Web objects. In the above example, if
the active session matches the left hand side of
the association rule, the recommendation engine
could recommend other Web pages that contain
the feature Hugh Grant.
The post-mining integration of semantic features involves combining the results of mining
(performed independently on usage and content
data) during the online recommendation phase.
An example of this approach was presented in Mobasher et al. (2000b), where clustering algorithms
were applied to both the transaction matrix TP
and the transpose of the feature matrix PF. Since
both matrices have pageviews as dimensions, the
centroids of the resulting clusters in both cases can
be represented as sets of pageview-weight pairs
where the weights signify the frequency of the
pageview occurrence in the corresponding cluster.
We call the patterns generated from content data
content profiles, while the patterns derived from
usage data are called usage profiles. Though
they share the same representation, they have different semantics: usage profiles represent a set of
transactions with similar navigational behavior,
while content profiles contain groups of Web
pages with (partly) similar content.
Specifically, given a transaction cluster (respectively, a feature cluster) cl, we can construct
the usage (respectively, content) profile prc l as a
set of pageview-weight pairs by computing the
centroid of cl:
prcl = { p, weight ( p, prcl ) | weight ( p, prcl ) }

where:
the significance weight, weight(p, prc l ), of
the page p within the usage (respectively,
content) profile prc l is given by:
1

weight ( p, prcl ) = | cl | w( p, s)
scl

748

w(p,s) is the weight of page p in transaction

(respectively, feature) vector s in the cluster
cl; and
the threshold is used to focus only on those
pages in the cluster that appear in a sufficient
number of vectors in that cluster.
Each such profile, in turn, can be represented
as a vector in the original n-dimensional space
of pageviews. This aggregate representation can
be used directly in the recommendation phase:
given a new user, u who has accessed a set of
pages, Pu , so far, we can measure the similarity
of Pu to the discovered profiles, and recommend
to the user those pages in matching profiles which
have not yet been accessed by the user. Note
that this approach does not distinguish between
recommendations emanating from the matching
content and usage profiles. Also note that there
are many other ways of combining usage profiles
and content profiles during the online recommendation phase. For example, we can use content
profiles as the last resort in the situation where
usage profiles can not provide sufficient number
of recommendations.

A Framework for OntologyBased Personalization

At a conceptual level, there may be many different kinds of objects within a given site that are
accessible to users. At the physical level, these
objects may be represented by one or more Web
pages. For example, our hypothetical movie site
may contain pages related to the movies, actors,
directors, and studios. Conceptually, each of these
entities represents a different type of semantic
object. During a visit to this site, a user may
implicitly access several of these objects together
during a session by navigating to various pages
containing them. In contrast to content features,
ontological representation of domain knowledge
contained in the site makes it possible to have a

Integrating Semantic Knowledge with Web Usage Mining for Personalization

uniform architecture to model such objects, their

properties, and their relationships. Furthermore,
such a representation would allow for a more
natural mapping between the relational schema for
the backend databases driving Web applications
and the navigational behavior of users.
In this section we will present a general framework for utilizing domain ontologies in Web usage
mining and personalization. Figure 3 lays out a
general process for such an integrated approach.
In keeping with our earlier discussion, it is composed of three main phases: preprocessing, pattern
discovery and online recommendation. Each of
these phases must take into account the object
properties and their relationships as specified in
a domain ontology.
We assume that the site ontology is already
available (either specified manually, or extracted
automatically using ontology learning techniques).
The goal of the preprocessing phase is to transform
users navigational transactions into semantic
transaction by mapping accessed pages and
resource to concepts and objects of the specified
ontology. The goal of the pattern discovery phase

is to create aggregate representation of groups of

semantic objects that are implicitly accessed by
similar users, thus providing a semantic characterization of user segments with common behavior
or interests. Finally, in the recommendation phase,
the discovered semantic patterns are utilized in
conjunction with an ongoing record of a current
users activity (including, possibly the users stored
profile) to recommend new resources, pages, or
objects to that user.

Knowledge Representation
General ontology representation languages such
as DAML+OIL (Horrocks, 2002) provide formal syntax and semantics for representing and
reasoning with various elements of an ontology.
These elements include individuals (or objects),
concepts (which represent sets of individuals),
and roles (which specify object properties).
In DAML+OIL, the notion of a concept is quite
general and may encompass a heterogeneous set
of objects with different properties (roles) and
structures. We, on the other hand, are mainly

Figure 3. A General Framework for Personalization Based on Domain Ontologies

C u rre n t U s e r

749

Integrating Semantic Knowledge with Web Usage Mining for Personalization

interested in the aggregate representations for

groups of objects that have a homogenous concept
structure (i.e., have similar properties and data
types). For example, we may be interested in a
group of movie objects, each of which has specific
values for properties such as year, genre, and
actors. We call such a group of objects a class.
Thus, in our framework, the notion of a class
represents a restriction of the notion of a concept
in DAML+OIL. It should be noted, however, that
the users of a Web site, in general, access a variety
of objects belonging to different classes. Thus,
this homogeneity assumption would imply that
semantic objects within user transactions must
first be classified into homogenous classes as a
preprocessing step.
More specifically, we define a class C as a
set of objects together with a set of attributes.
These attributes together define the internal
properties of the objects in C or relationships
with other concepts that involve the objects in
C. Thus attributes of a class correspond to a
subset of the set of roles in the domain ontology.
We denote the domain of values of an attribute r
as Dr . Furthermore, because we are specifically
interested in aggregating objects at the attribute
level, we extend the notion of a role to include
a domain-specific combination function and an
ordering relation.
More formally, a class C is characterized by
a finite set of attributes AC, where each attribute
a in AC is defined as follows.
Definition: Let C be a class in the domain ontology. An attribute a AC is a 4-tuple , where
a = Ta , Da , a , a

Ta is the type for the values for the attribute a.

Da is the domain of the values for a;

a is an ordering relation among the
values in Da ; and

a is a combination function for the
attribute a.

750

The type of an attribute in the above definition may be a concrete datatype (such as string
or integer) or it may be a set of objects (individuals) belonging to another class.
In the context of data mining, comparing and
aggregating values are essential tasks. Therefore,
ordering relations among values are necessary
properties for attributes. We associate an ordering
relation a with elements in Da for each attribute
a. The ordering relation a can be null (if no
ordering is specified in the domain of values), or
it can define a partial or a total order among the
domain values. For standard types such as values
from a continuous range, we assume the usual
ordering. In cases when an attribute a represents
a concept hierarchy, the domain values of a are a
set of labels, and a is a partial order representing
the is-a relation.
Furthermore, we associate a data mining operator, called the combination function, a , with each
attribute a. This combination function defines an
aggregation operation among the corresponding
attribute values of a set of objects belonging to
the same class. This function is essentially a generalization of the mean or average function
applied to corresponding dimension values of a
set of vectors when computing the centroid vector.
In this context, we assume that the combination
function is specified as part of the domain ontology for each attribute of a class. An interesting
extension would be to automatically learn the
combination function for each attribute based on
a set of positive and negative examples.
Classes in the ontology define the structural
and semantic properties of objects in the domain
which are instances of that class. Specifically,
each object o in the domain is also characterized
by a set of attributes Ao corresponding to the
attributes of a class in the ontology. In order to
more precisely define the notion of an object as
an instance of a class, we first define the notion
of an instance of an attribute.

Integrating Semantic Knowledge with Web Usage Mining for Personalization

Definition: Given an attribute a = Ta , Da , a ,

a and an attribute b = Tb , Db , b , b, b
is an instance of a, if Db Da , Tb = Ta , b
= a , and b is a restriction of a to Db .
The attribute b is a null instance of a, if Db
is empty.
Definition: Given a class C with attribute set
AC = {a1 C , a2 C , ..., an C }, we say that an object
o is an instance of C, if o has attributes Ao =
{a1 o , a2 o , ..., an o} such that each is a, possibly
null, instance of a1 C .
Based on the definitions of attribute and
object instances, we can now provide a more
formal representation of the combination function a . Let C be a class and {o1 , o2 , ..., om } a set
of object instances of C. Let aAC be an attribute
of class C. The combination function a can be
represented by:
a

({a , w
o1

, ao2 , w2 , , aom , wm

})=

aagg , wagg

where each belonging to object oi is an instance

of the attribute a, and each wi is a weight associated with that attribute instance (representing
the significance of that attribute relative to the
other instances). Furthermore, aa g g is a pseudo
instance of a meaning that it is an instance of
a which does not belong to a real object in the
underlying domain. The weight wa g g of aa g g is a
function of w1 , w2 , ..., wn .
Given a set of object instances, of a class C,
a domain-level aggregate profile for these instances is obtained by applying the combination
function for each attribute in C to all of the corresponding attribute instances across all objects
o1 , o2 , ..., on .

Ontology Preprocessing
The ontology preprocessing phase takes as input
domain information (such as database schema
and metadata, if any) as well as Web pages, and

generates the site ontology. For simple Web sites,

ontologies can be easily designed manually or
derived semi-automatically from the site content.
However, it is more desirable to have automatic
ontology acquisition methods for a large Web
site, especially in Web sites with dynamically
generated Web pages. E-commerce Web sites,
for instance, usually have well-structured Web
content, including predefined metadata or database schema. Therefore it is easier to build
automatic ontology extraction mechanisms that
are site-specific.
There have been a number of efforts dealing
with the ontology learning problem (Clerkin et
al., 2001; Craven et al., 2000; Maedche & Staab,
2000). A wide range of information, such as
thesauri, content features, and database schema
can help to identify ontologies. Many of these approaches have focused on extracting ontological
information from the Web, in general. In Berendt
et al. (2002) the notion of Semantic Web Mining
was introduced, including a framework for the
extraction of a concept hierarchy and the application of data mining techniques to find frequently
occurring combinations of concepts.

An Example
As an example, let us revisit our hypothetical
movie Web site. The Web site includes collections of pages containing information about
movies, actors, directors, etc. A collection of
pages describing a specific movie might include
information such as the movie title, genre, starring actors, director, etc. An actor or directors
information may include name, filmography (a set
of movies), gender, nationality, etc. The portion
of domain ontology for this site, as described,
contains the classes Movie, Actor and Director
(Figure 4). The collection of Web pages in the
site represents a group of embedded objects that
are the instances of these classes.
In our example, the class Movie has attributes
such as Year, Actor (representing the relation

751

Integrating Semantic Knowledge with Web Usage Mining for Personalization

Figure 4. Ontology for a Movie Web Site

Movie
Name

Year

Genre

Name
Romance

Director

Actor

Genre-All
Action

Actor

Movie

Nationality

Comedy
Director

Romantic
Comedy

Black
Comedy

Kid &
Family

acted by), Genre, and Director. The Actor and

Director attributes have values that are other objects in the ontology, specifically, object instances
of classes Actor and Director, respectively. The
attribute Year is an example of an attribute whose
datatype is positive integers with the usual ordering. The attribute Genre has a concrete datatype
whose domain values in DG e n r e are a set of labels
(e.g., Romance and Comedy). The ordering
relation G e n r e defines a partial order based on
the is-a relation among subsets of these labels
(resulting in a concept hierarchy of Genres, a
portion of which is shown in Figure 4).
Figure 5 shows a Movie instance About
a Boy and its related attributes and relations
extracted from a Web page. The schema of the
class Movie is shown at the bottom left portion
of the figure. Here we treat the classes Genre
and Year as attributes of the class Movie. The
instances of the ontology are shown at the bottom
right of the figure. The Genre attribute contains a
partial order among labels representing a concept
hierarchy of movie genres. We use a restriction of
this partial order to represent the genre to which
the Movie instance belongs. The diagram also
shows a keyword bag containing the important
keywords in that page.
An attribute a of an object o has a domain Da .
In cases when the attribute has unique a value for

752

Name

Movie

Nationality

an object, Da is a singleton. For example, consider

an object instance of class Movie, About a Boy
(see Figure 5). The attribute Actor contains two
objects H. Grant and T. Collette that are
instances of the class Actor (for the sake of presentation we use the actors names to stand for
the object instances of Actor). Therefore, DA c t o r
= {H. Grant, T. Collette}. Also, a real object
may have values for only some of the attributes.
In this case the other attributes have empty domains. For instance, the attribute Director in the
example has an empty domain and is thus not
depicted in the figure. We may, optionally, associate a weight with each value in the attribute
domain Da (usually in the range [0,1]). This may
be useful in capturing the relative importance of
each attribute value.
For example, in a given movie the main actors
should have higher weights than other actors in
the cast. In our example, the object H. Grant
has weight 0.6 and the object Toni Collette has
weight 0.4. Unless otherwise specified, we assume
that the weight associated with each attribute
value is 1. In the object o shown in Figure 5, the
domain for the attribute Genre is the set of labels
{Genre-All, Action, Romance, Comedy, Romantic
Comedy, Black Comedy, Kids & Family}. The
ordering relation o G e n r e is a restriction of G e n r e

Integrating Semantic Knowledge with Web Usage Mining for Personalization

Figure 5. Example of Ontology Preprocessing

K eyw ord bag : { B oy, H ug h, G rant,
W e itz, w itty, crow dp le asing , r om a ntic,
com edy, m ovie , s tarrin g, e m otiona lly,
dating, B ritish, f rien dship, U niversal,
P G -13}

A bo ut a boy
F rom h ttp://w w w .reel.com /m ovie.asp?M ID = 134706

M ovie

G enre

G enreA ll

S tarring

G enre

A ctor

Y ear

S tarrin g Y

{H .G ra n t 0 .6 ;
T o n i C olle tte : 0 .4 }

ear

2002

C om e dy
R o m a n tic
Com edy

to the subset {Genre-All, Comedy, Romantic

Comedy, Kids & Family}.

Pattern Discovery
As depicted in Figure 3 domain ontologies can be
incorporated into usage preprocessing to generate
semantic user transactions, or they can be integrated into pattern discovery phase to generate
semantic usage patterns. In the following example,
we will focus on the latter approach.
Given a discovered usage profile (for example,
a set of pageview-weight pairs obtained by clustering user transactions), we can transform it into
a domain-level aggregate representation of the
underlying objects (Dai & Mobasher, 2002). To
distinguish between the representations we call
the original discovered pattern an item-level
usage profile, and we call the new profile based on
the domain ontology a domain-level aggregate

K id &
F a m ily

S tep 1: O ntology
P reprocessin g

profile. The item-level profile is first represented

as a weighted set of objects: pr = {o1 , w1, o2 ,
w2, ..., on , wn in which each oi is an object in the
underlying domain ontology and wi represents oi s
significance in the profile pr. Here we assume that,
either using manual rules, or through supervised
learning methods, we can extract various object
instances represented by the pages in the original
page- or item-level usage profile. The transformed
profile represents a set of objects accessed together
frequently by a group of users (as determined
through Web usage mining). Objects, in the usage
profile, that belong to the same class are combined
to form an aggregated pseudo object belonging to
that class. An important benefit of aggregation is
that the pattern volume is significantly reduced,
thus relieving the computation burden for the
recommendation engine. Our goal is to create an
aggregate representation of this weighted set of
objects to characterize the common interests of

753

Integrating Semantic Knowledge with Web Usage Mining for Personalization

the user segment captured by the usage profile at

the domain level.
Given the representation of a profile pr as a
weighted set of objects, the objects in pr may be
instances of different classes C1 , C2 , ..., Ck in the
ontology. The process of creating a domain-level
aggregate profile begins by partitioning pr into
collections of objects with each collection containing all objects that are instances of a specified
class (in other words, the process of classifying the
object instances in pr). Let Gi denote the elements
of pr that are instances of the class Ci .
Having partitioned pr into k groups of homogeneous objects, G1 , ..., Gk , the problem is
reduced to creating aggregate representation of
each partition Gi . This task is accomplished with
the help of the combination functions for each of
the attributes of Ci some of whose object instances
are contained in Gi . Once the representatives for
every partition of objects are created, we assign
a significance weight to each representative to
mark the importance of this group of objects in
the profile. In our current approach the significance weight for each representative is computed
as the weighted sum of all the object weights in
the partition. However, significance weight can
be computed using other numeric aggregation
functions.

Examples Continued: Generating

Domain-Level Aggregate Profiles
To illustrate the semantic aggregation process, let
us return to our movie site example. The aggregation process requires that a combination function
be defined for each attribute of an object in the
domain ontology. Figure 6 and 8 show an example
of such process. Each movie object has attribute
Name, Actor, Genre and Year. For the
attribute Name, we are interested in all the movie
names appearing in the instances. Thus we can
define N a m e to be the union operation performed
on all the singleton Name attributes of all movie
objects. On the other hand, the attribute Actor con-

754

tains a weighted set of objects belonging to class

Actor. In fact, it represents the relation Starring
between the actor objects and the movie object.
In such cases we can use a vector-based weighted
mean operation as the combination function. For
example, we will determine the aggregate weight
of an actor object o by:
wo' =

( w w )/ w .
i

Applying A c t o r in our example will result in

the aggregate actor object {S, 0.58, T, 0.27, U,
0.09}. As for the attribute Year, the combination
function may create a range of all the Year values
appearing in the objects. Another possible solution
is to discretize the full Year range into decades
and find the most common decades that are in the
domains of the attribute. In our example, using
the range option, this may result in an aggregate
instance [1999, 2002] for the Year attribute.
The attribute Genre of Movie contains a
partial order representing a concept hierarchy
among different Genre values. The combination function, in this case, can perform tree (or
graph) matching to extract the common parts of
the conceptual hierarchies among all instances.
Extracting the common nodes from this hierarchy
may also depend on the weights associated with
the original objects leading to different weights
on the graph edges. For example, given that the
higher weight Movies 1 and 2 have Romance
in common, this node may be selected for the aggregate instance, even though it is not present in
Movie 3. However, the weight of Romance may
be less than that of Comedy which is present
in all three movies.
Figure 6 shows the item-level usage profile
and its representation as a weighted set of objects,
as well as the resulting domain-level aggregate
profile. Note that the original item-level profile
gives us little information about the reasons why
these objects were commonly accessed together.
However, after we characterize this profile at the
domain-level, we find some interesting patterns:

Integrating Semantic Knowledge with Web Usage Mining for Personalization

Figure 6. Creating an Aggregate Representation of a Set of Movie Objects

they all belong to Genre Comedy (and to a

lesser degree Romance), and the actor S has a
high score compared with other actors.

Online Recommendation Phase

In contrast to transaction-based usage profiles,
semantic usage profiles capture the underlying
common properties and relations among those
objects. This fine-grained domain knowledge,
captured in aggregate form enables more powerful approaches to personalization. As before,
we consider the browsing history of the current
user, i.e., active session, to be a weighted set of
Web pages that the user has visited. The same
transformation described in the last subsection can
be used to create a semantic representation of the
users active session. We call this representation
the current user profile.
Figure 7 presents the basic procedure for
generating recommendations based on semantic
profiles. The recommendation engine matches the
current user profile against the discovered domainlevel aggregate profiles. The usage profiles with

matching score greater than some pre-specified

threshold are considered to represent this users
potential interests. A successful match implies
that the current user shares common interests
with the group of users represented by the profile.
The matching process results in an extended user
profile which is obtained by applying the aggregation process described above to the domain-level
profiles and the original user profile.
The recommendation engine then instantiates
the users extended profile to real Web objects and
will recommend them to the user. We can also
exploit structural relationships among classes
during the recommendation process. For example,
if a concept hierarchy exists among objects, and
the recommendation engine can not find a good
match for a user profile at a certain concept level,
then it can generalize to a more abstract level (e.g.,
from romantic comedy to romance).
This approach has several advantages over
traditional usage-based personalization. First,
it retains the user-to-user relationships that can
be captured by the discovered usage profiles.
Secondly, in contrast to standard collaborative

755

Integrating Semantic Knowledge with Web Usage Mining for Personalization

Figure 7. Online Recommendation Enhanced by

Domain Ontologies
Current User
Profile

Aggregate Semantic
Usage Patterns

Match Profiles
Extended User Profile

Instantiate to Real
Web Objects
Recommendations
of Items

filtering, it provides more flexibility in matching

aggregate usage profiles with the current users
activity because the matching process involves
comparison of features and relationships, not exact
item identities. Thirdly, the items do not have to
appear in any usage profiles in order to be recommended, since fine-grained domain relationships
are considered during the instantiation process.
The previous example shows that this approach
can also be used to solve the new item problem.
Furthermore, it can alleviate the notorious sparsity problem in collaborative filtering systems by
allowing for fuzzy comparisons between two
user profiles (or ratings). The basis for matching
profiles does not have to be similar ratings on
the same items. The comparison can be based on
showing interest in different objects with similar
properties (for example, purchasing items that have
same brand). Therefore, even if the raw transaction
or rating data is sparse, the semantic attributes
of items or users can be used to indirectly infer
potential interest in other items.

756

Conclusion
We have explored various approaches, requirements, and issues for integrating semantic
knowledge into the personalization process based
on Web usage mining. We have considered approaches based on the extraction of semantic
features from the textual content contained in a
site and their integration with Web usage mining
tasks and personalization both in the pre-mining
and the post-mining phases of the process. We
have also presented a framework for Web personalization based on full integration of domain
ontologies and usage patterns. The examples
provided throughout this chapter reveal how such
a framework can provide insightful patterns and
smarter personalization services.
We leave some interesting research problems
for open discussion and future work. Most important among these are techniques for computing
similarity between domain objects and aggregate
domain-level patterns, as well as learning techniques to automatically determine appropriate
combination functions used in the aggregation
process.
More generally, the challenges lie in the successful integration of ontological knowledge at
every stage of the knowledge discovery process.
In the preprocessing phase, the challenges are in
automatic methods for the extraction and learning of the ontologies and in the mapping of users
activities at the clickstream level to more abstracts
concepts and classes. For the data mining phase,
the primary goal is to develop new approaches that
take into account complex semantic relationships
such as those present in relational databases with
multiple relations. Indeed, in recent years, there
has been more focus on techniques such as those
based relational data mining. Finally in the personalization stage, the challenge is in developing
techniques that can successfully and efficiently
measure semantic similarities among complex
objects (possibly from different ontologies).

Integrating Semantic Knowledge with Web Usage Mining for Personalization

In this chapter we have only provided an

overview of the relevant issues and suggested a
road map for further research and development
in this area. We believe that the successful integration of semantic knowledge with Web usage
mining is likely to lead to the next generation of
personalization tools which are more intelligent
and more useful for Web users.

topic-specific Web resource discovery. Proceedings of the 3rd World Wide Web Conference,
Toronto.

References

Clerkin, P., Cunningham, P., & Hayes, C. (2001).

Ontology discovery for the semantic Web using
hierarchical clustering. Semantic Web Mining
Workshop at ECML/PKDD-2001, Freiburg,
Germany.

Anderson, C., Domingos, P., & Weld, D. (2002).

Relational markov models and their application
to adaptive Web navigation. Proceedings of the
Eighth ACM SIGKDD International Conference
on Knowledge Discovery and Data Mining (KDD2002), Edmonton, Alberta, Canada.
Berendt, B., & Spiliopoulou, M. (2000). Analysing navigation behaviour in Websites integrating
multiple information systems. VLDB Journal,
Special Issue on Databases and the Web, 9(1),
56-75.
Berendt, B., Hotho, A., & Stumme, G. (2002).
Towards semantic Web mining. Proceedings of
the First International Semantic Web Conference
(ISWC02), Sardinia, Italy.
Bollacker, K., Lawrence, S., & Giles, C.L. (1998).
Citeseer: An autonomous Web agent for automatic
retrieval and identification of interesting publications. Proceedings of the Second International
Conference on Autonomous Agents, Minneapolis,
Minnesota.
Chakrabarti, S., Dom, B., Gibson, D., Kleinberg,
J., Raghavan, P., & Rajagopalan, S. (1998). Automatic resource list compilation by analyzing
hyperlink structure and associated text. Proceedings of the 7th International World Wide Web
Conference, Brisbane, Australia.
Chakrabarti, S., van den Berg, M., & Dom, B.
(1999). Focused crawling: A new approach to

Claypool, M., Gokhale, A., Miranda, T., Murnikov,

P., Netes, D., & Sartin, M. (1999). Combining content-based and collaborative filters in an online
newspaper. Proceedings of the ACM SIGIR 99
Workshop on Recommender Systems: Algorithms
and Evaluation, Berkeley, California.

Cooley, R., Mobasher, B., & Srivastava, J. (1997).

Web mining: Information and pattern discovery
on the World Wide Web. Proceedings of the International Conference on Tools with Artificial
Intelligence, 558-567, Newport Beach. IEEE
Press.
Cooley, R., Mobasher, B., & Srivastava, J. (1999).
Data preparation for mining World Wide Web
browsing patterns. Journal of Knowledge and
Information Systems, 1(1), 5-32.
Craven, M., DiPasquo, D., Freitag, D., McCallum, A., Mitchell, T., Nigam, K., & Slattery, S.
(2000). Learning to construct knowledge bases
from the World Wide Web. Artificial Intelligence,
118(1-2), 69-113.
Dai, H., & Mobasher, B. (2002). Using ontologies
to discover domain-level Web usage profiles.
Proceedings of the 2nd Semantic Web Mining
Workshop at ECML/PKDD 2002, Helsinki,
Finland.
Ganesan, P., Garcia-Molina, H., & Widom, J.
(2003). Exploiting hierarchical domain structure
to compute similarity. ACM Transactions on
Information Systems, 21(1), 63-94.
Getoor, L., Friedman, N., Koller, D., & Taskar, B.
(2001). Learning probabilistic models of relational

757

Integrating Semantic Knowledge with Web Usage Mining for Personalization

structure. Proceedings of the 18th International

Conference on Machine Learning.

tional Conference on Information and Knowledge

Sharing, Scottsdale, Arizona.

Ghani, R., & Fano, A. (2002). Building recommender systems using a knowledge base of
product semantics. Proceedings of the Workshop
on Recommendation and Personalization in
E-Commerce, 2nd International Conference on
Adaptive Hypermedia and Adaptive Web Based
Systems, Malaga, Spain.

Joachims, T., Freitag, D., & Mitchell, T. (1997).

Webwatcher: A tour guide for the World Wide
Web. Proceedings of the 15th International
Conference on Artificial Intelligence, Nagoya,
Japan.

Giugno, R., & Lukasiewicz, T. (2002). PSHOQ(D): A probabilistic extension of SHOQ(D)

for probabilistic ontologies in the semantic Web.
Proceedings of the 8th European Conference on
Logics in Artificial Intelligence, Cosenza, Italy.
Han, J., & Fu, Y. (1995). Discovery of multiplelevel association rules from large databases.
Proceedings of the 1995 International Conference
on Very Large Data Bases (VLDB95), Zurich,
Switzerland.
Herlocker, J., Konstan, J., Borchers, A., & Riedl, J.
(1999). An algorithmic framework for performing
collaborative filtering. Proceedings of the 22nd
ACM Conference on Research and Development
in Information Retrieval (SIGIR99), Berkeley,
CA.
Horrocks, I. (2002). DAML+OIL: A description
logic for the semantic Web. IEEE Data Engineering Bulletin, 25(1), 4-9.
Horrocks, I., & Sattler, U. (2001). Ontology reasoning in the SHOQ(D) description logic. Proceedings of the 17th International Joint Conference
on Artificial Intelligence, Seattle, WA.
Hotho, A., Maedche, A., & Staab, S. (2001). Ontology-based text clustering. Proceedings of the
IJCAI-2001 Workshop Text Learning: Beyond
Supervision, Seattle, WA.
Jin, X., & Mobasher, B. (2003). Using semantic
similarity to enhance item-based collaborative
filtering. Proceedings of The 2nd IASTED Interna-

758

Lieberman, H. (1995). Letizia: An agent that

assists Web browsing. Proceedings of the 1995
International Joint Conference on Artificial Intelligence, Montreal, Canada.
Loh, S., Wives, L.K., & de Oliveira, J.P. (2000).
Concept-based knowledge discovery in texts
extracted from the Web. SIGKDD Explorations,
2(1), 29-39.
Maedche, A., & Staab, S. (2000). Discovering
conceptual relations from text. Proceedings of the
European Conference on Artificial Intelligence
(ECAI00), Berlin.
Maedche, A., Ehrig, M., Handschuh, S., Volz,
R., & Stojanovic, L. (2002). Ontology-focused
crawling of documents and relational metadata.
Proceedings of the 11th International World Wide
Web Conference (WWW02), Honolulu, Hawaii.
McCallum, A., Rosenfeld, R., Mitchell, T., &
Ng, A. (1998). Improving text classification by
shrinkage in a hierarchy of classes. Proceedings
of the 15th International Conference on Machine
Learning, Madison, Wisconsin.
Melville, P., Mooney, R.J., & Nagarajan, R.
(2002). Content-boosted collaborative filtering for
improved recommendations. Proceedings of the
18th National Conference on Artificial Intelligence
(AAAI02), Edmonton, Alberta, Canada.
Mobasher, B., Cooley, R., & Srivastava, J. (2000a).
Automatic personalization based on Web usage
mining. Communications of the ACM, 43(8),
142-151.

Integrating Semantic Knowledge with Web Usage Mining for Personalization

Mobasher, B., Dai, H., Luo, T., & Nakagawa, M.

(2001). Effective personalization based on association rule discovery from Web usage data.
Proceedings of the 3rd ACM Workshop on Web
Information and Data Management (WIDM01),
Atlanta, Georgia.
Mobasher, B., Dai, H., Luo, T., & Nakagawa, M.
(2002). Discovery and evaluation of aggregate usage profiles for Web personalization. Data Mining
and Knowledge Discovery, 6, 61-82.
Mobasher, B., Dai, H., Luo, T., Sun, Y., & Zhu,
J. (2000b). Integrating Web usage and content
mining for more effective personalization. ECommerce and Web Technologies: Proceedings
of the EC-WEB 2000 Conference, Lecture Notes
in Computer Science (LNCS) 1875, 165-176.
Springer.
Palopoli, L., Sacca, D., Terracina, G., & Ursino, D.
(2003). Uniform techniques for deriving similarities of objects and subschemes in heterogeneous
databases. IEEE Transactions on Knowledge and
Data Engineering, 15(1), 271-294.
Palpanas, T., & Mendelzon, A. (1999). Web
prefetching using partial match prediction. Proceedings of the 4th International Web Caching
Workshop (WCW99), San Diego, CA.
Parent, S., Mobasher, B., & Lytinen, S. (2001). An
adaptive agent for Web exploration based on concept hierarchies. Proceedings of the International
Conference on Human Computer Interaction,
New Orleans, LA.
Pazzani, M. (1999). A framework for collaborative,
content-based and demographic filtering. Artificial
Intelligence Review, 13(5-6), 393-408.

Perkowitz, M., & Etzioni, O. (1998). Adaptive

Websites: Automatically synthesizing Web pages.
Proceedings of the 15th National Conference on
Artificial Intelligence, Madison, WI.
Pitkow, J., & Pirolli, P. (1999). Mining longest
repeating subsequences to predict WWW surfing. Proceedings of the 2nd USENIX Symposium
on Internet Technologies and Systems, Boulder,
Colorado.
Rodriguez, M.A., & Egenhofer, M.J. (2003).
Determining semantic similarity among entity
classes from different ontologies. IEEE Transactions on Knowledge and Data Engineering,
15(2), 442-456.
Sarwar, B.M., Karypis, G., Konstan, J., & Riedl,
J. (2000). Analysis of recommender algorithms
for e-commerce. Proceedings of the 2nd ACM
E-Commerce Conference (EC00), Minneapolis,
MN.
Srivastava, J., Cooley, R., Deshpande, M., & Tan,
P. (2000). Web usage mining: Discovery and
applications of usage patterns from Web data.
SIGKDD Explorations, 1(2), 12-23.
Spiliopoulou, M. (2000). Web usage mining for
Web site evaluation. Communications of ACM,
43(8), 127-134.
Stumme, G., Taouil, R., Bastide, Y., Pasquier, N.,
& Lakhal, L. (2000). Fast computation of concept
lattices using data mining techniques. Proceedings of the Knowledge Representation Meets
Databases Conference (KRDB00), Berlin.

This work was previously published in Web Mining: Applications and Techniques, edited by A. Scime, pp. 276-306, copyright
2005 by IGI Publishing, formerly known as Idea Group Publishing (an imprint of IGI Global).

759

760

Chapter 2.24

Modeling Variant User Interfaces

for Web-Based Software
Product Lines
Suet Chun Lee
BUSINEX, Inc., USA

ABSTRACT
Software product line (SPL) is a software engineering paradigm for software development. SPL
is important in promoting software reuse, leading
to higher productivity and quality. A software
product within a product line often has specific
functionalities that are not common to all other
products within the product line. Those specific
functionalities are termed variant features in a
product line. SPL paradigm involves the modeling
of variant features. However, little work in SPL
investigates and addresses the modeling of variant features specific to UI. UML is the de facto
modeling language for object-oriented software
systems. It is known that UML needs better support in modeling UIs. Thus, much research developed UML extensions to improve UML support
in modeling UIs. Yet little of this work is related
to developing such extensions for modeling UIs
for SPLs in which variant features specific to user
interfaces (UI) modeling must be addressed. This

research develops a UML extension, WUIML, to

address these problems. WUIML defines elements
for modeling variant features specific to UIs for
Web-based SPLs. The model elements in WUIML
extend from the metaclass and of the UML2.0
metamodel. WUIML integrates the modeling
of variant features specific to UIs to UML. For
example, in a Web-based patient registration SPL,
member products targeting British users may use
British date format in the user interface, while
member products targeting United States users
may use United States date format in the user
interface. Thus, this is a variant feature for this
product line. WUIML defines a model element,
XOR, to represent such exclusive or conditions
in a product line user interface model. WUIML
would reduce SPL engineers efforts needed in UI
development. To validate the WUIML research
outcome, a case study was conducted. The results
of this empirical study indicate that modeling UIs
for Web-based SPLs using WUIML is more effective and efficient than using standard UML.

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Modeling Variant User Interfaces for Web-Based Software Product Lines

Introduction
Software product line (SPL) (Chastek, Donohoe,
Kang, & Thiel, 2001; Clements & Northrop, 2002;
SEI, 2005a) is a software engineering paradigm
to develop software products. One important
step in the SPL paradigm is the modeling of the
functional features of software products across
the product line. The features are called common
core. An even more important step in the SPL
paradigm is the modeling of the specific functional
features within a particular member product in a
product line. These specific functional features
are called variant features because they are the
features that differentiate member products in the
product line. Then based on the model, a product
is assembled by reusing the common core and
selected variant features.
Unified Modeling Language (UML) (OMG,
2003b, 2004; Rumbaugh, Jacobson, & Booch,
2005) is a standard object-oriented modeling language. UML includes multiple views and diagram
types to capture software functionalities from
user perspective. However, UML seems to have
not been developed for modeling user interface
specific issues (Kovacevic, 1998; Silva & Paton,
2003). One of the usages of user interface models
is that, in model-based user interface management
systems (MB-UIMSs) (Griffiths et al., 2001; Szekely, Sukaviriya, Castells, Muthukumarasamy,
& Salcher, 1996), user interface models can be
used to generate user interface codes. There are
extensions of UML (Blankenhorn & Jeckle, 2004;
Nunes, 2003; Silva, 2002) to make UML better
support user interface modeling. Yet, these extensions often assume the modeling of a single system
instead of a SPL. On the other hand, although
standard UML (OMG, 2003b, 2004) seems to
have not been developed to support the modeling
of SPLs, there are works (Gomaa, 2004; Gomaa
& Gianturco, 2002; Ziadi, Hlout, & Jzquel,
2004) on extending UML to improve UML
supports in modeling SPLs. Yet, these works do
not focus on user interface modeling. Currently,

many software products are Web-based. However,

some (Silva, 2002) observe that there are specific
modeling challenges for modeling user interfaces
of Web-based software systems.
Thus, it is not clear how to model variant
features for user interface specific issues in Webbased software product lines. This is an important
barrier to overcome if software product line development of Web-based products is to take greater
advantage of software reuse objectives: increased
quality, decreased effort, or decreased time to
market. Therefore, this paper is concerned with
reporting research about developing a UML extension, Web User Interface Modeling Language
(WUIML) that decreases effort by increasing effectiveness and efficiency needed in using UML
to model user interfaces for Web-based software
product lines. User interface development has been
found (Myers, 1989) to account for a significant
amount of overall software development work.
WUIML would improve SPL software engineering paradigm in the user interface development
perspective by reducing engineers efforts needed
in user interface modeling.

Background and
related work
Unified Modeling Language
Unified Modeling Language (UML) (Booch,
Rumbaugh, & Jacobson, 1999; OMG, 2003b, 2004;
Scott, 2004) is a graphical language for specifying
software systems. UML is a standard of the Object
Management Group (OMG; see http://www.omg.
org). The most current version for UML is UML
2.0 (OMG, 2003b, 2004). This research considers
UML in UML 2.0 context.
UML is a standardized notation for objectoriented development. UML consists of views,
diagrams, model elements, and general mechanisms. Views are used to present different aspects
of complex systems from both the system in the

761

Modeling Variant User Interfaces for Web-Based Software Product Lines

problem and the system in the solution. Tacitly,

UML defaults to system as meaning the system
in the solution. This is consistent with the classical
use cases modeling and design process model.
Therefore, whenever the term system is used
in this paper, this usual UML use cases default
as system in the solution is implied. Moreover,
each view requires a number of diagrams, such as
use case diagram, class diagram, and interaction
diagrams. Each diagram captures a particular
aspect of a system. Views that are often used
with UML include: use case view, logical view,
implementation view, process view, and deployment view. Use case views describe the functionality a system should deliver from external
actors point of view. An actor is a human user
or another system external to the system. A use
case is often used to describe functions requested
for a system. Logical views are used to describe
a systems functionality. Use cases often are the
first set of models being created for a system.
However, some found common pitfalls in use
cases for one who is new to use case modeling
(Lilly, 2000; Rosenberg & Scott, 2001).
Modeling concepts are captured in model elements. Example model elements are class, object,
state, node, package, and component (Eriksson,
Penker, Lyons, & Fado, 2004). Model elements
are used in diagrams. A model element can have
a visual representation. There are rules that limit
types of model elements for different diagrams.
Relationships between model elements are specified by relationships, such as association, generalization, dependency, and aggregation.
Additional information that cannot be represented using model elements is represented using
UML general mechanisms, such as adornments,
comments, tagged values, and properties (Eriksson et al., 2004). Adornments are used to add visual
impact to the element. For example, underline
an instance of a type is an adornment. Comments
can be placed anywhere in a diagram. Comments
often contain explanations or questions to resolve issues at a later time. Model elements have

762

properties to store data values about an element.

A property is defined with a name and a value
called a tagged value. Properties are used to add
additional specifications about element instances
that are not normally shown in the diagram.
UML can be extended or adapted by its extension mechanisms: stereotype, tagged values, and
constraints. A stereotype defines a new kind of
model element based on existing model elements.
A stereotype is described by placing its name as a
string and within a pair of guillemots. For example,
a class with the stereotype <<Radio>> is read as
a class of the Radio stereotype, meaning that
it is a radio type of class. The particular characteristics of a Radio class must have been defined
when the stereotype is defined. A constraint is a
restriction on an element that limits the usage of
the elements. A constraint is enclosed in a curly
bracket, for example, {student = Dorothy}.
Alternatively, constraints can be written in Object Constraint Language (OCL) (OMG, 2003a).
A UML profile (OMG, 2003b) is a stereotyped
package that contains model elements that have
been customized for a specific purpose using extension mechanisms, such as stereotypes, tagged
values, and constraints. A profile can only extend
an existing UML model element using stereotypes, tagged values, or constraints. A profile
cannot create new model elements that were not
previously defined in the UML 2.0 infrastructure
specification (OMG, 2003b).
UML 2.0 is defined in Meta-Object-Facility
(MOF) (OMG, 2002). MOF is an abstract language
for defining modeling languages. Therefore, another way to extend UML is using MOF. MOF
extends UML by defining new metamodel elements. This approach is often called metamodel
approach. In metamodel approach, new model
elements can be created. With metamodel approach, one can also create model elements that
change UML symbols and semantics.

Modeling Variant User Interfaces for Web-Based Software Product Lines

Software Product Line

Software product line (SPL) (SEI, 2005a) is a
software engineering paradigm (SEI, 2005b).
The paradigm mainly includes two software
engineering processes: product line engineering
and member product engineering. The product
line engineering process involves a number of
techniques in software engineering, such as
domain analysis, requirements engineering,
product architecture, and component development (Clements & Northrop, 2002). The product
engineering process involves the configuration
(assemble) of member products from the product
line assets. The SPL paradigm depends on the
notion of variability and variability mechanisms.
In SPL, the term variability is used to indicate
the extent of differences in user visible software
properties among member products. A user visible
software property is a feature. A feature of a member product that is different from other member
products is a variant feature. Therefore, a variant
feature is a feature only to one or more (but not
all) products in a SPL. Variability mechanisms
are a set of methods and steps that can be applied
to develop variant features for SPLs.
The user interface of member products of a
SPL may vary in one or more user interface specific features. Examples of user interface specific
features are the layout of the user interface, the
functionality of the user interface, the interaction
of the user interface to and from the user, and so
forth. The user interface specific variant features,
like other non-user interface specific features, also
need to be captured in a SPL model.
It seems that issues in user interface developments have been largely ignored by the SPL
researches. It is not clear how to go about applying the software product line paradigm in
user interface modeling. WUIML developed in
this research is differentiated from other related
works (Gomaa, 2004; Gomaa & Gianturco, 2002;
Shin, 2002; Webber, 2001; Ziadi et al., 2004) in
SPL researches in that WUIML is addressing the

modeling of user interfaces in a SPL using UML.

It is known (Silva & Paton, 2003) that UML needs
better support in modeling user interfaces. The
related works extend UML to improve support
on modeling SPL, yet those works have not addressed the modeling issues in user interfaces
for SPL. User interface development has been
found (Myers, 1989) to account for a significant
amount of overall software development work.
WUIML would improve SPL software engineering paradigm in the user interface development
perspective by reducing engineers efforts needed
in user interface modeling.

Web User Interface Modeling

User interface is a computer-based media to
facilitate communication between users and a
software system (Marcus, 2002). Early user interfaces in software systems are text or form-based.
Modern software systems however use graphical
user interfaces (GUI) implemented according
to programming packages, such as Java Swings
classes (Eckstein, Loy, & Wood, 1998). More
recently, HTML is used to create Web-based
user interfaces that can be rendered by Internet
browsers (Sommerville, 2001).
The user interfaces of a Web-based software
product are called Web user interface (WUI)
in this research. The modeling of a WUI is the
development of an abstract understanding of the
requirements of the WUI and represents it in an
abstract notation. The modeling of WUI using
WUIML does not provide a visual look of a
WUI, thus, WUI modeling is not the design of a
WUI. Currently, a major use of WUI modeling
is to provide inputs for WUI code generation.
In particular, WUI models resulting from WUI
modeling are fed into a model-based tool. The
tool either auto-generates (or semi-auto-generates) WUI implementations for the WUI models
(Behrens, Ritter, & Six, 2003; Gmez & Cachero,
2003).

763

Modeling Variant User Interfaces for Web-Based Software Product Lines

UML-Based User Interface Modeling

One approach to user interface modeling and
specification methods is to extend UML notations
(Behrens et al., 2003; Blankenhorn & Jeckle, 2004;
Hennicker & Koch, 2001; Jacobson, Christerson,
Jonsson, & Overgaard, 1992; Nunes, 2003; Phillips, Kemp, & Kek, 2001; Scogings & Phillips,
2001; Silva & Paton, 2003).
A Web UI development method is proposed
in Behrens et al. (2003). The method consists of
two tasks: UI requirements modeling and mapping UI requirements model to UI architecture.
The modeling of UI requirements includes the
specification of use cases (called work units in
the paper), and then the work units are modeled
using user-defined UML stereotypes representing
user interface elements. A UI element is modeled
as a <<scene>>. A <<scene>> may consist of zero
or more <<class view>>. A user interface is then
the composition of different <<scene>> elements.
The navigations between user interfaces are based
on the parent-child relationship between the user
interfaces, domain requirements, and usability
aspects. In Hennicker and Koch (2001), a UML
profile for modeling user interfaces for Web applications is proposed. The UML profile includes
UML stereotypes to model the presentation (look
and feel) and navigation aspects of Web UIs. The
Object-Oriented Hypermedia (OO-H) Method
(Gmez & Cachero, 2003) extends UML with a
set of new views for modeling Web user interface
model. In Silva and Paton (2003), a UML extension, UMLi, was developed to support user interface modeling for interactive systems. However,
UMLi has yet to address Web UI as indicated in
Silva (2002) that there are additional user interface
modeling properties for Web applications.
The major difference between this work and
the related works (Behrens et al., 2003; Gmez
& Cachero, 2003; Hennicker & Koch, 2001; Jacobson et al., 1992; Silva, 2002; Silva & Paton,
2003) in user interface modeling is that WUIML
addresses the modeling of user interfaces for SPLs

764

and the related works focus on the modeling of

user interfaces for a single system. In addition, the
modeling elements in WUIML capture the Web
user interface properties specific to Web software
products according to important Web standards
(Axelsson, Epperson, Ishikawa, McCarron, Navarro, & Pemberton, 2003; Bos, Celik, Hickson,
& Lie, 2004; Dubinko, Leigh, Klotz, Merrick,
& Raman, 2002; Lie & Bos, 1999; Meyer, 2000;
W3C). WUIML improves SPL software engineering paradigm by reducing efforts needed by SPL
engineers in user interface modeling for SPLs.

Web User Interface Modeling

Language (WUIML)
WUIML improves UML support on user interface
modeling for Web-based software product lines.
User interface modeling falls into the scope of
user interface development. The user interface
models specified in WUIML form one generic
user interface software product line asset for the
entire main software product line. The idea is to
have user interface models as core assets. Then
to develop user interfaces for member products,
one would reuse all common user interface components and reuse selectively the variant user
interface components.
In this research, the requirements analysis
process for user interfaces is as follows. Given
software requirements for a software product line,
user interface related use cases from product line
requirements are derived. A use case that is user
interface related can be developed into many user
interface scenarios. Each scenario is due to one or
more variant user interface aspects. A user interface scenario involves one or more user interfaces.
Any interaction between a human user and a user
interface or between a user interface and a backend application is also captured in the use case
scenario. Once the user interfaces are identified,
one can decide on the user interface components.
After eliciting and describing the scenarios, the

Modeling Variant User Interfaces for Web-Based Software Product Lines

next step is to identify the relationships between

user interfaces. The interacting relationships
between user interfaces are captured in extended
activity diagrams where nodes in the activity
diagrams can represent variant user interfaces.
The extended activity diagram shows the variant
interactions between user interfaces.
In order to specify common and variant
features for user interfaces, WUIML defines
stereotyped relationships between modeling
elements. These new relationships are selected
syntheses and extensions from Jacobson, Griss,
and Jonsson (1997) and Anastasopoulos and
Gaceks (2001) variability mechanisms. In addition, the variant notations and rules of the FODA
(Cohen, Stanley, Peterson, & Krut, Jr., 1992) technique have been adopted and incorporated into
WUIML. In addition, this research developed a
new variability mechanism called WUIAggregation. WUIAggregation models a WUI that is an
aggregation of different parts of other WUIs. For
example, a portal WUI is a WUIAggregation of
a number of other WUIs. The difference between
WUIAggregation and the UML aggregation
is that WUIAggregation limits its aggregated
components to be instances of WUIComposite
or specific WUIElements while the standard
UML aggregation does not have this restriction.
This restriction helps modelers on deciding what
components are appropriate for the aggregation.
The variability mechanisms are used to generate
variants on user interface aspects. User interface
aspects include (but not limited to) user functional
requirements, style, presentation, layout, events,
data model, and constraints.
The structural contents of WUIML are built
from various World Wide Web Consortium (W3C;
see http://www.w3c.org) specifications: XHTML
(Altheim, Boumphrey, McCarron, Schnitzenbaumer, & Wugofski, 2001; Altheim & McCarron,
2001; Axelsson et al., 2003; Powell, 2003; Raggett,
Hors, & Jacobs, 1999; Sauers & Wyke, 2001;
W3C, 2002), XForms (Dubinko, 2003; Dubinko
et al., 2002; Khare, 2000), and Cascading Style

Sheets (CSS) (Bos et al., 2004; Lie & Bos, 1999;

Meyer, 2000, 2003; Schengili-Roberts, 2004;
Schmitt, 2003).
WUIML is also built on the UML 2.0 specifications (OMG, 2003b, 2004) and is an extension
to the metamodeling approach of the UML 2.0.
In particular, WUIML extends metaclass and
BasicActivity of the UML 2.0 metamodel. The
metaclass extension is achieved via stereotype
extension mechanism. The extension to the BasicActivity is achieved via MOF.
A WUI structurally consists of multiple user
interface elements, such as user interface visual
elements (such as a button, a checkbox, etc.),
hypertext links, images, text, and so forth. These
elements are mapped to stereotyped UML
classes that extend the UML2.0 metaclass.
Events (sometimes termed triggers) occur
when user interface elements interact with user
actions. Actions are mapped to UML interfaces.
An interface defines a set of operations triggered by events. Events and actions characterize
a user interface interaction and are modeled as
operations of metamodel elements. Interactions
between user interfaces are modeled using UML
activity diagrams. In addition, UML activity
diagram is extended to model variant interactions
in WUIML.

Basic WUIML Elements

Figure 1 defines the Basic WUIML Elements in
UML 2.0 notations. These model elements extend
from the metaclass of the UML2.0 metamodel.
These model elements have attributes and operations. The attributes are used to model the static
features of user interfaces; for example, an attribute can be used to describe the background
color of a user interface element. The operations
are used to model the dynamic aspects of user
interfaces, for example, events (e.g., a user clicks
a button can be tracked by an event listener) and
actions (e.g., an error message box popup can
be implemented by a method) can be modeled

765

Modeling Variant User Interfaces for Web-Based Software Product Lines

Figure 1. Basic WUIML elements

< < m e ta cla s s> >
C la s s

<<stereotype>>

D imension

0..*
<<stereotype>>

Position

<<stereotype>>

StyleD ecorator
1

0..*

<<stereotype>>

StyleS trategy

0..* 1

<<stereotype>>

0..1

0..1
0..*

WUITemplate

1
1

<<stereotype>>

<<stereotyp>> 0..*

W UIC omposite

Style

1
1
<<stereotype>>

<<stereotype>>

0..*

0..*

Table

<<stereotype>>

Upload

W UIElement

<<stereotype>>

PasswordInput

<<stereotype>>

C ommand

<<stereotype>>
<<stereotype>>

<<stereotype>>

Range

TextArea

<<stereotype>>
<<stereotype>>

SingleSelect

<<stereotype>>
<<stereotype>>

Submit

as operations. Figure 2 shows the class diagram

of the WUIElement and the Command model
elements.
A WUI is modeled as a WUITemplate. A
WUITemplate may be composed of one or more
of WUIComposite. A WUITemplate may also
be composed of one or more WUIElement and
one or more WUIComposite. A WUIComposite
has one or more specific WUI modeling elements (such as Submit, Input, etc.) that extend
from the WUIElement. The style, layout, and
position within a WUI of a WUIElement can be
described using the StyleDecorator model element. A StyleDecorator composes of a number
of Dimension element, Position element, and the

766

Input

C ontent
1

MultiSelect

Output

<<stereotype>>

0..*
<<stereotype>>

Hyperlink

Style element. The Dimension (how large or small

the WUI element is), Position (where the WUI
element is positioned within a WUI), and Style
(such as what is the foreground color of the WUI
element) elements together model abstractly the
style of a WUI element. A set of StyleDecorator
elements forms a StyleStrategy. A StyleStrategy
element describes the style of a WUIComposite
element. A set of StyleStrategy elements models
the style of a WUITemplate.
These WUIML elements integrate Web
properties according to important Web standards
(Axelsson et al., 2003; Bos et al., 2004; Dubinko
et al., 2002; Lie & Bos, 1999; Meyer, 2000; W3C)
to the modeling of Web-based user interfaces for

Modeling Variant User Interfaces for Web-Based Software Product Lines

Figure 2. The WUIElement and the Command element

< < s tereoty pe> >

W UIE le m e nt

id
label
value
tabIndex
im age

getId()
s etId()
getLabel()
s etLabel()
getV alue()
s etV alue()
ac tivate()
foc us In()
foc us O ut()
getHelpM es s age()
getHintM es s age()
notify V alueChanged()
notify ControlV alid()
notify ControlInvalide()
notify ReadO nly ()
notify ReadW rite()
notify Required()
notify O ptional()
notify E nabled()
notify Dis abled()
s howM es s age()
s etF oc us ()
res etV alue()
loadW indow()
s etIndex ()
getIndex ()
getIm age()
s etIm age()

SPLs. With regard to the modeling of the static

aspects of user interfaces, what WUIML adds to
the standard UML is like what application programming interfaces (APIs) add to a programming
language. For example, the Swing API (Eckstein
et al., 1998) provides the building blocks for
programmers to code user interfaces. Similarly,
WUIML provides the building blocks for modelers
to model user interfaces. When using Swing, a programmer does not have to construct fundamental
user interface elements, such as textboxes, radio
buttons, and so forth. The programmer would

< < s tereoty pe> >

C o m m a nd
ac tion
m ethod
s end()

just use them. Similarly, when using WUIML, a

modeler does not have to model the fundamental
user interface elements, such as textboxes, radio
buttons, and so forth. When using standard UML,
those fundamental user interface elements must
be modeled first. In addition, WUIML provides
a way to conveniently model the layout, size, and
position (where it is within a user interface) for a
particular user interface element. Currently, there
is no standard way in using UML to model the
layout and style of a Web-based user interface.
WUIML defines a way for modeling the layout
and style of a Web-based user interface.
767

Modeling Variant User Interfaces for Web-Based Software Product Lines

Figure 3. An example WUI model using WUIML basic elements

Example I: Web User Interface

Modeling Using Basic WUIML
Elements
In a SPL, a user interface feature may vary among
the member products. For example, consider a SPL
of online patient registration systems. A member
product to be deployed for U.S. customers would
have a user interface customized for the U.S. address format. On the other hand, member products
to be deployed for Hong Kong customers would
have a user interface customized for the Hong
Kong address format.
Figure 3 shows a product line Web user
interface model that consists of instances of
<<WUIComposite>> elements. An instance of
<<WUIComposite>> may compose of a number
of fundamental user interface elements (such as a
textbox, an input box, a listbox, etc.) but they are
not shown in Figure 3. A <<WUIComposite>>
can be considered as a portion of a user interface
called user interface fragment in this research. A

768

user interface is composed of a number of user

interface fragments.

Modeling Variant Interactions

WUIML defines elements to model the different
interactions between a user interface and its backend process. If an MVC architecture (Gamma,
Helm, Johnson, & Vlissides, 1995) is the example,
then the back-end application is the controller and
the model. For the same functional task, such as a
WUI for remittance of an online banking service
SPL, each member product may implement it in
a slightly different way. The difference would
lead to differences in the WUI related to the task
among member products. The differences in WUI
would in turn lead to different patterns of user
interactions to the WUI as well as different implementation of back-end processes. For example,
to send money to a foreign country, the WUI
must collect the destination country (as country
code) and currency (currency code) information
from the user. On the other hand, to send money

Modeling Variant User Interfaces for Web-Based Software Product Lines

Figure 4. Extension to UML 2.0 Activity

Re d e fina b le E le m e nt
(fro m K e rne l)

A ctivi ty (fro m
Ba sicBe ha vio rs)

is ReadO nly : B oolean = fals e

0..1

{subs ets o wnedE lemen t}

*

+a ctivity

{s ubs ets o wner}

0.. 1
0..1

+no d e

0..1

A ctivity

A ctivityNo d e

0..*

A ctio n
E xe cuta b le No d e

0..*

E le m e nt (fro m
K e rne l)

V a ri antA ctio n
0.. *
0.. 1

0..*

S P L A ctio n

is Co m pos ite : B oolean = t rue

domestically within the U.S., the WUI does not

need to collect country code or currency information. The different user interactions to a user
interface in a SPL are called variant interactions.
User actions, such as clicking a submit button in
a WUI, may trigger transitions between WUIs.
In this research, UML activity diagram is used
to model the interactions between WUIs.
WUIML defines two new metamodel elements using MOF to model variant interactions.
The two new elements are VariantAction and
SPLAction. In UML, a node in an activity diagram is represented by the ActivityNode element.
The ActivityNode element has three children:
ExecutableNode, ObjectNode, and ControlNode.
Object Nodes represent the objects involved in the
activity diagram. The ControlNode signifies the
control flow on the activity diagram; for example,
a fork bar is a ControlNode. The ExecutableNode
has a child, the Action element. The Action element in UML is the representation of a node
that characterized by one or more actions. Since

interactions are characterized by actions, WUIML

extends the UML Action element to model variant interactions.
Figure 4 shows the new metamodel elements,
VariantAction and SPLAction, and how they relate
to other elements in the activity class diagram.
The VariantAction element is a specialization of
the UML Action element. The SPLAction is a
specialization of the Element metamodel element
from the UML2.0 kernel package. A SPLAction
element contains one or many Variant Action
elements. A UML Activity is now also consisting
of one or more of the SPLAction and VariantAction elements.
Suppose we have an activity called submit
order. The submit order activity is very common in Web-based shopping applications because
a user must submit an order so that the order can
be processed. But there are variants in the submit
order activity. For example, most online shopping sites accept only credit card payments, thus,
those applications may want the submit credit

769

Modeling Variant User Interfaces for Web-Based Software Product Lines

Figure 5. SPLAction symbol

S u bm it C re d it C a rd O rd e r
S u bm it C O D O rd e r
S u b m it C h e ck O rde r
S u b m it P u rch ase O rd er
S u b m it O rde r

card order. In another case, corporate purchasing often submits purchase orders, and then pays
the bill after receiving the invoice. In this case,
submit purchase order may be desired. Yet
another variant is in the case where a customer
wants to receive the purchase and check it first
before making a payment, these customers may
want the submit Cash-On-Delivery (COD) order.
Yet some merchants may offer a combination of
these payment methods. There are a number of
variants in the submit order activity, and we
need to be able to model the variants. This paper
extends UML activity diagram to model the
requirements as follows.
Figure 5 shows a graphical symbol for the
SPLAction. The symbol exposes the base action
and its four variants actions. The shape of the base
submit order action is filled, while the variants
are not. The base Submit Order action defines
the common logics that are applicable across all
of the variants. Each variant differs from other
variants by some features. Our concern is on
user interface modeling; it is clear that the user
interface to collect information on a credit card
order is different from the user interface to collect
a COD order in some way.
Sometimes, there are variant logics and behaviors within a variant member product that capture
different execution scenarios. Those logics and
behaviors can be modeled using standard UML
activity diagrams. For example, suppose there
is a WUI that provides user interface elements

770

to accept either credit card based order option

or purchase order option. If a customer chooses
credit card order, the next WUI displayed is to
collect the customers credit card information. If
a customer chooses purchase order based order
option, then the next WUI displayed is to collect
the customers purchase order related information. Thus, the softwares behavior varies due
to different execution scenarios. This variant is
within a product.
These variant logics and behaviors, due to
different execution scenarios, often are easy for
one to confuse them with the product line variant
logics and behaviors. To identify the product line
variants, one must consider the same particular
variant logics and behaviors across multiple member products at the same time. If the variant logics
and behaviors work the same across the member
products, then the variant logics and behaviors
are not a product line variant.

Example II: Modeling Variant

Interactions
Figure 6a shows an activity diagram for a SPLAction for an online e-retailer SPL. The SPLAction
is about Submit Order. The SPLAction consists
of four variant actions: Submit Credit Card
Order, Submit COD Order, Submit Check
Order, and Submit Purchase Order. A member
product of this product line would have the same
activity but with only one variant of the submit
order action. Figure 6b shows the activity diagram
for a member product that allows users to submit
Credit Card Orders.

WUIML Elements for Variability

Mechanisms
Figure 7 shows the WUIML elements for variability mechanisms that are based on Jacobson,
Griss, and Jonsson (1997) and Anastasopoulos
and Galeks (2001) variability mechanisms and the
variant notations and rules of the FODA (Cohen

Modeling Variant User Interfaces for Web-Based Software Product Lines

Figure 6a. Exemple SPL variant interactions

S ubm it C redit C ard O rder
S ubm it C O D O rder
S elect item s

C heck out

S ubm it C heck O rder

S ubm it P urchase O rder
S ubm it O rder

Figure 6b. Example of a member product variant interaction

S elect item s

<< V ariantA ction> >

S ubm it C redit C ard O rder

C heck out

Figure 7. Model elements for variability mechanisms

< < m etaclass> >

D ependency

< < stereotyp e > >

< < stereo typ e > >

D elegation

< < stereotyp e > >

W U ID erive

R equireP resen ceO f

< < stereo typ e > >

W U IE xten d

< < ste reotyp e > >

U se

< < ste reotyp e > >

T em plateInstantiatio n

< < ste reotyp e > >

C lass

< < stereo typ e > >

W U IA g gre gatio n
1

1
< < m e ta class> >

E xp ression

< < stereotyp e > >

< < stere otyp e > >

P aram e terization

C onfigura tion
1
0..*
< < m etacla ss> >

C o nstraint

1
0..*

1 ..*
< < m etaclass> >

P roperty

< < m etaclass> >

G ene ralization

< < stereo typ e > >

W U IG eneralization

< < C onstraint> >

XOR

< < C onstra int> >

O ptio n a lE le m ent

< < C onstrain t> >

OR

771

Modeling Variant User Interfaces for Web-Based Software Product Lines

et al., 1992) technique. Delegation, WUIDerive,

WUIExtend, Use, TemplateInstantiation, and
RequirePresenceOf are stereotypes extended
from Dependency metaclass. WUIAggregation,
Configuration, and Parameterization are defined
as stereotypes of Class stereotype. WUIGeneralization is a stereotype of the Generalization
metaclass. OptionalElement is a constraint to
express that an element is optional in the model.
XOR and OR are Constraints. A Parameterization
element is composed of one or more Property. A
Configuration element is composed of one or more
Property and zero or more Constraint.
The Use model element is used to specify that
the source user interface element (e.g., a user interface fragment to capture the travel date) depends
on the target user interface element (e.g., a user
interface of a calendar for setting a travel date).
The Configuration model element models a set
of parameters and rules. These parameters and
rules manipulate the composition of the software
component of a software system or the setting
of software components or software systems, so
that variants can be created. The Parameterization model element models a set of parameters
that vary the features of a software product. The
TemplateInstantiation model element is a special
case of the Parameterization model element. In
TemplateInstantiation, the parameters are templates. WUIExtend allows small extensions in
functionality or look to a user interface due
to new requirements. WUIDerive extends UML
derive stereotype to apply on all basic WUIML
elements. Delegation is used to model the situation in which a Web user interface is acting as a
representative of another Web user interface in
performing certain operations. WUIAggregation
allows one to model the situation in which various user interface fragments from various Web
user interfaces are used to compose another Web
user interface based on a business goal. WUIAggregation differs from UML aggregation in that
the aggregated components in WUIAggregation
must be an instance of WUIComposite or specific

772

WUI elements. WUIGeneralization models the

is-a-kind-of relationship between two user
interfaces. WUIGeneralization differs from
UML generalization in that the generalization
in WUIGeneralization applies to WUI elements
only. OptionalElement is used to specify that a
user interface element is optional in the model. OR
models the inclusive or constraint. XOR models
the exclusive or constraint. RequirePresenceOf
models the situation in which a user interface
feature depends on the presence of another user
interface feature. For example, to model the style
of a column within a table, it requires the presence
of a table element.

Example III: Modeling Variant Web

User Interface Features
To illustrate WUIML for modeling variant WUI
features, let us continue the example shown in
Figure 3. But this time, we apply the variability
mechanisms using the corresponding variability
mechanism model elements defined in WUIML.
Figure 8 shows a more detailed WUIML model
for the online patient registration systems SPL.
Figure 3 shows the common user interface features
across the product line. Figure 8 shows the variant
user interface features among member products
within the product line in addition to the common
features across the product line.
Applying variability mechanisms and the
corresponding WUIML notations, variant WUI
features are specified in Figure 8. The variant
notations in WUIML are only used in product
line WUIML models to show the variant features among member products. Thus, the variant notations in WUIML are not to appear in
the WUIML model of a member product. For
example, notice that the USAddress and HKAddress WUIComposite elements appear in Figure
3 are now specified with exclusive or constraint.
This indicates that if a member products WUI
uses USAddress format, then it will not use the
HKAddress format. TemplateInstantiation vari-

Modeling Variant User Interfaces for Web-Based Software Product Lines

Figure 8. WUIML model of the online patient registration systems SPL

ability mechanism is used to create two variant

WUIComposites of PatientInfoReporter. The patient information report can be in graphic format
or summary format.

Modeling Dynamic Aspects of Web

User Interfaces
In addition to interactions as described in the section Modeling Variant Interactions, the dynamic
aspects of WUIs also include events and actions.
In this paper, event literally means something
that takes place; an occurrence (Pickett et al.,
2000). In this paper, an event means an occurrence
that takes place in the user interface. Example
occurrence may be a mouse click on a button,
a selection on a selection box, and so forth. The
term action literally means the state or process
of acting or doing (Pickett et al., 2000). In this
paper, an action means doing some user interface
related tasks in response to user interface events.
An example action may be popping up a message

box on the user interface, setting a value for a

text input box, and so forth. Moreover, an event
is observed by a user interface control, such as a
button, and then the user interface control triggers
an action. For example, when a button is clicked,
a message box is popped up to display a message.
In this scenario, the button is the user interface
control that is observing the event: clicking the
button. The popping up of a message box is the
action being triggered by the button click event.
The concept of events and actions described in this
section are built on the XForms specification.
XForms specification includes a set of events
and actions. Many events and actions have to
do with XForms processing and are not suitable
for use at modeling level. For this reason, only
a selected set of events and actions are taken
as the bases for building WUIML elements for
modeling the dynamic nature of a user interface.
In WUIML, events are modeled as operations of
a basic WUIML metamodel element. Each basic
WUIML metamodel element has a distinct set

773

Modeling Variant User Interfaces for Web-Based Software Product Lines

Figure 9. Patient Registration WUIML Model

of events. Extra events can be added as needed

by the modeler at design time as extensions. An
action models the resulting behavior of a basic
WUIML element in response to events. Actions
are represented as operations of a basic WUIML
element.

Example IV: Modeling Product Web

User Interface Using WUIML
Figure 9 shows an even more detailed WUIML
model for the online patient registration systems
SPL than Figure 8. In Figure 9, more variability
mechanisms are applied to create variants. In
addition to the variability mechanisms shown in
Figure 8, the variability mechanisms applied in

774

Figure 9 include WUIExtend, Use, Configuration,

WUIAggregation, Delegation, and WUIDerive.
Once we have a SPL WUIML model, we can derive
product WUIML model from the SPL WUIML
model. A product WUIML model is a WUIML
model that reuses all the common features of a SPL
WUIML model but reuses only selected variant
features from a SPL WUIML model. Figure 10
and Figure 11 show two product WUIML models
derived from Figure 9.
The variant patient registration WUIML model
shown in Figure 10 is created by taking all the
common model elements (i.e., Email, LoginInfo,
Address, PhoneNum, NameInfo, OtherContent,
PatientInfoReporter, and SubmitRequest) and
selected variant user interface features (i.e.,

Modeling Variant User Interfaces for Web-Based Software Product Lines

Figure 10. A product WUIML model for the online patient registration systems

USAddress, USPhoneNum, and NameInfo_USStyle) from Figure 9.

Figure 11 shows another variant product
WUIML model derived from Figure 9. Compare
Figure 11 with Figure 10; Figure 11 shows more
variant user interface features. Figure 11 includes
the BritishDate WUIComposite that is extended
from DateRegistered WUIComposite.

Validation
The goal of this research is to improve UML support in modeling user interfaces for Web-based
SPLs. The improvement goal is to have the new
WUIML method exhibit decreased effort needed
by increasing effectiveness and efficiency. The
improvement will reduce the SPL engineers
efforts needed for developing user interfaces. To
exhibit that this improvement goal is met, a case
study research validation method is applied to
supply supporting evidence for the thesis (research
hypothesis).

Case Study Research Method

Using case study (Lynch, 1999; Tellis, 1997a,
1997b; Yin, 2003) as an empirical research validation method has increasingly been adopted in
software engineering research (Lee, 2004). This
research uses a case study research method to
validate the research hypothesis.
The rationale of a case study research method
is the notion of analytical generalization.
Analytical generalization depends on whether the
findings of a case study can be replicated. If the
findings of a case can be replicated, an analytical
generalization may be drawn.
In analytical generalization, the results of a
case study should be compared with a proposition. For example, in this case study, the results
of a multiple-case study are to compare with
proposition 1 and proposition 2 (proposition
1 and proposition 2 will be discussed later).
Multiple cases are needed to test a proposition
through replications of findings in order to lead
to an analytical generalization (Yin, 2003). Many
research methods, such as a survey, depend on
statistical generalization (Yin, 2003). In statisti-

775

Modeling Variant User Interfaces for Web-Based Software Product Lines

Figure 11. Variant patient registration Web user interface

cal generalization, an inference is made about

a population (or universe) base on the empirical data collected about a sample (i.e., surveys)
(Yin, 2003). Since a case study is not based on
statistical generalization, a sampling logic should
not be used in a case study research method. As
a result, the number of cases in a multiple-case
design and the typical criteria regarding sample
size are irrelevant (Yin, 2003).
The evidence from multiple-case studies
makes the overall study more robust (Yin, 2003).
In this empirical study, multiple case studies
were conducted. In a multiple-case study, each
case must be selected to predict similar results
(a literal replication) or to produce contrasting
results but for predictable reasons (a theoretical
replication) (Yin, 2003). A literal replication
explains the conditions under which a particular
phenomenon is likely to be found. A theoretical
replication explains the conditions when it is not
likely to be found. This empirical study relies on
literal replications. The two case studies conducted

776

are expected to predict similar results. The study

is trying to show that under the condition (i.e.,
using WUIML in modeling WUIs for a SPL), a
particular phenomenon (i.e., increased modelers
effectiveness and efficiency in modeling) is likely
to be found. For each case in the multiple-case
study, the study indicates the way a particular
proposition was demonstrated (or not demonstrated). Across cases in the multiple-case study,
the study indicates the extent of the replication
logic and the reasons on the prediction of supporting or contrasting results (Yin, 2003).

Case Study Design

The goal of this empirical study is to investigate
the effectiveness and efficiency of WUIML.
This empirical study uses a case study research
method (Lee, 2004; Tellis, 1997a, 1997b; Yin,
2003) as the validation method. The design of a
case study is characterized by five important case
study components (Yin, 2003): a studys research

Modeling Variant User Interfaces for Web-Based Software Product Lines

questions, study propositions, units of analysis,

the logic linking data (results) to the propositions,
and the criteria for interpreting the findings.

Studys Research Questions

The studys research questions define the validation goal of a case study. The studys research
questions should be clarified precisely (Yin, 2003).
For example, in this empirical study, the studys
research question is Does Web user interface
modeling for a Web-based medical SPL using
WUIML increase SMEs modeling efficiency
and effectiveness (thus decreasing work)? This
studys research question needs to be clarified
further because the notion of efficiency and effectiveness need further clarification. This studys
research question can be decomposed into a set
of propositions.

Study Propositions
Study propositions are derived from the studys
research questions (Yin, 2003) but are more specific than the studys research questions. Study
propositions quantify the quality variables (indirect metrics) in a studys research question into
directly measurable quantitative metrics (direct
metrics or indicators). For example, in this multiple-case study, the studys research question is
decomposed into two propositions:
1.

2.

SMEs are able to correctly model a larger

number of required modeling items using
WUIML than standard UML in modeling
WUIs for a Web-based medical SPL in
approximately the same amount of personhours.
SMEs are able to correctly model larger
numbers of required modeling items via
reuse using WUIML than standard UML in
modeling WUIs for a Web-based medical
SPL in approximately the same amount of
person-hours.

In this multiple-case study, for each WUI, the

investigator has identified from the requirements
a set of modeling items, called required modeling
items that must be modeled by the SMEs. The
resulted models produce by SMEs are inspected
by the investigator. The investigator first checks
the models for required modeling items. Then,
based on his modeling experience, the investigator
decides the correctness of the required modeling
items found in the resulted models.

Units of Analysis
Units of analysis are materials such as documents
or other resources that the subject matter experts
(SMEs) use as inputs or materials to apply the
method or tools being validated. In this study,
the method under investigation is WUIML. The
units of analysis are the Web user interface requirements for a medical SPL.
In the first case study, the requirements for
the Pediatric Medical Profile Login WUI, the
Adolescent Medical Profile Login WUI, and the
Adult Medical Profile Login WUI are provided.
These WUIs are each from a different member
product (Pediatric Medical Management System,
Adolescent Medical Management System, and
Adult Medical Management System) of a medical product line (Medical Management Systems).
The WUIs to model are extracted from three
Web-based medical products under development
in BUSINEX Inc.
The WUIs for the medical Web software
products are based on actual medical forms from
health-care providers in the United States. For
the first case study, three member products of the
product line are considered for WUI modeling.
In particular, this case study requires the SMEs
to model the Medical Profile Login WUI and a
related activity across three member products of
the product line. This WUI is chosen because it
allows one to exercise the modeling of commonality and variability found in product lines.

777

Modeling Variant User Interfaces for Web-Based Software Product Lines

In the second case study, the requirements

for the Pediatric Exercise Record WUI, the
Adolescent Exercise Record WUI, and the Adult
Exercise Record WUI are provided. These WUIs
are each from the same member products in the
same product line as in the first case study. These
requirements are the units of analysis for the
second case study.
Note that within each case study, there are
two SMEs. One SME applies WUIML to the
requirements to generate results while the other
SME applies standard UML to the requirements
to generate results. The results generated by ap-

plying standard UML are used as the baseline

for analyzing the results generated by applying
WUIML.
In this research, the SMEs are well trained
in software engineering; they are professional
software engineers with experience ranging from
7 to 15 years. Their specialties are focused on
user interface development. They are representative users for the new WUIML. The investigator
provides training on WUIML to the SMEs before
they begin conducting the case studies. In each
case, all SMEs participate in the case study only
once. Not having the same SME to participate in

Table 1. Required modeling items for Pediatric Medical Profile Login WUI
Required modeling items for Pediatric Medical Profile Login WUI
Page title: Pediatric Medical Profile Login
Label 1: Profile ID:
A textbox for label 1.
Label 2: Password:
A textbox for label 2.
Label 3: Role:
A radio button.
The radio button must default to be checked.
A label for the radio button: Parent/Guardian. (Only parent or legal guardian who are previously registered with the health provider
can login for the child.).
A submit button with name Login.
The Page title must placed on the top of the page.
The Profile ID and its related textbox must be immediately next to each other.
The Password and its related textbox must be immediately next to each other.
The Role and its related radioButton must not be immediately next to each other.
Profile ID must be placed on top of the Password and its related textbox.
The Role and the radio button must be on a line that is on top of the Login button.
The Login button must be placed at the lower left hand side of the page.
The activity diagram should include an action node: Pediatric Medical Profile Login.
The activity diagram should include an action node: Login Error.
The activity diagram should include an action node: Parent Welcome.
The activity diagram should include an action node: Customer Service.
The activity diagram should include a start node.
The activity diagram should include an end node.
The activity diagram should include a decision node: whether validation is successful.
The activity diagram should indicate the condition that Profile ID, Password, and Role values are available.

778

Modeling Variant User Interfaces for Web-Based Software Product Lines

more than one case is to prevent the introduction

of bias due to the familiarity of WUIML by SMEs
participating in multiple case studies. The SMEs
are to carry out the case study without help from
the investigator. The SMEs are also not given
information on the expected results of the case
study. This is to prevent bias by the SMEs in
performing the case studies.
Table 1 shows the required modeling items for
the Pediatric Medical Profile Login WUIs. The
result of whether these items are correctly modeled by SMEs, the number of correctly modeled
items over total number of required modeling
items by SMEs with or without product line reuse,
and the approximate person-hours spent will be
used to support/reject the study propositions 1

and 2. There are specific required modeling items

for Adult Medical Profile Login WUI, Pediatric
Medical Profile Login WUI, Adult Exercise Record WUI, Adolescent Exercise Record WUI,
and Pediatric Exercise Record WUI; for brevity,
they are not shown here.

Linking Data to Propositions

Data analysis is done in the step of linking data
(results) to propositions. In a case study, data are
collected using formal case study worksheets. A
case study worksheet consists of specific concrete
questions associate with each one of the study
propositions. The answers to those questions are
measured to concrete criteria metrics that can

Figure 12. Case study worksheet for modeling Pediatric Medical Profile Login WUI using WUIML
Case Study Worksheet
Using WUIML to model the Pediatric Medical Profile Login WUI.
SME ID: _____________________
Please answer the following questions:
1. What tool did you use to perform the modeling? Pediatric_WUIML_modelingTool
-----------------------------------------------------------------------------------------------------------2. What is the filename of the resulted model(s)? Pediatric_WUIML_modelsFilename
---------------------------------------------------------------------------------------------------------3 Please list below the model elements (e.g. classes and activity nodes, etc.) in your
resulted models. For each element, first provide its name and then provide a brief
description on its purpose. List as many as you wish in additional pages (if needed).
Pediatric_WUIML_ required_modeling_ itemi (i = 1 to the total number of modeled
elements in the resulted models for the WUI.)
-----------------------------------------------------------------------------------------------------------4 Among the items listed in 3, which of the model elements are developed via reuse?
List them below.
Pediatric_WUIML_reusei (i = 1 to the total number of modeled elements in the resulted
models for the WUI.)
-----------------------------------------------------------------------------------------------------------5. How many person-hours did you spend in complete the modeling of this WUI?
Pediatric_WUIML_personHours
------------------------------------------------------------------------------------------------------------

779

Modeling Variant User Interfaces for Web-Based Software Product Lines

be analyzed to support (or reject) the propositions. Thus, the most concrete criteria metrics
are measured terms found within the questions
on the case study worksheets. Figure 12 shows
a case study worksheet used in the first case
study for modeling the Pediatric Medical Profile
Login WUI. In Figure 12, each concrete criteria
metrics in questions that link to propositions is
identified by a name formed by three sections.
For example, Pediatric_WUIML_personHours

is the metric about person-hours spent in using

WUIML to model the Pediatric Medical Profile
Login WUI. The same format is used in naming
other metrics found in the questions in other case
study worksheets in the case studies.
Table 2 summarizes the evidence collected
through the case study worksheets in the case
study about Medical Profile Login WUIs.
In each case study, WUIML and standard
UML are applied to the units of analysis (that is

Table 2. Evidence collection through the case study worksheets

Questions (Units)

Evidence captured

Propositions to
support/reject

Pediatric_WUIML_modelingTool

The modeling tool used to perform the modeling

in WUIML for the Pediatric Medical Profile
Login WUI.

Proposition 1 and
Proposition 2

Pediatric_WUIML_modelsFilename

The name of the software copy of the resulted

models in WUIML for the Pediatric Medical
Profile Login WUI.

Proposition 1 and
Proposition 2

Pediatric_WUIML_required_modeling_itemi

The correctly modeled required modeling items

in WUIML for the Pediatric Medical Profile
Login WUI.

Proposition 1

Pediatric_WUIML_reusei

The correctly modeled required modeling items

via reuse in WUIML for the Pediatric Medical
Profile Login WUI.

Proposition 2

Pediatric_WUIML_personHours

The person-hours spent to model the Pediatric

Metrical Profile Login WUI using WUIML.

Proposition 1 and
Proposition 2

Pediatric_UML_modelingTool

The modeling tool used to perform the modeling

in UML for the Pediatric Medical Profile Login
WUI.

Proposition 1 and
Proposition 2

Pediatric_UML_modelsFilename

The name of the software copy of the resulted

models in UML for the Pediatric Medical Profile
Login WUI.

Proposition 1 and
Proposition 2

Pediatric_UML_required_modeling_itemi

The correctly modeled required modeling items

in UML for the Pediatric Medical Profile Login
WUI.

Proposition 1

Pediatric_UML_reusei

The correctly modeled required modeling items

via reuse in UML for the Pediatric Medical
Profile Login WUI.

Proposition 2

Pediatric_UML_personHours

The person-hours spent to model the Pediatric

Metrical Profile Login WUI using UML.

Proposition 1 and
Proposition 2

Adolescent_WUIML_modelingTool

The modeling tool used to perform the modeling

in WUIML for the Adolescent Medical Profile
Login WUI.

Proposition 1 and
Proposition 2

Adolescent_WUIML_modelsFilename

The name of the software copy of the resulted

models in WUIML for the Adolescent Medical
Profile Login WUI.

Proposition 1 and
Proposition 2

continued on following page

780

Modeling Variant User Interfaces for Web-Based Software Product Lines

Table 2. (continued)
Adolescent_WUIML_required_modeling_itemi

The correctly modeled required modeling items

in WUIML for the Adolescent Medical Profile
Login WUI.

Proposition 1

Adolescent _WUIML_reusei

The correctly modeled required modeling items

via reuse in WUIML for the Adolescent Medical
Profile Login WUI.

Proposition 2

Adolescent _WUIML_personHours

The person-hours spent to model the Adolescent

Metrical Profile Login WUI using WUIML.

Proposition 1 and
Proposition 2

Adolescent _UML_modelingTool

The modeling tool used to perform the modeling

in UML for the Adolescent Medical Profile
Login WUI.

Proposition 1 and
Proposition 2

Adolescent _UML_modelsFilename

The name of the software copy of the resulted

models in UML for the Adolescent Medical
Profile Login WUI.

Proposition 1 and
Proposition 2

Adolescent_WUIML_required_modeling_itemi

The correctly modeled required modeling items

in UML for the Adolescent Medical Profile
Login WUI.

Proposition 1

Adolescent _UML_reusei

The correctly modeled required modeling items

via reuse in UML for the Adolescent Medical
Profile Login WUI.

Proposition 2

Adolescent _UML_personHours

The person-hours spent to model the Adolescent

Metrical Profile Login WUI using UML.

Proposition 1 and
Proposition 2

Adult_WUIML_modelingTool

The modeling tool used to perform the modeling

in WUIML for the Adult Medical Profile Login
WUI.

Proposition 1 and
Proposition 2

Adult_WUIML_modelsFilename

The name of the software copy of the resulted

models in WUIML for the Adult Medical Profile
Login WUI.

Proposition 1 and
Proposition 2

Adult _WUIML_required_modeling_itemi

The correctly modeled required modeling items

in WUIML for the Adult Medical Profile Login
WUI.

Proposition 1

Adult_WUIML_reusei

The correctly modeled required modeling items

via reuse in WUIML for the Adult Medical
Profile Login WUI.

Proposition 2

Adult _WUIML_personHours

The person-hours spent to model the Adult

Metrical Profile Login WUI using WUIML.

Proposition 1 and
Proposition 2

Adult _UML_modelingTool

The modeling tool used to perform the modeling

in UML for the Adult Medical Profile Login
WUI.

Proposition 1 and
Proposition 2

Adult _UML_modelsFilename

The name of the software copy of the resulted

models in UML for the Adult Medical Profile
Login WUI.

Proposition 1 and
Proposition 2

Adult _UML_required_modeling_itemi

The correctly modeled required modeling items

in UML for the Adult Medical Profile Login
WUI.

Proposition 1

Adult _UML_reusei

The correctly modeled required modeling items

via reuse in UML for the Adolescent Medical
Profile Login WUI.

Proposition 2

Adult _UML_personHours

The person-hours spent to model the Adult

Metrical Profile Login WUI using UML.

Proposition 1 and
Proposition 2

781

Modeling Variant User Interfaces for Web-Based Software Product Lines

the Web user interface requirements of a Webbased medical SPL) respectively. The results,
that is, the resulting models and the completed
use case worksheets, are analyzed to find out the
following:
D1. How many of the required modeling items
are correctly modeled when the modeling
was done in WUIML?
D2. How many of the required modeling items
are correctly modeled when the modeling
was done in standard UML?
D3. How many of the required modeling items
are correctly modeled via reuse when the
modeling was done in WUIML? To model
via reuse is to create new models by re-using previously created models or model
elements. For example, suppose one previously created a model that includes a class
representing fruit. Now one can reuse the
fruit class to create a class that represents a
specific fruit, such as an apple, by extending the fruit class. Both standard UML and
WUIML allow SMEs to model via reuse.
D4. How many of the required modeling items
are correctly modeled via reuse when the
modeling was done in standard UML?
D5. How many person-hours spent to generate
the WUIML models?
D6. How many person-hours spent to generate
the standard UML models?
D7. The total number of required modeling items.
D1, D2, D5, D6, and D7 link to proposition
1. D3, D4, D5, D6, and D7 link to proposition 2.

Criteria for Interpreting a Case Studys

Findings
In this empirical study, the criteria for interpreting
a case studys findings correspond to the metric
and measures used in evaluating the results of applying WUIML and standard UML respectively
to the units of analysis. A measure provides a

782

quantitative indication of the extent, amount,

dimension, capacity, or size of some attribute of
a product or process (Pressman, 2005). A metric
is an important, directly-measurable attribute of a
software product, a software service, a software
process or a software resource. Direct metrics
may be either used as indicators (predictors), or
of other more valuable outcomes that are indirect
metrics. For example, the indirect metrics in this
case study are efficiency and effectiveness. The
direct metrics are:
m1. Number of the required modeling items that
are correctly modeled when the modeling
was done in WUIML;
m2. Number of the required modeling items that
are correctly modeled when the modeling
was done in standard UML;
m3. Number of the required modeling items that
are correctly modeled via reuse when the
modeling was done in WUIML;
m4. Number of the required modeling items that
are correctly modeled via reuse when the
modeling was done in standard UML;
m5. Person-hours spent to generate the WUIML
models;
m6. Person-hours spent to generate the standard
UML models;
m7. The total number of required modeling
items.

The Case Study Procedure for Both

Case Studies
0.

1.
2.

3.

For those SMEs that need to perform the

modeling in WUIML, teach them WUIML.
Identify the required modeling items from
the software requirements.
Provide SMEs who are to conduct the
modeling using WUIML with the software
requirements.
Provide SMEs who are to conduct the modeling using standard UML with the software
requirements.

Modeling Variant User Interfaces for Web-Based Software Product Lines

4.

5.

6.

7.

8.

9.

Collect the results (WUIML models and

the data from the completed case study
worksheets) generated from step 2.
Collect the results (standard UML models
and the data from the completed case study
worksheets) generated from step 3.
Perform data analysis on the results and data
collected from step 4 to find out the following: (a) number of the correctly modeled
required modeling items when the modeling
was done in WUIML (the measure of m1);
(b) number of the correctly modeled required
modeling items via reuse when the modeling
was done in WUIML (the measure of m3);
and (c) person-hours spent to generate the
WUIML models (the measure of m5).
Perform data analysis on results and data
collected from step 5 to find out the following: (a) number of the correctly modeled
required modeling items when the modeling
was done in standard UML (the measure of
m2); (b) number of the correctly modeled
required modeling items via reuse when the
modeling was done in standard UML (the
measure of m4); and (c) person-hours spent
to generate the standard UML models (the
measure of m6).
Evaluate the outcome from step 6 to determine whether or not the proposition 1 is
supported or rejected.
Evaluate the outcome from step 7 to determine whether or not the proposition 2 is
supported or rejected.

Results and Analysis

WUIML is developed to improve SPL software
engineering paradigm in modeling user interfaces for Web-based SPLs. The improvement
provided by WUIML should decrease the work
needed by the SPL engineers in the WUI modeling perspective.

In the first case study, each SME has to

model the WUIs for three product lines: Pediatric
Medical Profile Login, Adolescent Medical Profile
Login, and Adult Medical Profile Login. One SME
is asked to model each WUI using WUIML while
the other SME is asked to model the WUIs using
standard UML. Each WUI is from a different
member product (Pediatric Medical Management System, Adolescent Medical Management
System, and Adult Medical Management System)
of a medical product line (Medical Management
Systems).
Table 3 shows the results for modeling Pediatric
Medical Profile Login WUI using standard UML
and WUIML, respectively. The last row in Table
3 shows the ratio of the number of correctly modeled required modeling items to the total number
of required modeling items in standard UML and
WUIML, respectively.
Note that in each case study, there are two
SMEs. One SME is asked to model using standard
UML while the other SME is asked to model using
WUIML. In the following tables (Tables 4, 5, 6,
and 7), the numbers shown in the second column
from the left were derived from the models created by the SME which was asked to model using
standard UML. The numbers shown in the third
column from the left were derived from the models
created by the SME which was asked to model
using WUIML. The sample size is irrelevant in
case study research method because case study
research method is based on analytical generalization instead of statistical generalization (Yin,
2003). However, the results shown in these tables
must replicate (or be replicated by) the results
from the second case study in order to lead to an
analytical generalization.
Table 4 shows the ratio of the number of correctly modeled required modeling items to the
total number of required modeling items in modeling the three WUIs and the person-hours spent
on each approach. Notice that the person-hours
spent for modeling the Pediatric Medical Profile
Login WUI was more than the person-hours spent

783

Modeling Variant User Interfaces for Web-Based Software Product Lines

Table 3. Results for modeling the Pediatric Medical Profile Login WUI
Required modeling items for Pediatric Medical Profile Login WUI

Modeled correctly in
Standard UML

Modeled correctly
in WUIML

Page title: Pediatric Medical Profile Login

Yes

Yes

Label 1: Profile ID:

Yes

Yes

A textbox for label 1.

Yes

Yes

Label 2: Password:

Yes

Yes

A textbox for label 2.

Yes

Yes

Label 3: Role:

Yes

Yes

A radio button.

Yes

Yes

The radio button must default to be checked.

Yes

A label for the radio button: Parent/Guardian. (Only parent or legal

guardian who are previously registered with the health provider can login
for the child.).

Yes

Yes

A submit button with name Login.

Yes

Yes

The Page title must placed on the top of the page.

Yes

The Profile ID and its related textbox must be immediately next to each
other.

Yes

The Password and its related textbox must be immediately next to each
other.

Yes

The Role and its related radioButton must not be immediately next to
each other.

Yes

Profile ID must be placed on top of the Password and its related textbox.

Yes

The Role and the radio button must be on a line that is on top of the Login
button.

Yes

The Login button must be placed at the lower left hand side of the page.

Yes

The activity diagram should include an action node: Pediatric Medical

Profile Login.

Yes

Yes

The activity diagram should include an action node: Login Error.

Yes

Yes

The activity diagram should include an action node: Parent Welcome.

Yes

Yes

The activity diagram should include an action node: Customer Service.

Yes

Yes

The activity diagram should include a start node.

Yes

Yes

The activity diagram should include an end node.

Yes

Yes

The activity diagram should include a decision node: whether validation

is successful.

Yes

Yes

The activity diagram should indicate the condition that Profile ID,
Password, and Role values are available.

Yes

Yes

18/25

24/25

for Adolescent Medical Profile Login WUI and

Adult Medical Profile Login WUI respectively
in WUIML approach. To model the WUIs for a
member product (such as the Pediatric Medical

784

Profile Login WUI), one must first develop the

WUI models for the SPL.
Table 5 shows the number of correctly modeled required modeling items per person-hours

Modeling Variant User Interfaces for Web-Based Software Product Lines

Table 4. Ratio of correctly modeled required modeling items to the total number of required modeling
items
WUI

Standard UML (Person-hours)

WUIML (Person-hours)

Pediatric

18/25 (8)

24/25 (7)

Adolescent

23/30 (10)

29/30 (4)

Adult

18/25 (7)

24/25 (2)

Table 5. Number of correctly modeled required modeling items

Standard UML model

WUIML model

Pediatric

2.25

3.4

Adolescent

2.3

7.25

Adult

2.57

12

Average

2.37

9.08

Table 6. Number of correctly modeled required modeling items via reuse

WUI

Standard UML (Person-hours)

WUIML (Person-hours)

Pediatric

0/10 (8)

10/11 (3+4)

Adolescent

8/10 (10)

10/11 (4)

Adult

8/10 (7)

10/11 (2)

for the models in UML and WUIML for the three

member products WUIs respectively.
Table 6 shows the ratio of the number of correctly modeled required modeling items via reuse
(or product line reuse when WUIML is used) to
the total number of reusable required modeling
items in modeling the three WUIs and the personhours spent on each approach.
Table 7 shows the number of correctly modeled required modeling items per person-hours
via reuse for the models in UML and WUIML
respectively.
The data in Table 3 show that 18 out of 25
required modeling items were correctly modeled
for the Pediatric Medical Profile Login WUI using
standard UML. On the other hand, the data in
Table 3 show that 24 out of 25 required modeling items were modeled for the same Pediatric
Medical Profile Login WUI using WUIML. This
result in Table 3 indicates that an SME was able

to correctly model more required modeling items

using WUIML than using standard UML. Table
4 shows the correctly modeled required modeling
items and the approximate person-hours for modeling the Pediatric, Adolescent, and Adult Medical
Profile Login WUI respectively in standard UML
and WUIML. Table 5 shows the calculated value
of modeled items per person-hour based on the
data from Table 4. The data in Table 5 shows
that when using WUIML, an SME was able to
correctly model about four times more required
modeling items per person-hour (i.e., 9.08) than
that in standard UML (i.e., 2.37). Thus, this result indicates WUIML is more efficient than the
standard UML in modeling user interfaces for
Web-based SPL; therefore, this result supports
proposition 1.
Table 6 shows the number of correctly modeled
required modeling items via reuse (or product line
reuse) and its corresponding person-hours used.

785

Modeling Variant User Interfaces for Web-Based Software Product Lines

Table 7. Number of correctly modeled required modeling items via reuse (or product line reuse) per
person-hours
Standard UML model

WUIML model

Pediatric

1.43

Adolescent

0.8

2.5

Adult

1.14

Average

0.65

2.98

For example, using standard UML to model the

Pediatric Medical Profile Login WUI, a SME was
not able to reuse any model items out of the ten
reusable items in the WUI model. For approximate
eight person-hours, the SME has to model without
reuse. This is because the Pediatric Medical Profile Login WUI was the first WUI for an SME to
model. There is nothing to reuse yet. Notice that
the number of reusable required modeling items
is only a sub-set of the total number of required
modeling items in a WUI model because some
items are unique to a particular WUI and cannot
be reused.
On the other hand, using WUIML, across all
three WUIs (Pediatric Medical Profile Login,
Adolescent Medical Profile Login, and Adult
Medical Profile Login.), as shown in Table 6,
number of correctly modeled required modeling
items via reuse are consistent. This is because in
WUIML, an SME is to first develop the product
line WUI model. The WUI model for the SPL
encapsulated all model items for its member
product WUI models. Therefore, reusing the
items in the product line WUI model creates the
WUI model for a member product. As shown
in Figure 7, the average number of correctly
modeled required modeling items via reuse per
person-hours using WUIML (i.e., 2.98) is about
four times to that when using standard UML (i.e.,
0.65). This result indicates that WUIML enables
higher level of reuse than standard UML, thus
it indicates increased effectiveness when using
WUIML than using standard UML. Therefore,
this result supports proposition 2.

786

The results of the second case study are not

shown due to the limitation of the space in the
paper. The second case study was conducted by
another two SMEs. The units of analysis for the
second case study were the software requirements for the Pediatric Exercise Record WUI,
the Adolescent Exercise Record WUI, and the
Adult Exercise Record WUI. The results from
the second case study show that using WUIML
to model the Exercise Record WUIs for the
three member products of the SPL increases the
modelers efficiency and effectiveness in modeling. This result is similar to the result from the
first case study. The results of the two case studies literally replicate. Based on the results, the
analytical generalization that WUIML increases
modelers efficiency and effectiveness in model
WUIs for SPL is drawn in this study.
The generalization thus provides a positive
answer to the studys research question that Web
user interface modeling for a Web-based medical
SPL using WUIML increases SMEs modeling efficiency and effectiveness (thus decreases
work).

Conclusion and Future Work

In standard UML, there is no standard way to
model user interface and no standard way to
model variant features of user interfaces for a
SPL. WUIML defines elements to propose a
standard way to model user interfaces as well as

Modeling Variant User Interfaces for Web-Based Software Product Lines

variant user interface features of Web-based SPLs.

WUIML improves UML in terms of modeling of
user interfaces and use interfaces for SPL.
Case study research method has been applied
to investigate whether WUIML method increases
modelers efficiency and effectiveness on modeling a Web-based medical SPL when compared
to standard UML method. The results indicate
that modeling WUIs for SPL using WUIML is
more effective and efficient than using standard
UML.
There are many ways to model software using UML; a human is needed to make the best
judgment on the correctness of a UML model.
In this empirical study, the way to determine
whether a required modeling item is modeled
correctly or not relies on a humans modeling
experience. Since the same human is using his
same modeling experience to make the judgment
on all models, the bias due to human judgment
should be insignificant.
Since the SMEs are of similar backgrounds
and technical experience, the impact on modeling efficiency and effectiveness due to individual
SMEs capability should be insignificant. Since the
same requirements are given for both approaches
(WUIML and standard UML), the reuse opportunity base on application logics (according to the
requirements) should be the same for all SMEs.
Future work is to develop a tool to support
using WUIML to model WUIs for SPLs and to
extend WUIML beyond the Web platform.

References
Altheim, M., Boumphrey, F., McCarron, S.,
Schnitzenbaumer, S., & Wugofski, T. (Eds.).
(2001). Modularization of XHTML. World Wide
Web Consortium. Retrieved April 1, 2005, from
http://www.w3.org/TR/xhtml-modularization/
xhtml-modularization.html

Altheim, M., & McCarron, S. (Eds.). (2001).

XHTML 1.1 - Module-based XHTML. World Wide
Web Consortium. Retrieved April 1, 2005, from
http://www.w3.org/TR/xhtml
Anastasopoulos, M., & Gacek, C. (2001). Implementing product line variabilities. ACM SSR
01, Toronto, Canada (pp. 109-117). New York:
ACM Press.
Axelsson, J., Epperson, B., Ishikawa, M., McCarron, S., Navarro, A., & Pemberton, S. (Eds.).
(2003). XHTML2.0. W3C Working Draft 6 May
2003. World Wide Web Consortium. Retrieved
April 1, 2004, from http://www.w3.org/TR/2004/
WD-xhtml2-20040722
Behrens, H., Ritter, T., & Six, H.-W. (2003).
Systematic development of complex Web-based
user interfaces. Computer Science in Perspective,
2598, 22-38.
Blankenhorn, K., & Jeckle, M. (2004). A UML
profile for GUI layout. Net.ObjectDays, 110121.
Booch, G., Rumbaugh, J., & Jacobson, I. (1999).
The unified modeling language user guide. Reading, MA: Addison Wesley Longman.
Bos, B., Celik, T., Hickson, I., & Lie, H. W. (Eds.).
(2004). Cascading style sheets, level 2 revision 1
CSS 2.1 Specification. W3C Candidate Recommendation. Retrieved April 1, 2005, from http://
www.w3.org/TR/2004/CR-CSS21-20040225
Chastek, G., Donohoe, P., Kang, K. C., & Thiel, S.
H., A. (2001). Product line analysis: A practical
description (CMU/SEI-2001-TR-001). Pittsburgh,
PA: Software Engineering Institute, Carnegie
Mellon University.
Clements, P., & Northrop, L. (2002). Software
product lines: Practices and patterns (2nd ed.).
Boston: Addition-Wesley.
Cohen, S. G., Stanley, J. J. L., Peterson, A. S., &
Krut, Jr., R. W. (1992). Application of feature-

787

Modeling Variant User Interfaces for Web-Based Software Product Lines

oriented domain analysis to the Army movement

control domain (CMU/SEI-91-TR-28). Pittsburgh,
PA: Software Engineering Institute, Carnegie
Mellon University.

Rigorous Development Methods - Countering

or Integrating the eXtremists, Workshop of the
pUML-Group at the UML 2001. Gesellschaft fr
Informatik, Kllen Druck+Verlag.

Dubinko, M. (2003). XForms essentials. Sebastopol, CA: OReilly & Associates.

Jacobson, I., Christerson, M., Jonsson, P., &

Overgaard, G. (1992). Object-oriented software
engineering: A use case driven approach. Reading, MA: Addison-Wesley.

Dubinko, M., Leigh, L., Klotz, J., Merrick, R., &

Raman, T. V. (2002). XForms 1.0. Retrieved April
1, 2005, from http://www.w3.org/TR/xforms/.
Eckstein, R., Loy, M., & Wood, D. (1998). Java
swing. (1st. ed.). Sebastopol, CA: OReilly and
Associates, Inc.
Eriksson, H.-E., Penker, M., Lyons, B., & Fado,
D. (2004). UML 2 toolkit. Indianapolis, IN: Wiley
Publishing.
Gamma, E., Helm, R., Johnson, R., & Vlissides,
J. (1995). Design patterns (1st ed.). Reading, MA:
Addison-Wesley Professional.
Gomaa, H. (2004). Designing software product
lines with UML: From use cases to pattern-based
software architectures. Boston: Addison-Wesley
Professional.
Gomaa, H., & Gianturco, M. (2002). Domain
modeling for World Wide Web based software
product lines with UML. IEEE ICSR-7, LNCS,
2319, 78-92.
Gmez, J., & Cachero, C. (2003). OO-H method:
Extending UML to model Web interfaces. In
Information modeling for Internet applications
(pp. 144-173). Hershey, PA, USA: Idea Group
Publishing.
Griffiths, T., Barclay, P. J., McKirdy, J., Paton, N.
W., Gray, P. D., Kennedy, J. B., et al. (2001). Teallach: A model-based user interface development
environment for object databases. Interacting
with Computers, 14(1), 31-68.
Hennicker, R., & Koch, N. (2001). Modeling the
user interface of Web applications with UML.
Paper presented at the In Practical UML-Based

788

Jacobson, I., Griss, M., & Jonsson, P. (1997).

Software reuse: Architecture, process, and organization for business success. Reading, MA:
Addison-Wesley Longman.
Khare, R. (2000, March-April). Can XForm transform the Web? Transcending the Web as GUI, Part
II. IEEE Internet Computing, 103-106.
Kovacevic, S. (1998, June). UML and user interface modeling. Paper presented at the UML98,
Mulhouse, France.
Lee, S.-W. (2004). Proxy viewpoints model-based
requirements discovery. Unpublished doctoral
dissertation, George Mason University, Fairfax,
VA.
Lie, H. W., & Bos, B. (1999). Cascading style
sheets, level 1 (11 Jan 1999). Retrieved on June 28,
2003, from http://www.w3.org/TR/REC-CSS1
Lilly, S. (2000). How to avoid use-case pitfalls.
Retrieved on April 9, 2005 from http://www.
sdmagazine.com/print/
Lynch, C. L. (1999). A methodological framework
for interface requirements identification and
specification. Unpublished doctoral dissertation,
George Mason University, Fairfax, VA.
Marcus, A. (2002). Dare we define user-interface
design? Interactions, 9(5), 19-24.
Meyer, E. A. (2000). Cascading style sheets:
The definitive guide. Sebastopol, CA: OReilly
& Associates, Inc.

Modeling Variant User Interfaces for Web-Based Software Product Lines

Meyer, E. A. (2003). Eric Meyer on CSS: Mastering the language of Web design (2nd ed.).
Indianapolis, IN: New Riders Publishing.

Rosenberg, D., & Scott, K. (2001). Top ten use

case mistakes. Retrieved on April 1, 2005, from
http://www.sdmagazine.com/print/

Myers, B. A. (1989, January). User-interface

tools: Introduction and survey. IEEE Interface
Systems, 15-23.

Rumbaugh, J., Jacobson, I., & Booch, G. (2005).

The Unified Modeling Language reference
manual (2nd ed.). Boston: Addison Wesley.

Nunes, N. J. (2003). Representing user-interface

patterns in UML. Lecture Notes in Computer
Science, 2817, 142-151.

Sauers, M., & Wyke, R. A. (2001). XHTML essentials. New York: Wiley Computer Publishing.

OMG. (2002). Meta object facility (MOF) Version

1.4. Retrieved May 1, 2005, from http://www.omg.
org/technology/documents/formal/mof.htm
OMG. (2003a). UML 2.0 OCL Specification,
ptc/03-10-14. Object Management Group.
OMG. (2003b). Unified Modeling Language: Infrastructure, version 2.0 (3rd revised submission
to OMG RFP ad/00-09-01). Object Management
Group.
OMG. (2004). Unified Modeling Language: Superstructure, version 2.0 (ptc/04-10-02).
Phillips, C., Kemp, E., & Kek, S. M. (2001).
Extending UML use case modelling to support
graphical user interface design. In Proceedings
of the 13th Australian Conference on Software
Engineering (pp. 48-57). Washington, DC: IEEE
Computer Society.
Pickett, J., et al. (2000). The American Heritage
Dictionary of the English Language (4th ed.).
Boston: Houghton Mifflin.
Powell, T. (2003). HTML & XHTML: The complete
reference (4th ed.). McGraw-Hill/Osborne.
Pressman, R. S. (2005). 15 product metrics for software. In Software engineering: A practitioners
approach (6th ed.). Boston: McGraw-Hill.
Raggett, D., Hors, L. E., & Jacobs, I. (1999). HTML
4.01 specification. Retrieved May 1, 2005, from
http://www.w3.org/TR/REC-html40/

Schengili-Roberts, K. (2004). Core CSS: Cascading style sheets (2nd ed.). Upper Saddle River, NJ:
Prentice Hall PTR.
Schmitt, C. (2003). Designing CSS Web pages (2nd
ed.). Indianapolis, IN: New Riders Publishing.
Scogings, C., & Phillips, C. (2001). A method for
the early stages of interactive system design using
UML and Lean Cuisine+. In Proceedings Second
Australasian User Interface Conference, 2001,
AUIC 2001, Gold Coast, Queensland, Australia
(pp. 69 -76). IEEE Computer Society.
Scott, K. (2004). Fast track UML 2.0: UML 2.0
reference guide. APress.
SEI. (2005a). A framework for software product
line practice, version 4.2. Software Engineering
Institute. Retrieved on March 23, 2005, from http://
www.sei.cmu.edu/productlines/framework.html
SEI. (2005b). Software product lines. Retrieved
on March 23, 2005, from http://www.sei.cmu.
edu/productlines/
Shin, E. M. (2002). Evolution in multiple-view
models of software product families. Fairfax, VA:
George Mason University.
Silva, P. P. d. (2002). Object modelling of interactive systems: The UMLi approach. Unpublished
doctoral dissertation, University of Manchester,
UK.
Silva, P. P. d., & Paton, N. W. (2003, July/August). User interface modeling in UMLi. IEEE
Software, 62-69.

789

Modeling Variant User Interfaces for Web-Based Software Product Lines

Sommerville, I. (2001). User interface design.

In Software engineering (chap. 15). New York:
Addison-Wesley.

W3C.The Extensible Stylesheet Language (XSL).

Retrieved on May 3, 2003, from http://www.
w3.org/Style/XSL/

Szekely, P. A., Sukaviriya, P. N., Castells, P.,

Muthukumarasamy, J., & Salcher, E. (1996).
Declarative interface models for user interface
construction tools: The MASTERMIND approach. Paper presented at the in Engineering
for Human-Computer Interaction (pp. 120-150).
London: Chapman & Hall.

W3C. (2002). XHTML 1.0: The Extensible Hypertext Markup Language (2nd edition).

Tellis, W. (1997a). Application of a case study

methodology. The Qualitative Report, 3(3).
Retrieved May 3, 2003, from http://www.nova.
edu/sss/QR/QR3-3/tellis2.html
Tellis, W. (1997b). Introduction to case study. The
Qualitative Report, 3(2).

Webber, D. L. (2001). The variation point model

for software product lines. Fairfax, VA: George
Mason University.
Yin, R. K. (2003). Case study research: Design
and methods (3rd ed., Vol. 5). Thousand Oaks:
Sage Publications, Inc.
Ziadi, T., Hlout, L., & Jzquel, J.-M. (May
2004). Towards a UML profile for software product lines. Lecture Notes in Computer Science,
3014, 129-139.

This work was previously published in the International Journal of Information Technology and Web Engineering, Vol. 1, Issue
1, edited by G. Alkhatib and D. Rine, pp. 1-34, copyright 2006 by IGI Publishing, formerly known as Idea Group Publishing
(an imprint of IGI Global).

790

791

Chapter 2.25

A User-Centered Approach
to the Retrieval of Information
in an Adaptive Web Site
Cristina Gena
Universit di Torino, Italy
Liliana Ardissono
Universit di Torino, Italy

Abstract
This chapter describes the user-centered design
approach we adopted in the development and
evaluation of an adaptive Web site. The development of usable Web sites, offering easy and
efficient services to heterogeneous users, is a
hot topic and a challenging issue for adaptive
hypermedia and human-computer interaction.
User-centered design promises to facilitate this
task by guiding system designers in making decisions, which take the users needs in serious account. Within a recent project funded by the Italian
Public Administration, we developed a prototype
information system supporting the online search
of data about water resources. As the system was
targeted to different types of users, including
generic citizens and specialized technicians, we
adopted a user-centered approach to identify their

information needs and interaction requirements.

Moreover, we applied query analysis techniques
to identify further information needs and speed
up the data retrieval activity. In this chapter, we
describe the requirements analysis, the system
design, and its evaluation.

Introduction
The development of a Web-based information
system targeted to different types of users challenges the Web designer because heterogeneous
requirements, information needs, and operation
modes have to be considered. As pointed out by
Nielsen (1999) and Norman and Draper (1986),
the users mental model and expectations have
to be seriously taken into account to prevent
her/him from being frustrated and rejecting the

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

A User-Centered Approach to the Retrieval of Information in an Adaptive Web Site

services offered by a Web site. Indeed, this issue is particularly relevant to Web sites offering
task-oriented services, because most target users
utilize them out of their leisure time, if not at work.
Being under pressure, these users demand ease of
use as well as efficient support to the execution
of activities.
The positive aspect of a technical Web site is,
however, the fact that the users can be precisely
identified and modeled; moreover, their information needs, representing strong requirements,
can be elicited by means of a suitable domain
analysis. Therefore, utilities, such as data search
and retrieval, can be developed to comply with
different goals and backgrounds. Of course, users involvement and testing have to be carried
out also in this case because they support the
development of effective and usable services (see
Dumas & Redish, 1999; Keppel, 1991).
In our recent work, we faced these issues in
the development of ACQUA, a prototype Webbased information system for the Italian Public
Administration presenting information about
water resources (a demo is available at http://acqua.di.unito.it). During the system design phase,
we put in practice traditional usability principles
and adaptive hypermedia best practices and we
derived general guidelines for the development
of usable Web-based systems for technical users
(see Brusilovsky, 1996, 2001; Fink, Kobsa, & Nill,
1999; Maybury & Brusilovsky, 2002). The system
described in the rest of this chapter is targeted to
two main classes of users:

Generic users, such as the citizen, who want

to be informed about the general health state
of rivers, lakes, and underground waters.
Technical users, such as the public administration employees, who retrieve specific
pieces of information for analysis purposes.

In this chapter we describe the requirements

analysis, the design, and the evaluation of ACQUA,

792

focusing on the user-centered approach adopted

in the prototype design and development phases.
We involved domain experts and end users since
the beginning of our work in order to assess the
usefulness and suitability of the functionality
offered by the system, as well as of its user interface. For further information about the system,
see Gena and Ardissono (2004).
The rest of this chapter is organized as follows:
Section Background provides an overview of
the relevant user-centered design research. Section The ACQUA Project presents our work.
Specifically, Section Application Requirements
describes the interaction and user interface requirements that emerged during the design phase;
Section Adaptive Features presents the adaptive
features we developed for our system; Section
Association Rules describes the techniques
supporting the personalized information search;
Section Evaluation of ACQUA presents the
results of an evaluation we carried out to test the
system functionality with real users; and Section
Comparison with Other Solutions compares our
proposal with some related work. Finally, section
Future Trends discusses some open technical
issues and suggests how to address them, and
Section Conclusion concludes the chapter.

Background
Several researchers suggested to address usability issues by developing adaptive systems.
For instance, Benyon (1993) proposed adaptivity
as a solution, because a single interface cannot
be designed to meet the usability requirements
of all the groups of users of a system. However,
it is possible to prove that adaptivity enhances
the usability of a system only if it can be shown
that, without the adaptive capability, the system
performs less effectively. Benyon identifies five
interdependent activities to be considered when
designing an adaptive system:

A User-Centered Approach to the Retrieval of Information in an Adaptive Web Site

1.
2.

3.

4.

5.

Functional analysis, aimed at defining the

main functions of the system.
Data analysis, concerned with understanding and representing the meaning and structure of data in the application domain.
Task knowledge analysis, focused on the
cognitive characteristics required by the
system users, such as the users mental
model, cognitive load, and the required
search strategy.
User analysis, aimed at determining the
scope of the user population to which the
system is targeted. This analysis concerns
the identification of the user attributes that
are relevant for the application, such as
required intellectual capability, cognitive
processing ability, and similar. The target
population is analyzed and classified according to the aspects of the application derived
from the above-mentioned points.
Environment analysis is aimed at identifying the characteristics of the environment
in which the system is going to operate.

Notice that these phases are similar to the steps

followed during the requirements analysis phase
of a generic software system (Preece, Rogers,
Sharp, & Benyon, 1994). Benyon underlined the
fact that adaptive systems should benefit more
than other systems from a requirements analysis
before starting any kind of evaluation, because the
development of these systems has to take a high
number of features into account. The recognition
that an adaptive capability may be desirable leads
to an improved system analysis and design. As a
demonstration, he reported an example of an adaptive system development, wherein he prototyped
and evaluated the system with a number of users.
Several user characteristics were examined to
determine their effects on the interaction. Then,
further task knowledge and functional analysis
were carried out.
Also Oppermann (1994) proposed a usercentered perspective and suggested a design-

evaluation-redesign approach. He noticed that the

adaptive features can be considered as the main
part of a system and thus have to be evaluated
during every development phase. The problem
is circular:

A problem solvable by means of the adaptivity has to be identified.

The user characteristics related to the problem have to be selected.
Ways of inferring user characteristics from
interaction behavior have to be found.
Adaptation techniques offering the right
adaptive behavior have to be designed.

This process requires a bootstrapping method:

first some initial adaptive behavior is implemented, then tested with users, revised, and tested
again. The reason is that it is hard to decide which
particular adaptations should be associated to
specific user actions. Furthermore, the adaptations must be potentially useful to the user. The
necessity of an iterative process is due to the fact
that the real behavior of users in a given situation
is hard to foresee; therefore, some evidence can
be shown only by monitoring the users activity.
From the iterative evaluation point of view, the
design phases and their evaluation have to be
repeated until good results are reached.
Oppermanns iterative process is very similar
to the user-centered system design approach
originally phrased by Gould and Lewis (1983) and
extended by Norman and Draper (1986).
Dix, Finlay, Abowd, and Beale (1998) pointed
out that the iterative design is also a way to
overcome the inherent problems of incomplete
requirements specification, as only a subset of
the requirements for an interactive system can
be determined from the start of the project. The
iterative evaluation process requires empirical
knowledge about the users behavior from the first
development phases. In the case of an adaptive
system, prior knowledge about the real users, the
context of use, and domain experts facilitates the

793

A User-Centered Approach to the Retrieval of Information in an Adaptive Web Site

selection of the relevant data for the user model,

such as personal features, goals, plans, domain
knowledge, and context. Deep knowledge about
users also offers a broad view of the application
goals and prevents the system designer from
serious mistakes, especially when dealing with
innovative applications.
Petrelli, De Angeli, and Convertino (1999)
proposed the user-centered approach to user modeling as a way to move from designer questions
to guidelines by making the best use of empirical
data; they advocated incremental system design as
a way to satisfy large sets of users. They reported
that at the early stage of development of a mobile
device presenting contextual information to museum visitors, they decided to revise some of their
initial assumptions about the user model. Indeed,
they made this decision after having analyzed
the results of a questionnaire distributed to 250
visitors. For instance, they discarded the former
user modeling techniques based on stereotypes
(because the sociodemographic and personal data
taken in consideration did not characterize the
users behavior in a satisfactory way) in favor of
a socially oriented and context-aware perspective.
For instance, they noticed that people do not like
to visit museums on their own and prefer looking
at paintings to interacting with a device.
As discussed by Hk (2000), intelligent user
interfaces may violate many usability principles
developed for direct manipulation systems. The
main problem is that these systems may violate
many good principles, such as enabling the user
to control the system, making the system predictable (given a certain input, the system always
generates the same response), and making the
system transparent so that the user understands
at least partially how it works. In addition, most
adaptive interface developers are more concerned
with defining inference strategies than with interface design. For Hk, intelligent user interfaces
sometimes require a new way of addressing usability, different from the principles outlined for
direct-manipulation systems. Instead of measur-

794

ing factors such as task completion time, number

of errors, or number of revisited nodes, other
aspects have to be considered. For instance, if
the system should do information filtering, then
we must check whether subjects find the most
relevant information with the adaptive system and
not necessarily whether they find it fast. This is
not to say that the traditional measurements are
always wrongthis of course depends upon the
task that user and (adaptive) system should solve
together (Hk, 2002, p. 12).
Finally, Palmquist and Kim (2000) investigated the effects of (field independent and field
dependent) cognitive style and online database
search experience on WWW search performance.
They concluded that cognitive style significantly
influences the performance of novice searchers.
In contrast, experienced searchers display a common behavior: they usually do not get lost in Web
pages including many links, but they are able to
choose useful navigation strategies. Therefore,
Palmquist and Kim suggested that novice users
should benefit from Web pages that have a simple
design and few links providing information necessary to perform analytic search.

The ACQUA Project

Application Requirements
In 2003, the Water Resources Division (Direzione Risorse Idriche) of the Piedmont Region
and the University of Torino started a project
for the development of ACQUA, a Web-based
information system presenting data about water
resources derived from the monitoring activities
on the territory. The goal was to make information available on a Web site that describes the
Division and supports a search for data in real
time, in order to limit the distribution of information on a one-to-one basis via e-mail messages
and paper publications. The technicians of the
Division guided us in the system development

A User-Centered Approach to the Retrieval of Information in an Adaptive Web Site

by specifying application requirements and by

sharing with us a repository of e-mail messages
they exchanged with users asking for information
throughout the years. The repository provided us
with evidence about the users interested in water
resources data, the inspected information, and the
regularities in the search for data. Most questions
were posed by the following:

Employees of other Public Administrations,

such as technicians and researchers, who are
often interested in environmental impact
studies, construction feasibility studies, and
historical data.
Technicians, such as companies working at
the construction of bridges and houses.
Attorneys, who are typically interested in
the examination of data concerning specific
regions, for example, as a consequence of
an environmental disaster.
Farmers, who wish to monitor the biochemical state of their fields.
Students attending secondary school, university, and also doctoral programs. These
users collect information for the preparation
of reports concerning, for example, historical
changes in biological and chemical composition of waters, or the evolution of the
capacity and hydrometric levels of rivers,
and similar.

Following a user-centered approach, we developed the system by involving domain experts

and end users since the first design phases. After
a requirements analysis phase, we developed a
number of mock-ups, which we discussed and
redesigned after several focus group sessions with
the experts and the users involved in the project.
We decided to adopt a cooperative design approach
(Greenbaum & Kyng, 1991) in order to utilize the
experience of domain experts and technicians
in the design of an effective user interface. We
based the development of our first prototype on
the collected feedback. As the ACQUA system

is devoted to the Public Administration, we had

to satisfy usability and predictability requirements that imposed the design of a simple user
interface. Specifically, our interlocutors suggested
the following:

The interface should be usable and intuitive

in order to satisfy user needs and expectations. This first requirement should be
followed in every interface design project;
however, Public Administrations have the
mandatory goal of satisfying all the citizens,
thus usability is also intended as a service
for the collectivity.
The system behavior should be highly
predictable (Dix et al., 1998) to support
first-time visitors in their search task, but
also to avoid frustrating professional users
who would regularly use it at work. Notice
that the predictability requirement has some
subtle aspects: for instance, not only the
user should foresee what is going to happen
next, but also what should not be expected
from the service. This is very important to
prevent the user from starting the exploration of paths that will not provide her/him
with the information, or the functionality
(s)he is looking for.
The system should provide the user with data
that can be analyzed without preprocessing.
Therefore, search results should be presented
in machine-processable formats, in addition
to the pictorial ones suitable for a generalpurpose presentation in Web pages.
For the sake of accessibility, the pages of the
user interface should be optimized for standard browsers, without the need of special
equipments or software environments.

In order to maximize the usefulness of the

information that can be retrieved from the Web
site, we decided to make the system generate the
search results in formats, such as MS Excel tables
and textual (TXT) files, directly supporting the

795

A User-Centered Approach to the Retrieval of Information in an Adaptive Web Site

data analysis and interpretation at the user side.

We also tried to address efficiency in the retrieval
of information by reconsidering the design of the
general Web site to be presented. We wanted to
offer the right information the user is looking
for; thus, we decided to show the main search
functions in the home page of the Web site, and
to move textual information, such as the pages
describing the Public Administration divisions,
in secondary pages, which can be reached by
following hypertextual links.
Moreover, having analyzed the data about the
users interested in water resources, we identified
two main targets to which the system should
adapt. For shortness, we denote these categories
as novices and experts.

Novice users, such as students and generic

citizens, visit the Web site on an occasional

basis and are not familiar with the content

presented by the information system.
Expert users, such as technicians, farmers,
and the personnel of other Public Administrations, frequently visit the site and are
familiar with the domain-specific information provided by the system.

In order to take the interaction requirements

of these users into account, we defined two search
functions:
i.

ii.

The simple search is a geographical search

modality and guides the user step by step
in the retrieval of information;
The advanced search offers forms where
the expert user may compose the queries in
single step. Figure 1 shows the user interface of the ACQUA prototype supporting

Figure 1. Searching quantitative data (continuous hydrometric and chemical-physical parameters) about
Po River in the Torino-Murazzi observation point

796

A User-Centered Approach to the Retrieval of Information in an Adaptive Web Site

the advanced search; the menus enable the

user to specify the river (Scegli il corso
dacqua), observation point (Scegli il punto
di monitoraggio), start date (Data Inizio),
and end date (Data Fine). Moreover, the
user interface enables the user to select the
hydrometric and chemical-physical parameters to be inspected.
Thus, novice users may search for information
in a friendly modality and the eligible choices are
restricted and presented along the path, while expert users benefit from a faster search function.
As a matter of fact, the information about water
resources exploited by the system is unavoidably
incomplete. For instance, some data are collected
by automatic stations, which have been set up at
different times over the years and sometimes are
out of order. Moreover, unfortunately, data collected in manual observation points have been
stored in unstructured formats and the historical
series has been reconstructed only for the very
recent past.
For the sake of predictability, the simple and
advanced search functions prevent the user from
composing any queries that are incorrect, or are
aimed at searching for unavailable data. The idea
is that, in both cases, the system should only
present the choices leading to available results.
For instance, as shown in Figure 1, the labels of
the parameters, which are not available for the Po
River, are shaded and cannot be selected by the
user to define a query.

Adaptive Features
The information about water resources concerns
rivers, lakes, and underground waters and includes
the following:

Descriptive data about resources and observation points: for example, maps of the
points, charts representing environmental
changes, pictures, documents, publications,

and descriptions of the monitoring stations.

For instance, Figure 2 (Caratteristiche della
stazione di monitoraggio TORINO [Features of the Torino monitoring station])
shows the coordinates and other information about the observation point on Po River
located in Torino, Parco Michelotti.
Measurement parameters concerning physical dimensions and other features, which
characterize the environmental state of the
resources. These parameters are grouped in
two main classes:

Qualitative parameters, which are
periodically measured: technicians
visit the observation points, collect
data, and take samples for laboratory
tests.

Quantitative parameters, which are
monitored by automatic stations.

These stations carry out the measurements

on a daily basis.
The ACQUA Web site is organized in four main
sections, respectively devoted to the presentation
of qualitative and quantitative information about
rivers, information about lakes, and information
about underground waters. The system enables
the user to retrieve data about water resources by
performing a simple or advanced search in all the
sections of the site. Therefore, a large amount of
heterogeneous data is accessible, ranging from
biological and chemical data to capacity measurement and hydrometric levels (for details, see Gena
& Ardissono, 2004).
We noticed that, by performing queries aimed
at selecting a large number of data items, belonging to different categories, the results returned
by the system were complex and hard to present
in an intuitive results table. However, as shown
by the repository of user requests we analyzed,
users often need to combine heterogeneous data
to accomplish their goals. For example, in construction feasibility studies, users are interested
in qualitative and quantitative parameters of

797

A User-Centered Approach to the Retrieval of Information in an Adaptive Web Site

Figure 2. Portion of the page describing the Torino observation point on Po River

rivers and underground waters, considering the

historical series. In order to keep the user interface
simple and to guarantee that the presented results
are not confusing, we decided to limit the users
freedom in composing the queries: to retrieve very
heterogeneous types of information, the user must
define more than one search query. For example,
as shown in Figure 1, the ACQUA query interface
enables the user to choose from different rivers,
observation points, years, and data types. Other
categories, such as qualitative and quantitative
data about rivers, lakes, and underground waters
are treated as separate sections of the Web site
and have their own query functions.
Unfortunately, although this approach enforces
the clarity of the results, it makes the search for

798

multiple types of information a lengthy task.

Therefore, a compromise between clarity and
efficiency must be found. In order to address this
issue, we extended the system with an intelligent
search component, which complements the users
explicit queries with follow-up queries (Moore
& Mittal, 1996) frequently occurring together
in navigation paths. When possible, the system
anticipates the users queries and makes the extended search results available as personalized
suggestions that can be downloaded on demand.
If the user is interested in the recommended
information, (s)he can retrieve it by clicking on
the adaptive suggestion links, without performing any further queries. At the same time, the
system retrieves the extended results only after

A User-Centered Approach to the Retrieval of Information in an Adaptive Web Site

the user clicks on a suggestion link in order to

avoid precaching possibly useless data.
For instance, Figure 3 shows the recommendations generated by the system in the lower portion
of the page (Ti consigliamo anche i valori dei
parametri chimici e microbiologici [We also
suggest results about chemical and microbiological parameters]).
During different interaction sections, the same
user may be interested in rather different types
of information; therefore, we decided to base the
systems recommendations on the analysis of her/
his navigation behavior, leaving the management
of a long-term user model apart. One immediate
advantage is the fact that the user can interact with
the system in an anonymous way, without signing up for the service. The follow-up queries are
generated as follows: the search queries performed
by the user while (s)he browses the Web site are
monitored and association rules which suggest
other, strictly related queries are applied.
Each association rule has a condition part
specifying constraints on the previous navigation
behavior, and an action part defining a relevant fol-

low-up query to be performed in order to retrieve

complementary information. The rules we defined
in our current prototype are mutually exclusive and
they are selected and fired by applying a very tiny
and efficient inference engine. This engine would
not be suitable to manage a large set of conflicting rules: a general-purpose rule-based engine
should be employed to that purpose. However,
we prefer to maintain a simple set of adaptation
rules, and to avoid embedding complex rule-based
engines in order to keep the adaptive features as
lightweight as possible. In fact, the management
of the interaction is subject to a relevant overload
due to the generation of results in multiple formats,
which is a main requirement for the information
system. In this situation, minimalist but efficient
adaptation to the user is strongly preferred to
flexible but complex one.

Association Rules
In order to define the association rules to be applied for anticipating the users information needs,
we analyzed a repository of requests, which real

Figure 3. Annotated link for the suggested information and descriptions of the monitoring stations

799

A User-Centered Approach to the Retrieval of Information in an Adaptive Web Site

users posed to the Water Resources Division

over the years; the requests consisted of e-mail
messages and fax documents. As we noticed
that different kinds of information frequently
occurred together in these requests, we decided
to analyze the frequency of co-occurrence in
order to identify the regularities. Specifically,
we analyzed 97 requests and we selected a set of
features describing the requests in a systematic
way. These features concerned rather different
aspects of the requests; thus, for clarity purposes,
we grouped them in subcategories. In the following, we report the subcategories we defined and
for each one we list some sample features:

Kind of request: for example, environmental

impact study, construction feasibility studies, and lawyers studies.
Request features: for example, information about one or more rivers, about lakes
or underground waters, about one or more
observation points on a river or lake.
Kind of data: for example, qualitative or
quantitative parameters, biological and
chemical data, hydrometric level, average
daily capacity.
Data features: for example, raw data, or
elaborated data such as medium, maximum,
and minimum values during a time interval.
User features: for example, research center,
Public Administration, technicians, farmers,
and attorneys.

We computed the frequency with which the

features co-occurred in the requests: if the frequency exceeded a given threshold, the set of
involved features became a possible candidate
for an association rule. Then we compared the
extracted associations with their original requests
in order to validate our findings with factual
knowledge, and finally we asked the technicians
of the Water Resources Division if our conclusions
were correct. After this last check, we selected

800

the correct associations and we encoded the rules

in the system.
For instance, a rule suggests to retrieve qualitative parameters about a water resource if the user
has asked for quantitative historical data for more
than one observation point on that resource, supposing that (s)he is looking for information for a
construction feasibility study. Another rule suggests retrieving the environmental state indexes
of a resource if the user has requested biological
and chemical data, under the hypothesis that (s)he
is involved in an environmental impact study.

Evaluation of ACQUA
We first evaluated the ACQUA prototype in a
usability test by involving external users who
were not cooperating at the project (see Dumas
& Redish, 1999, for methodological details). The
evaluation highlighted some usability problems
concerning the presentation of basic information,
such as the choice between simple and advanced
search and the background color of the menus.
After having solved those problems, we tested
the final prototype with real end users representative of the users the Web site is devoted to. In
particular, we involved technicians working at
the Water Resources Division in different fields
(rivers, lakes, underground rivers, etc.) and not
collaborating to the design of the project. We
carried out both an experimental evaluation and
a qualitative session to assess the suitability of
the adaptive features offered the system.
Subjects. We evaluated 10 potential users of
the ACQUA system, four females and six males,
aged 3050. All the users worked in the water
resource area and none of them was involved in
the project.
Procedure. The subjects were split up in two
groups (five subjects each) and randomly assigned
to one of the two groups. The experimental group
had to solve some tasks using the adaptive Web
site, which applies the association rules described
in Section Adaptive Features to compute the

A User-Centered Approach to the Retrieval of Information in an Adaptive Web Site

results of follow-up queries related to the users

explicit queries. Instead, the control group had
to solve the tasks without adaptation.
Experimental tasks. Every subject had to solve
seven tasks, each one representing a real task the
user can perform in the Web site. As suggested
by our correlation study, the tasks were strictly
correlated and could be grouped in three search
activities the user often performs together. The
first activity conveyed the whole information
useful to an environmental impact study. The
second one supported construction feasibility
studies. The third activity supported lawyers
studies and activities.

In the control group, the users had to submit

a new query for every task, in order to obtain
the requested results. The new queries were
submitted by filling in the query specification forms (see, e.g., Figure 1).
In the experimental group, the users could
obtain the extra information related to the
next task to be performed by clicking on
an adaptive suggestion link that supports
the immediate retrieval of the suggested
information (see, e.g., Figure 3).

Experimental design. Single-factor (the adaptivity) between-subjects design.

Measures. The subjects navigation behavior
was recorded by using Camtasia Studio. We
measured the task completion time and then the
subjects satisfaction, by means of a post-task
walk-through.
Hypothesis. We hypothesized that the users
working in the experimental group could obtain
better performance results than those of the
control group.
Results. The ANOVA (analysis of variance)
showed that the subjects of the experimental
group achieved the best performance results.
In addition, we calculated the effect size (treatment magnitude) and the power (sensitivity) as
suggested in Chin (2001). The effect size (2)

measures the strength, or the magnitude, of the

treatment effects in an experiment. In behavioral
sciences, small, medium, and large effects of 2
are 0.01, 0.06, and >0.15, respectively. The power
of an experiment (n) is the ability to recognize
treatment effects and the power can be used for
estimating the sample size. In social science, the
accepted value of the power is equal to 0.80, which
means that the 80% of repeated experiments will
give the same results. In the following, we show
a summary of the results:
Task 2.

ANOVA: F(1.8) = 12.45 p<0.01;
2=0.53; n=3.49
Task 3.

ANOVA: F(1.8) = 12.12 p<0.01;
2=0.53; n=3.60
Task 5.

ANOVA: F(1.8) = 14.16
2=0.57; n=3.04

p<0.01;

Task 7.

ANOVA: F(1.8) = 9.23
2=0.45; n=4.86

p<0.05;

It should be noticed that all the results are significant and have a large estimate of the magnitude
of the treatment effect. In addition, by exploiting
a power of 0.80 and the corresponding 2 for each
task we could determine the requested sampled
size, which fits our sample size (n=5) (for details
about statistics, see Keppel, 1991).
Post-task walk-through. During any post-task
walk-through, test subjects are asked to think about
the event and comment on their actions. Thus,
after each test we talked to the subjects to collect
their impression and to discuss their performance
and the problems encountered during the test. In
this session, we also aimed at retrieving useful
feedback for a qualitative evaluation of the site.
In fact, although our experimental evaluation

801

A User-Centered Approach to the Retrieval of Information in an Adaptive Web Site

reported significant results supporting our hypothesis, the actual user behavior could be different. As recently pointed out by Nielsen (2004),
statistical analyses are often false, misleading,
and narrow; in contrast, insights and qualitative
studies are not affected by these problems because
they strictly rely to the users observed behavior
and reactions.
In most cases, the interviewed users were satisfied with the site. Most of them encountered some
problems in the execution of the starting query of
task 2, thus we modified the interface form.

All the users of the experimental group followed the adaptive suggestion link provided
by the system but they did not realize that
it represented a personalization feature.
When we explained the adaptations, they
noticed the particularity of the suggestion
(We also recommend you ...). Anyway,
they were attracted from the suggestions and
they appreciated the possibility of skipping
the execution of a new query. The adaptive
suggestions were considered visible and not
intrusive.
The users of the control group reported
similar considerations when we described
the adaptive features offered by the Web site.
Even if they did not receive any suggestions
during the execution of tasks, they explored
the result pages in order to find a shortcut to
proceed in the task execution. After having
followed some links, they went back to the
previous query page or to the home page
by clicking on the Back button of the
browser.

Both groups displayed a common behavior

pattern: the users explored a results page before
starting a new search. Nevertheless, their behavior
could be influenced by the test condition, because
tested users tend to pay a lot of attention to their
own actions and to the page design.

802

We conclude by admitting that although the

test subjects were satisfied with the adaptation
features, only the real system usage can demonstrate our hypothesis. However, both quantitative
and qualitative test results are encouraging and
we think that the adaptations are correctly placed.
After this test, we presented the adaptive version
of the Web site to the technicians of the Water
Resources Division collaborating on the project.
They confirmed the correctness of association
rules we defined and they decided to replace the
non-adaptive version of the prototype system with
the adaptive one.

Comparison with Other Solutions

The ACQUA system has a plain user interface,
designed to meet simplicity, usability, and predictability requirements, but it offers advanced
interactive features enabling the user to create a
personal view of the information space. Two search
features, targeted to novice and expert users, are
available, and the search results are presented in
both pictorial and machine-readable formats in
order to support direct data manipulation at the
user side. Moreover, the system analyzes the users
queries to identify her/his information needs, and
it employs association rules to propose follow-up
queries complementing the search results with
strictly related information. The follow-up queries
are applied on demand; thus, the user can ignore
them if (s)he is not interested in the additional
data, and the system does not need to retrieve
any uninteresting information.
The advanced search features we presented
differ from the related work in various aspects.
On the one hand, the inferences performed by
our system are simpler than the probabilistic
ones applied in other automated assistants, such
as Lumire (Horvitz, Breese, Heckerman, Hovel,
& Rommelse, 1998) and ACE (Bunt & Conati,
2003), which exploit Bayesian networks to capture the dependencies among the user actions.

A User-Centered Approach to the Retrieval of Information in an Adaptive Web Site

The point is that the user interacting with the

ACQUA system does not carry out a complex task
requiring a problem-solving activity. Therefore,
lightweight rules associating contextually related
search queries are sufficient to predict the implicit
information needs and to complement the search
for information accordingly. Our approach also
differs from the follow-up question answering
techniques proposed by Moore and Mittal (1996):
in order to efficiently manage the query selection
process, our follow-up queries are precompiled in
a set of association rules, instead of being generated by a planner.
On the other hand, we apply query analysis
techniques to identify regularities in search patterns. This differs from the typical inferences
carried out in recommender systems, which reason
about the features of the selected items to identify
the users priorities (see, e.g., Billsus & Pazzani,
1999), or about the regularities in the selection of
individual items (see, e.g., the work by Cotter &
Smyth, 2000; GroupLens, 2002).
Liu, Yu, and Meng (2002) propose other query
analysis strategies for personalized Web search.
However, instead of personalizing the proposed
results, their system supplies a small set of categories as a context for each query. The system
combines the users search history with a general
user profile automatically extracted from a category hierarchy to offer a personalized context
for disambiguating the proposed query results.
In ACQUA, we do not manage long-term user
preferences because we noticed that, in different
interaction sections, the same users are interested
in rather different types of information. We thus
decided to base the recommendations only on the
analysis of the users search behavior.

Future Trends
It is worth mentioning that the manual definition
of the first set of association rules supporting the

users search task was a lengthy work and might

not be easily replicated to revise the rules along
time. However, if the Water Resources Division
employs the ACQUA system as its official Web
site, the log files generated by the system will
provide structured evidence about user behavior
(in addition to e-mails and faxes). Thus, data-mining techniques could be exploited to automatically
recognize usage patterns and revise the association
rules accordingly.
Indeed, we believe that these techniques can
support the analysis of user behavior in an effective way, but they still have to be coupled with
human analysis, in order to validate and interpret
results: in several cases, these techniques have
generated some very interesting results, but also
other irrelevant or hardly understandable findings,
which have been discarded.
At any rate, Web usage mining techniques,
derived from machine learning methods such as
knowledge discovery in data (KDD or data mining) can contribute to automate the adaptation
of Web-based systems to the users. According
to the scheme proposed by Pierrakos, Paliouras,
Papatheodorou, and Spyropoulos (2003), ACQUA can be classified as a Web personalization
system offering task performance support: this
functionality involves the execution of a particular action on behalf of the user. In our case, the
system generates queries and makes the results
available as links to some files storing them. This
functionality is considered as the most advanced
personalization function and it is seldom offered
by Web-based personalized services.
The most suitable data-mining technique,
given the adaptive goals of the ACQUA system,
is the sequential pattern discovery, which is
aimed at identifying navigational patterns (event
sequences) in the analyzed data (in our case, Web
usage data). This methodology supports the discovery of event sequences that can be summarized
as follows: If event A, B, and C occur in that
order, then events D, E, and F always follow.

803

A User-Centered Approach to the Retrieval of Information in an Adaptive Web Site

Two types of methods are generally applied to

discover sequential patterns: deterministic techniques, which record the navigational behavior
of the users and extract knowledge from the
analyzed data, and stochastic methods, which
use the sequence of already-visited Web pages to
predict the behavior occurring in the next visits.
Once sequential patterns have been discovered,
the extracted knowledge can be automatically
integrated in the personalization process, and the
system behavior adapted accordingly.

Conclusion
We presented our experience in the design and
development of ACQUA, an interactive prototype
Web site for the Public Administration. The system
presents information about water resources and
supports the user in the search for generic information, as well as technical information about the
rivers, lakes, and underground waters.
The usability and functional requirements that
emerged during the design of the ACQUA system
were very interesting and challenging, as they
imposed the development of functions supporting
the efficient retrieval of data by means of a simple
user interface. We found out that the introduction
of basic adaptivity features, aimed at understanding the users information needs in detail, was
very helpful to meet these requirements.
We were asked to develop a system having a
simple user interface, designed to meet usability
and predictability requirements. This fact limited
our freedom to add advanced interaction features,
desirable in a Web site visited by heterogeneous
users; however, it challenged us to find a compromise between functionality and simplicity.
In order to address this issue, we developed two
interactive features enabling the user to create a
personal view on the information space:

804

The system offers a simple and an advanced

search functions targeted to novice and
expert users, respectively.
Moreover, the system carries out a query
analysis aimed at identifying the users
information needs, and applies association rules to extend the users queries and
complete the search results with data that
is usually retrieved together by end users.

Qualitative and quantitative evaluation results

showed that the adaptive user interface was more
successful than the nonadaptive one. The reason
was probably the concrete help offered by the
adaptive suggestions, which speed up the execution of time-consuming search tasks. Moreover,
the adaptive features were not perceived as intrusive and the user was allowed to skip useless
suggestions. Furthermore, the system did not
impose a previous annoying and discouraging
registration phase.
As discussed in Section Future trends, the
adaptive features offered by the ACQUA system
could be improved by the integration of Webusage mining techniques aimed at discovering
real usage patterns. In that way, the association
rules employed to identify the users implicit information needs could be automatically updated
along time. However, we believe that the rules we
manually defined provide a knowledge base that
cannot be replaced with automatically extracted
rules. In principle, both kinds of rules could be
integrated in order to enhance the effectiveness
of the system adaptations.

Acknowledgments
This work was funded by Regione Piemonte,
Direzione Risorse Idriche. We thank Giovanni Negro, Giuseppe Amadore, Silvia Grisello, Alessia
Giannetta, Maria Governa, Ezio Quinto, Matteo

A User-Centered Approach to the Retrieval of Information in an Adaptive Web Site

Demeo, and Vincenzo Pellegrino, who assisted

us during the system development and provided
the domain-specific knowledge.

References
Benyon, D. (1993). Adaptive systems: A solution
to usability problems. International Journal of
User Modeling and User-Adapted Interaction,
3, 6587.
Billsus, D., & Pazzani, M. (1999). A personal
news agent that talks, learns and explains. In
Proceedings of 3rd International Conference on
Autonomous Agents (pp. 268275).
Brusilovsky, P. (1996). Methods and techniques
of adaptive hypermedia. International Journal
of User Modeling and User-Adapted Interaction,
6(23), 87129.
Brusilovsky, P. (2001). Adaptive hypermedia.
International Journal of User Modeling and
User-Adapted Interaction, 11(12), 87110.
Bunt, A., & Conati, C. (2003). Probabilistic student modelling to improve exploratory behaviour.
International Journal of User Modeling and
User-Adapted Interaction, 13(3), 269309.
Chin, D. N., (2001). Empirical evaluation of user
models and user-adapted systems. International
Journal of User Modeling and User-Adapted
Interaction, 11(12), 181194.
Cotter, P., & Smyth, B. (2000). WAPing the Web:
Content personalization for WAP-enabled devices.
Proceedings of International Conference on
Adaptive Hypermedia and Adaptive Web-Based
Systems (pp. 98108).
Dix, A., Finlay, J., Abowd, G., & Beale, R. (1998).
Human computer interaction (2nd ed.). Prentice
Hall.

Dumas, J. S., & Redish, J. C. (1999). A practical

guide to usability testing. Norwood, NJ: Ablex.
Fink, J., Kobsa, A., & Nill, A. (1999). Adaptable
and adaptive information for all users, including
disabled and elderly people. New Review of Hypermedia and Multimedia, 4, 163188.
Gena, C., & Ardissono, L. (2004). Intelligent support to the retrieval of information about hydric
resources. Proceedings of 3rd International Conference on Adaptive Hypermedia and Adaptive
Web-Based Systems (pp. 126135).
Gould, J. D., & Lewis, C. (1983). Designing for
usability: Key principles and what designers think.
Proceedings of CHI 83 (pp. 5053).
Greenbaum, J., & Kyng, M. (1991). Design at
work: Cooperative design of computer systems.
Hillsdale, NJ: Lawrence Erlbaum.
GroupLens. (2005). GroupLens Research. Retrieved November 2, 2005, from www.GroupLens.org
Hk, K. (2000). Steps to take before IUIs become
real. Journal of Interacting With Computers,
12(4), 409426.
Horvitz, E., Breese J., Heckerman D., Hovel D.,
& Rommelse, K. (1998). The Lumire project:
Bayesian user modeling for inferring the goals
and needs of software users. Proceedings of 14th
Conference on Uncertainty in Artificial Intelligence, San Francisco.
Keppel, G. (1991). Design and analysis: A
researchers handbook. Englewood Cliffs, NJ:
Prentice-Hall.
Liu, F., Yu, C., & Meng, W. (2002). Personalized
Web search by mapping user query to categories. Proceedings of the 2002 ACM Conference
on Information and Knowledge Management,
McLean, VA.

805

A User-Centered Approach to the Retrieval of Information in an Adaptive Web Site

Maybury, M., & Brusilovsky, P. (Eds.). (2002).

The adaptive Web. Communications of the ACM,
45.
Moore, J. D., & Mittal, V. O. (1996). Dynamically
generated follow-up questions. IEEE Computer,
29(7), 7586.
Nielsen, J. (1999). Web usability. Indianapolis,
IN: New Riders Publishing.
Nielsen, J. (2004). Risks of quantitative studies.
Retrieved March 9, 2004, from www.useit.com/
alertbox/20040301.html
Norman, D. A., & Draper, S. W. (1986). User
centered system design: New perspective on HCI.
Hillsdale, NJ: Lawrence Erlbaum.
Oppermann, R. (1994). Adaptively supported
adaptivity. International Journal of HumanComputer Studies, 40, 455472.

Palmquist, R. A., & Kim, K. S. (2000). Cognitive

style and on-line database search experience as
predictors of Web search performance. Journal
of the American Society for Information Science,
51(6), 558566.
Petrelli, D., De Angeli, A., & Convertino, G.
(1999). A user centered approach to user modeling.
Proceedings of the 7th International Conference
on User Modeling (pp. 255264).
Pierrakos, D., Paliouras, G., Papatheodorou, C., &
Spyropoulos, C. D. (2003). Web usage mining as
a tool for personalization: A survey. International
Journal of User Modeling and User-Adapted
Interaction, 13(4), 311372.
Preece, J., Rogers, Y., Sharp, H., & Benyon, D.
(1994). Human-computer interaction. Addison
Wesley.

This work was previously published in Cognitively Informed Systems: Utilizing Practical Approaches to Enrich Information
Presentation and Transfer, edited by E. M. Alkhafia, pp. 142-166, copyright 2006 by IGI Publishing, formerly known as Idea
Group Publishing (an imprint of IGI Global).

806

807

Chapter 2.26

Auto-Personalization
Web Pages
Jon T. S. Quah
Nanyang Technological University, Singapore
Winnie C. H. Leow
Singapore Polytechnic, Singapore
K. L. Yong
Nanyang Technological University, Singapore

Introduction
This project experiments with the designing of a
Web site that has the self-adaptive feature of generating and adapting the site contents dynamically
to match visitors tastes based on their activities
on the site. No explicit inputs are required from
visitors. Instead a visitors clickstream on the
site will be implicitly monitored, logged, and
analyzed. Based on the information gathered,
the Web site would then generate Web contents
that contain items that have certain relatedness to
items that were previously browsed by the visitor.
The relatedness rules will have multidimensional
aspects in order to produce cross-mapping between items.
The Internet has become a place where a vast
amount of information can be deposited and also

retrieved by hundreds of millions of people scattered around the globe. With such an ability to
reach out to this large pool of people, we have seen
the expulsion of companies plunging into conducting business over the Internet (e-commerce).
This has made the competition for consumers
dollars fiercely stiff. It is now insufficient to just
place information of products onto the Internet
and expect customers to browse through the Web
pages. Instead, e-commerce Web site designing is
undergoing a significant revolution. It has become
an important strategy to design Web sites that are
able to generate contents that are matched to the
customers taste or preference. In fact a survey
done in 1998 (GVU, 1998) shows that around 23%
of online shoppers actually reported a dissatisfying experience with Web sites that are confusing
or disorganized. Personalization features on the

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Auto-Personalization Web Pages

Web would likely reverse this dissatisfaction and

increase the likelihood of attracting and retaining visitors.
Having personalization or an adaptive site can
bring the following benefits:
1.
2.

3.

4.

Attract and maintain visitors with adaptive

contents that are tailored to their taste.
Target Web contents correspondingly to their
respective audience, thus reducing information that is of no interest to the audience.
Advertise and promote products through
marketing campaigns targeting the correct
audience.
Enable the site to intelligently direct information to a selective or respective audience.

Currently, most Web personalization or adaptive features employ data mining or collaborative
filtering techniques (Herlocker, Konstan, Borchers, & Riedl, 1999; Mobasher, Cooley, & Srivastava, 1999; Mobasher, Jain, Han, & Srivastava,
1997; Spiliopoulou, Faulstich, & Winkler, 1999)
which often use past historical (static) data (e.g.,
previous purchases or server logs). The deployment of data mining often involves significant
resources (large storage space and computing
power) and complicated rules or algorithms. A
vast amount of data is required in order to be able
to form recommendations that made sense and
are meaningful in general (Claypool et al., 1999;
Basu, Hirsh, & Cohen, 1998).
While the main idea of Web personalization
is to increase the stickiness of a portal, with the
proven presumption that the number of times a
shopper returns to a shop has a direct relationship
to the likelihood of resulting in business transactions, the method of achieving the goal varies.
The methods range from user clustering and time
framed navigation sessions analysis (Kim et al.,
2005; Wang & Shao, 2004), analyzing relationship
between customers and products (Wang, Chuang,
Hsu, & Keh, 2004), performing collaborative filtering and data mining on transaction data (Cho

808

& Kim, 2002, 2004; Uchyigit & Clark, 2002;

Jung, Jung, & Lee, 2003), deploying statistical
methods for finding relationships (Kim & Yum,
2005), and performing recommendations bases
on similarity with known user groups (Yu, Liu,
& Li, 2005), to tracking shopping behavior over
time as well as over the taxonomy of products.
Our implementation works on the premise that
each user has his own preferences and needs, and
these interests drift over time (Cho, Cho, & Kim,
2005). Therefore, besides identifying users needs,
the system should also be sensitive to changes in
tastes. Finally, a truly useful system should not
only be recommending items in which a user had
shown interest, but also related items that may
be of relevance to the user (e.g., buying a pet =>
recommend some suitable pet foods for the pet,
as well as suggesting some accessories that may
be useful, such as fur brush, nail clipper, etc.).
In this aspect, we borrow the concept of category management use in the retailing industry
to perform classification as well as linking the
categories using shared characteristics. These
linkages provide the bridge for cross-category
recommendations.

Description of System
In this article, we seek to provide an adaptive
feature using a fast and cost-effective means.
The aim is to provide adaptiveness in the sense
that when a visitor selects the next link or a new
page, the contents of the page generated will
have relatedness to previous pages contents.
This adaptive feature will be immediate and will
not experience delay or repetitive computational
filtering problems, as compared to using mining
or collaborative filtering (Claypool et al., 1999;
Basu et al., 1998).
The rules-based technique offers an excellent and flexible mechanism to specify rules that
map categories that exhibit relatedness among
themselves (IBM, 2000). Adding new product

Auto-Personalization Web Pages

lines is simple, by just adding new sets of rules

to map the new products accordingly. For direct
item-to-item relatedness mapping, it is not so
scalable and feasible to implement through use
of the rules-based technique. Instead we will
use content-based filtering for generating direct
item-to-item mappings. The content-based technique (Claypool et al., 1999; Basu et al., 1998)
allows item-to-item mapping to be implemented
in a scalable manner by just defining the items
attribute, and the recommendation engine will
automatically generate or match items of same
attribute without involving user efforts (Basu et
al., 1998).
In order to facilitate the deployment of these
recommendation techniques, the Web domain is
structured into their respective categories that
exhibit relatedness among them. For example, pet
dog would have relatedness to books on dogs. Each
of the categories is given a unique ID value. The
relatedness rules make use of these IDs to generate
recommendations. The Web site domain is structured into supernodes (SA, SB) which branch
into child nodes (A1,A2An;K1,K2Kn).
These supernodes are a representation of products
on the Web site, and the child nodes are used to
represent the breakdown of the products into
categories. Below each of the child nodes are the
sub-child nodes (Aa1,Axm,Ka1,Ka2,Kxm)
that represent the items. Each of the child nodes
(A1,A2An;K1,K2Kn) is identified with
its corresponding ID value. With this structure,
rules-based mapping can be easily identified and
applied among the child nodes by defining the IDs
that will result into a recommended page.
The syntax of a relatedness rule is:
IDa:IDb:=>Target page
The entries IDx represent the IDs of Web pages
that have relatedness and thus can be mapped
directly to the Web page (link) identified as target
page. A rule is considered matched when any
of the Web page IDs in the rule is also found to

exist in the selectors list (visitors profile). The

selectors list is compared against all the rules in
the rule file. Only one of the rules will be used,
and that is the rule that has the most number of
IDx elements matching the selectors list. In the
event of tie, the rule which matches with selector
entry that carries the higher points will be used,
or if they still tie, then precedence of rule entry
in the rule file will be used to determine the final
rule to use.
Mappings across sub-child nodes are done
based on their attributes (content based). A file
will be used to define the list of items (or contents)
of each Web page (at category level) and also any
of attributes for each of the items. The syntax for
this item list file is:
Idxa:idxb::attra:attrb=>display.html
The entries idxx are indexes to the html page
that contains all the information regarding the
item. An example of these indexes could be the
type of pet. The information about each type of
pet is found in the corresponding html file. An
item sorter engine will, at startup, sort out the
indexes and their corresponding html files. This
will allow the page contents to be arranged or
displayed based on any of these indexes. The
other set of entries attrx define the attributes
of the item. When the visitor selects any of the
items for browsing, the corresponding attrx will
be tracked. Whenever the visitor browses a new
page, the recommendation engine will check for
items that have the same attrx and automatically
include the corresponding html file for display
(see Figure 1).
Our prototype system also incorporates means
of generating a dynamic profile that changes as
the visitor browses through the Web site. Implicit
tracking does not require any explicit inputs or
intervention by the visitor. Conventionally this
is done either through use of user authorization,
hidden form fields, URL rewriting, or cookies
(Hunter et al., 2002). Although cookies offer an

809

Auto-Personalization Web Pages

Figure 1. Web domain structure

SA

Aa2

Aam

An

A2

A1

Aa1

SK

Ab1

Ab2

Abm

Ax1

Ax2

Axm

elegant solution to the stateless HTTP protocol

when compared to the other three, it is also
frowned upon by some visitors and as such are
disabled (Mobasher et al., 1999; Cooley, Mobasher,
& Srivastava, 1999). Instead, the tracking of visitors activities is implemented through use of a
session object on the server side (Hunter et al.,
2002). This object will be used to record all the
Web pages (tagged with IDs on the server side)
browsed by the visitor during the current session,
hence obtaining the profile of the visitor. However, a cookies mechanism is still implemented to
provide the remember feature, in the sense that
if the cookies feature is enabled on the visitors
browser, the activities on the Web site from the
most recent visit will be available (and retrieved)
during the next visit. This allows the system to
build the visitors profile immediately. Hence
we are able to combine the static profiling (from
cookies) with the dynamic one (tracked by session
object) to build a profile that can remember a
visitors previous preferences and dynamically
change itself as the visitor continues browsing
the Web site.

810

Ka1

Ka2

Kam

Kn

K2

K1

Kb1

Kb2

Kbm

Kx1

Kx2

Kxm

In order to give priority to activities that occur more often and recent, a point system is used
whereby the current session log is given heavier
weight than those retrieved from the cookies so
that the current activities will be more likely to
be nominated into the visitors latest profile. The
activities log tracked by the session object will
be given higher consideration during profiling in
order for the system adaptiveness to be reflected
accordingly and immediately to the changes in
the visitor browsing behavior. In this design, a
total of three activities logs are used (two from
cookies if available, and the remaining one is
from the current session object that tracks the
current activities).
In order to demonstrate the adaptiveness feature, the Web site domain should consist items or
products that have the following characteristics:
1. Products that can be easily categorized.
2. Selection of items or products should be able
to reflect the taste or preference of visitor.
3. Certain form of relatedness between the
products.

Auto-Personalization Web Pages

With the above considerations taken into the

account, we implemented a Web portal using an
online shop named My-Pet that sells the following products:
1.
2.
3.
4.

Pets
Books on pets
Accessories for pets
Food for pets

To demonstrate how the system functions, we

selected the following list of categories:
Pets: Bird, Dog, Cat, Hamster, Fish
Books:
How to care, Diseases of,
Nutrition for
Accessories: Pet wears, Toys, Restraints
Foods:
Meals, Snacks, Supplements
The My-Pet recommendation engine is based
on the relatedness determination among its products. The relatedness determination for My-Pet
is defined at three levels:

1.

Relatedness between categories across products (e.g., pet dog and dog food and books
on dogs);
2. Relatedness between categories in the same
product (e.g., books on how to care for dogs
and books on diseases of dogs and books on
nutrition for healthy dogs); and
3. Relatedness at items level across products
(e.g., food and accessory items from the
same company/brand).
Items (1) and (2) are achieved based on the
rules-base technique, while content-based filtering is used for item (3).
The server software architecture is decomposed into independent modules that have very
similar structure. Each product has its own module
that is responsible for presenting and generating
its dynamic Web page contents. The development
platform chosen is Java Server Pages (JSP), which
offers object-oriented components deployment.
By developing the whole Web program in JSP

Figure 2. Software architecture

Logs
Processor
(Profiler)

Visitor's
PC

Internet
network

Java
Web Server

JSP
Page

Recommendation
Engine
(Rule-based)

Items Sortor
(Content-based)

811

Auto-Personalization Web Pages

and Java, the server program is OS-platform

independent.
Each module is made up of the following
components:

JSP Page
Logs Processor (JavaBean)
Recommendation Engine (JavaBean)
Item Sorter (JavaBean)

The JSP page is responsible for delegating the

tasks to the respective components, and at the
end presenting the results into HTML format for
viewing by the visitor.
The Logs Processor component serves to log
all the activities by the visitor and process the
activities logs before submitting to the Recommendation Engine. It plays the role of profiling
the visitors activities. The Recommendation
Engine is mainly used for determining productto-product mapping based on the selectors list
(profile) generated by the Logs Processor. The
Items Sorters main function is to read the list of
product items and sort them such that they can
be retrieved via means of indexing or key. This
component is also used to retrieve an item attribute
that is used for Content-Based filtering, based
on the input determined by the Logs Processor
component (see Figure 2).

Impact of System
The prototype system is implemented using a Java
Server from JRUN which provides the container
for running JSP and JavaBeans classes. The database stores all external data inputs. The Web
server PC is installed with JRUN and the Web
application. Customers can access the Web site
using any Web browser.
To illustrate the effectiveness of this system in
providing dynamic content generation, a series of
user inputs and the subsequent system responses
are tabulated.

812

From the trial run results obtained, it was

observed that the system behaved as expected,
providing the adaptive feature of generating
Web contents based on users activities on the
Web site (see Table 1).Browsing Activities Web
Content Generated
The system is able to demonstrate the following adaptiveness features:
1.

the ability to automatically present the choice

of viewing items in a page (e.g., what type
of pets) based on visitors previous preference;
2. the ability to automatically present Web
contents that display the category that exhibits relatedness (as defined in Rule Files)
to other products categories browsed by the
visitor previously;
3. the ability to generate an option to view list
of items that has relatedness to other items
that the visitor has visited or selected; and
4. the ability to generate a recommendation
based on cookies retrieved from the visitors
PC.

Conclusion
We have developed a system that can adapt its
Web contents based on visitors activities on the
Web site through combining rule-based with content-based filtering techniquesresulting in an
implementation that is both flexible and can rapidly
adjust its recommendations. Rule-based structure
offers cross-product mapping. Content-based
filtering takes the items attributes into account
when generating recommendations. The system
transparently and seamlessly tracks the visitor
on the server side and does not require explicit
inputs (ratings or purchases or login account) to
determine the visitors profile dynamically.
My-Pets system design utilizes the concept of
category management, which is widely practiced
in brick-and-mortar shop fronts and maps product

Auto-Personalization Web Pages

Table 1. Results obtained from trial runs

Browsing Activities
Customer initially selects Dog category under the Pet
section and chooses from the choice of displaying
dogs based on friendliness/loyalty. The customer then
chooses Cat category under Pet section.

Web Content Generated

The system automatically displays the choices of
cats for selection. The cats are displayed based on
the degree of friendliness/loyalty to owners.

In continuation from above, the customer then selects

the choice of displaying Cats based on feeding habits.
User then re-selects the Dog category.

The system automatically switches from presenting

cats to presenting dogs based on their feeding
habits.

Customer selects Dog category under the Pet section.

Customer then chooses Book product.

The system automatically presents books that are

related to dogs.

Customer selects Supplements category under Food

section. Visitor then selects Petsfriend (from the list
of brands) and clicks the item Best Choice Dog
Cookies. Customer then chooses Book product.

The system automatically brings the customer to

Nutrition category under the Book section and also
displays the option for viewing book titles related to
dog nutrition from Petsfriend.

Customer selects Dog category under Accessories

product and then chooses Dog category under Books
as well. Customer subsequently closes the browser
before restarting again to browse the Web site again,
but this time around selecting the Food product
instead.

The system automatically forwards the visitor to the

Dog category when the Food product was selected
in this new browsing session.

taxonomy into a cyberspace virtual environment.

The key to using category management is that it
makes system expansion systematic and easy. In
our illustration, we included a handful of products
and categories. However, the system architecture
provides a structure that allows more products or
categories to be added easily. Sub-categories can
also be added as long as each of them is given a
unique ID page. A new set of relatedness rules
can then be defined for these newly added elements. This is in fact a strength of rule-based

implementation, where new linkages between

product categories and subcategories can be added
and removed as need arises. Rules may also carry
different precedence values.
Such a highly adaptive system does have a
weaknessit has taken away some navigation
control from the user. It was noted in a 1998 survey
(GVU, 1998) that 17% of Web users experience
difficulty in returning to pages visited before. A
system with an auto-recommendation feature is
likely to fair worse in this aspect. Therefore, for

813

Auto-Personalization Web Pages

future improvement to the system, a feature could

be added to the system whereby a link page is
created and this page contains all links to pages
that were previously browsed by the user.
In addition, a search option can also be included. This search option indirectly provides a
way for the visitor to tell the system his/her taste
or preference. This serves as an excellent and accurate input in building a profile that even more
closely reflects the visitors preferences. Hence the
system will be able to use a more accurate visitors
profile when generating recommendations.

References
Basu, C., Hirsh, H., & Cohen, W. (1998). Recommendation as classification: Using social and
content-based information recommendation.
Proceedings of the 1998 Workshop on Recommender Systems (pp. 11-15).
Cho, Y. B., Cho, Y. H., & Kim, S. H. (2005).
Mining changes in customer buying behavior for
collaborative recommendations. Expert Systems
with Applications, 28(2), 359-369.
Cho, Y. H., & Kim, J. K. (2004). Application of Web
usage mining and product taxonomy to collaborative recommendations in e-commerce. Expert
Systems with Applications, 26(2), 233-246.
Cho, Y. H., Kim, J. K., & Kim, S. H. (2002). A
personalized recommender system based on Web
usage mining and decision tree induction. Expert
Systems with Applications, 23(3), 329-342.
Claypool, M., Gokhale, A., Miranda, T., Murnikov,
P., Netes, D., & Sartin, M. (1999). Combining
content-based and collaborative filters in an online
newspaper. Proceedings of the ACM SIGIR Workshop on Recommender Systems.
Cooley, R., Mobasher, B., & Srivastava, J. (1999).
Data preparation for mining World Wide Web

814

browsing patterns. Journal of Knowledge and

Information System, 1(1), 5-32.
GVU. (1998). GVUs 8th WWW user survey.
Retrieved from http://www.cc.gatech.edu/gvu/
user_surveys/
Herlocker, J. L., Konstan, J. A., Borchers, A. L.,
& Riedl, J. (1999). An algorithmic framework for
performing collaborative filtering. Proceedings
of the Research and Development in Information
Retrieval Conference (pp. 230-237).
Hunter, J., & William, C. (2002). Java servlets
programming. Sebastopol, CA: OReilly & Associates, Inc.
IBM. (2000). Web site personalization. Highvolume Web sites. Retrieved April 19, 2001,
from http://www.128.ibm.com/developerworks/
websphere/library/techarticles/hipods/personalize.html
Jung, K. Y., Jung, J. J., & Lee, J. H. (2003).
Discovery of user preference in personalized
design recommender system through combining
collaborative filtering and content based filtering. Lecture Notes in Computer Science, 2843,
320-327.
Kim, D., & Yum, B. J. (2005). Collaborative
filtering based on iterative principal component
analysis. Expert Systems with Applications, 28(4),
823-830.
Kim, Y. S., Kim, B. J., Song, J., & Kim, S. M.
(2005). Development of a recommender system
based on navigational and behavioral patterns of
customers in e-commerce sites. Expert Systems
with Applications, 28(2), 381-393.
Mobasher, B., Cooley, R., & Srivastava, J. (1999).
Automatic personalization based on Web usage
mining (Technical Report TR99010). Department of Computer Science, DePaul University
(pp. 558-567).

Auto-Personalization Web Pages

Mobasher, B., Jain, N., Han, S., & Srivastava,

J. (1997). Web mining: Pattern discovery from
World Wide Web transactions. Proceedings of
the 9th IEEE International Conference on Tools
with Artificial Intelligence.
Spiliopoulou, M., Faulstich, L. C., & Winkler,
K. (1999). A data miner analyzing the navigational behaviour of Web users. Proceedings of
the International Conference of AIWorkshop
on Machine Learning in User Modelling.
Uchyigit, G., & Clark, K. (2002). Agents that
model and learn user interests for dynamic collaborative filtering. Lecture Notes in Computer
Science, 2446, 152-163.
Wang, Y. F., Chuang, Y. L., Hsu, M. H., & Keh, H.
C. (2004). A personalized recommender system
for the cosmetic business. Expert Systems with
Applications, 26(3), 427-434.
Wang, F. H., & Shao, H. M. (2004). Effective
personalized recommendation based on timeframed navigation clustering and association
mining. Expert Systems with Applications, 37(3),
365-377.
Yu, L., Liu, L., & Li, X. (2005). A hybrid collaborative filtering method for multiple-interests
and multiple-content recommendation in e-com-

merce. Expert Systems with Applications, 28(1),

67-77.

Key Terms
Category Management: Classify and manage
items based on some predetermined categories.
Clickstream: A sequence of mouse clicks.
Collaborative Filtering: Unveiling general
patterns through sniffing through users past
activities.
Personalization: Customization to individual
users preferences and needs.
Portal: The entry node/point of navigation
semantic unit for a Web site.
Profiling: Capturing individual users interests and needs.
Self-Adaptive: Ability of a Web portal to
automatically adjust its presentations to perceived
users preferences.
Session Object: Information items that capture characteristics and activities of a user during
a Web session.

This work was previously published in the Encyclopedia of E-Commerce, E-Government, and Mobile Commerce, edited by
M. Khosrow-Pour, pp. 20-25, copyright 2006 by Information Science Reference, formerly known as Idea Group Reference (an
imprint of IGI Global).

815

816

Chapter 2.27

A Qualitative Study in
Users Information-Seeking
Behaviors on Web Sites:
A User-Centered Approach to
Web Site Development
Napawan Sawasdichai
King Mongkuts Institute of Technology, Ladkrabang, Thailand

ABSTRACT
This chapter introduces a qualitative study of
users information-seeking tasks on Web-based
media, by investigating users cognitive behaviors
when they are searching for particular information
on various kinds of Web sites. The experiment,
which is a major part of the recently completed
doctoral research at the Institute of Design-IIT,
particularly studies cognitive factors including
user goals and modes of searching in order to
investigate if these factors significantly affect
users' information-seeking behaviors. The main
objective is to identify the corresponding impact
of these factors on their needs and behaviors in
relation to Web site design. By taking a userbased qualitative approach, the author hopes
that this study will open the door to a careful
consideration of actual user needs and behav-

iors in relation to information-seeking tasks on

Web-based media. The results may compliment
the uses of existing quantitative studies by supplying a deeper user understanding and a new
qualitative approach to analyze and improve
the design of information on Web sites.

INTRODUCTION
When visiting a Web site, each user has a specific
goal that relates to a pattern of needs, expectations, and search behaviors. They also approach
with different modes of searching based on varied
knowledge, experience, and search sophistication.
This leads to differences in information-seeking
strategies and searching behaviors. Since information on Web sites is traditionally structured and
presented based on Web sites goals and contents,

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

it may or may not match with user goals or search

behaviors.
Because of these problems, information
structuring is the essence of Web design since
these problems cannot be solved by the development of technically sophisticated systems alone.
User search behaviors need to be studied and
deeply understood in order to design systems
that allow them to perform their informationseeking tasks easily, without struggle and
frustration. The contents need to be authored,
organized, structured, and presented to fit their
needs, expectations, and search behaviors,
while being able to carry out the goal of the
Web site simultaneously. Both the provider and
user must benefit at the same time to ensure
the Web site success. As a result, user-centered
design process is important in Web development
to help people succeed within an information
context that seeks to achieve business goals
(Brinck, Gergle, & Wood, 2002).
In attempts to move toward user-centered
design, many studies have been developed to
establish design principles that better serve
Web-based media. Among these attempts, Web
usability, grounded in human-computer interaction (HCI), has currently assumed a significant
role underpinning the design of many Web sites
in order to maximize efficient use. Web usability
studies and practices are primarily concerned with
people performing a task accurately, completely,
and easily. These may involve making information accessible, retrievable, legible, and readable,
ensuring that all Web pages are reachable and
practically navigated, or dealing with technical
aspects of media interface and Web system by
ensuring that all system functionality can be
operated correctly and easily.

User Research in Web Development

User research in relation to Web site development is mostly conducted by using quantitative
methods or automated programs, such as data

mining and Logs File Analysis (analyze usage

data), GOMS analysis (predict execution and
learning time), and Information Scent modeling (mimic Web site navigation) serve different
purposes. These automated methods are particularly essential to usability testing (evaluation),
especially in cases where numerous users are
involved since they can reveal a substantial amount
of information with regard to usage patterns by
representing the actual usage characteristics.
Some also provide in-depth statistical analysis of
usage. For example, logs file analysis can show
overall hits, conversion rates, entrance pages,
search terms, peak times, demographics, and
system down-time (see Figure 1 and 2). These
develop an understanding of how the Web site
is being used by the actual users, which helps
identify potential problems of the Web site, and
may assist in suggesting a change or directing the
future design (Brinck et al., 2002).
However, the limitations of these automated
methods are that they cannot be employed
without an existing Web site; the Web site
needs to be prototyped or implemented at
some level before these methods can be applied
since they are intended as an analytical means
rather than a generative one. More importantly,
these automated methods cannot capture important qualitative and subjective information
such as user preferences and misconceptions
(Ivory & Hearst, 2001). They tend to yield a
higher level of user data what they do or
what they do not do but they usually fail to
capture and analyze user cognitive behaviors
such as their satisfaction, decision-making
pattern, or reasons that underpin their needs
and behaviors.
Therefore, qualitative study using nonautomated methods such as user observation, focus
groups, user interviews and surveys still play
an important role in Web development. These
nonautomated methods can be used in the design
process to capture, analyze, and conceptualize
Web structure before usability evaluation takes

817

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

Figure 1. An example page from logs file analysis: Visits

Figure 2. An example page from logs file analysis example: Top entry pages

place. They can be employed in the evaluation

process as complements to the automated methods as well, in order to help capture and analyze
qualitative and subjective user information that
is missing from the automated methods alone.
Applying both quantitative and qualitative studies

818

can significantly improve the quality and quantity

of user input and feedback that may help suggest
a change or direction that better supports user
satisfaction. As a result, this study investigates a
relatively new area of user research in Web-based
media, offering a user-centered perspective with

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

consideration of user goals, user modes of searching and their search behaviors by taking a userbased qualitative approach. The study expects
to expand understanding within the area of user
studies, and accordingly investigates how these
user cognitive factors contribute to differences
in user information needs and their information-seeking behaviors on Web-based media,
particularly in view of user search strategies and
user search methods. Understanding within this
area will contribute to the further development of
information architecture and interface design.

SCOPE OF THE RESEARCH

The practices of Web site development are fundamentally concerned with two equally important
and interrelated parts: (1) Web functionality and
(2) Web information. In the user-centered perspective, usability, accessibility, sustainability,
suitability, credibility, and usefulness of both Web
functionality and Web information for its intended
users are important for the Web to succeed. In
most current practices, the user-centered approach
is usually taken into design consideration in a
general sense; for example, by conducting user
studies to establish who are the intended users of
the Web site (user profiles), and what do they want
to achieve (user goals)? Others may perform user
testing with regard to usability evaluation in terms
of what is working, and what is not (usability of
Web functionality). These current user-centered
approaches are concerned with Web functionality
rather than Web information. Furthermore, they
pay considerably more attention to usability aspects while the Web information content receives
less attention. Therefore, this research primarily
focuses on the design of Web information, particularly in view of the importance, suitability,
and usefulness of information design provided
on Web sites (see Figure 3).
The study is particularly concerned with user
information needs and user information-seeking

behaviors; it also investigates whether the design

of information provided on the Web site supports
these needs and behaviors. Secondly, the research
is also concerned with the suitability and usefulness of Web functionality necessary for users to
gain access to the information they need. The
research also looks into different types of search
methods or search tools provided on Web sites to
investigate whether these search methods or tools
are useful and suitable to user search strategies.
When searching, each user has a different
search plan: For instance, they lightly explore or
seriously search and they select search methods
and tools in order to easily achieve their goal.
This search plan is primarily based on their
search strategy, which is the scheme that generally distinguishes user search patterns. Besides
search strategies, each user may use a particular
search method, which is the procedure for how
they actually perform their search. User search
method is based on the types of search tools they
choose for their search, which may include menu
bar, table of contents, index, or search engine. User
search strategies can range from general or lessfocused search to more specific or purposeful
search. Furthermore, it may change from time
to time based on the current context or situation
that unfolds while they are searching as well
as the search results that they find or retrieve.
Based on user search strategies and its results,
user search methods are changed accordingly.
As a result, this research is aimed to uncover
the primary factors governing or influencing
these user search strategies and methods.
More importantly, in order to understand and
eventually determine what types, characteristics,
formats, and presentation methods for information
is suitable and useful for users, the research needs
to investigate the relatively new areas of user-centered approaches to Web site design: user goals
and user modes of searching (see Figure 4).
Information scientists have studied user modes
of searching for decades, and these ideas are well
categorized by Rosenfeld and Morville (1998) in

819

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

Figure 3. Primary focuses of the research: The design of Web information-based on user-centered approach, particularly with regard to usefulness and suitability

their book Information Architecture for the World

Wide Web. However, since the notion of user modes
of searching has never been elaborated in terms
of what to expect from their differences, it needs
further investigation this becomes the primary
focus of this research to uncover its substantial
impact on user needs, expectations, search

820

strategies, or information-seeking behaviors

in this medium. In addition, typical user profiles are also categorized in order to determine
whether these profiles exert any substantial
impact on user needs, expectations, or search
patterns. These profiles particularly focus on
user demographic and techno-graphic data,

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

Figure 4. Primary focus of the research: User

goals and user modes of searching

the context of specific Web site goals to reveal

common user search patterns, search strategies,
and search methods associated with each case
and to identify the primary problems that occur
in each pattern.

User Goals
Each user has a specific goal when visiting a Web
site. Different goals suggest different kinds of
needs, expectations, and search behaviors, which
are factors in Web usage and success. Further, users may access the same Web site with different
goals at different times; moreover, they often link
several goals and explore them sequentially. User
goals may be categorized as follows:

including prior knowledge in the content they

are searching, prior experience in the particular
or similar Web site interface, and sophistication
in general Internet use.

RESEARCH ANALYTIC FRAME

The research begins with defining and categorizing the important elements or factors of the study,
including user goals, user modes of searching, as
well as Web site goals and contents. Accordingly,
the research analytic frame is established to help
identify potential cases for the study. User goals
and modes of searching were investigated within

To seek specific information: In this

category, users may engage a Web site to
search for specific information that helps
them to stay updated, make decisions, fulfill
a specific inquiry, perform tasks, learn, or
conduct research.
To fulfill personal interests: Some users
may engage a Web site as a resource for
pleasure to fulfill a personal interest (e.g.,
watching a video clip or listening to music
on an entertainment Web site).
To communicate and/or perform tasks:
Others may use Web sites as a channel for
communicating or as a means for performing tasks (e.g., connecting to a community
Web site or paying bills on a company Web
site).

Among these three categories of user goals, the

information-seeking goal is prevalent and poses
the greatest problem for users. Consequently this is
the primary investigative focus in this research.

User Modes of Searching

Besides user goals, users also approach a Web
site with varied levels of specification of their

821

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

needs and different levels of search specification

and determination, this leads to differences in
information-seeking behaviors including search
strategies, search methods, and selection of search
tools. Some users may know exactly what they
are looking for and where to find it, while others
are without a clue. Since these search behaviors
and user expectations vary widely, it is important
to recognize and distinguish among them noting
their differences.
A current study (Rosenfeld & Morville, 1998)
has delineated users different modes of searching as known-item searching, existence searching, exploratory searching, and comprehensive
searching (research). Based on Rosenfeld and
Morvilles model, user modes of searching are
modified and extended in this research to include
topical searching which falls between existence
and known-item searching. User modes of searching may be categorized as follows:

822

Exploratory searching (browsing): Users

have a vague idea of their information needs.
They do not know exactly what they are
hoping to find, but some may know how to
phrase their question. They want to explore
and learn more.
Existence searching: Users have an abstract
idea or concept of what they are hoping to
find, but do not know how to describe it
clearly or whether the answer exists at all.
They want to search for what matches their
idea or mental image.
Topical searching: Users know what they
want in general. Some may want to search
for an answer to their specific question.
They know what they are hoping to find,
but do not know where/which categories
they should look for.
Known-item searching: Users know exactly what they want, and usually know
where/which categories they should look
for. Users information needs are clearly
defined and have a single, correct answer.

Comprehensive searching (research): Users want to search for specific information,

and they want everything available regarding
this information. Users information needs
are clearly defined, but might have various
or many answers.

Users Search Behaviors

When users are searching for information and trying to accomplish their goals, they move between
two cognitive states (thoughts/decisions with
regard to their goal and interest) and physical states
(interactions concern with functions, navigation, and computer performance) with regard to
information provided on each Web page.
For instance, some users may want to keep
searching because they need detailed information,
while others may be satisfied with only a short
descriptive text presented on the first page. Some
may prefer textual information, but others may
feel more satisfied with visual information. These
search behaviors may be identified as shown in
Table 1. These cognitive and physical behaviors
with regard to user search, previously observed
from initial user observation, will be further used
to establish a coding scheme used in an analytical
process of the research.

Web Site Goals and Contents

While this research focuses on the relatively new
areas of user studies: user goals and user modes
of searching, other factors such as site contents,
site goals, and site functions nevertheless play a
significant role in determining the design of a Web
site. Each Web site contains unique contents and
goals depending on the nature of the company,
institution, or individual that owns that Web site.
Based on the book Web Navigation: Designing
the User Experience (Fleming, 1998), these Web
sites can be distinguished and generalized by the
similarities of their goals and contents into six
categories: (1) commercial Web site, (2) identity

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

Table 1. Users cognitive and physical search behaviors

Users Cognitive Behaviors (thoughts/decisions)

Users Physical Behaviors (interactions)

Some information is found, and they want to learn

more, or want to know the details of the retrieval
documents.

Users keep searching in the current retrieval

results.

The intended information is found. Users

primary information needs are fulfilled, but users
are interested in finding other relevant or related
information.
The intended information is found. Users primary
information needs are fulfilled. Users are ready to use
information they found to take further action(s).

Users keep searching by changing search strategy.

Users record the information they found.
Users go back to the selected (bookmarked) page
or results.
Users give up.

The intended information is not found, or not

enough to take further action(s). Users primary
needs are not fulfilled. Users need to keep searching.
Users make a positive decision (decide to
proceed) about something according to information
they found.
Users make a negative decision (decide not to
proceed) about something according to information
they found.
Users are satisfied. All users needs are fulfilled.
Users are able to accomplish their goals based on the
information they found.
Users are not satisfied. Users needs are not
fulfilled. Users are unable to accomplish their goal(s).

Web site (Web site for company or institution),

(3) information Web site, (4) education Web site,
(5) entertainment Web site, and (6) community
Web site. However, only the first four categories,
in which the problems of information-seeking
tasks are primarily found, will be investigated

in this study. Moreover, entertainment and community Web sites are quite different from other
Web sites because of their unique goals, contents,
and functions.
By simultaneously considering the three
important factors of information design on

823

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

Web sites: (1) Web site goals and contents, (2)

user goals, and (3) user modes of searching, an
analytic frame is constructed. Different aspects
of each factor are systematically combined with
one another to establish prominent cases or
scenarios for the study; each of which presents
a unique combination of the three factors: (1)

Web site goals and contents, (2) user goals, and

(3) user modes of searching.
Nevertheless, these cases are not mutually
exclusive; they might overlap or combine since
one Web site may consist of two or more combinations (see Figure 5). As shown in Figure 6,
case 1 represents the scenario in which users with

Figure 5. The research analytic frame: Generating 10 different study cases for the research

824

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

Figure 6. An example scenario

exploratory searching mode visit a commercial

Web site in order to find information to make a
decision. Case 2 represents a similar scenario to
case 1; however, users in case 2 arrive with an
existence mode of searching. Cases 3, 4, and 5
represent similar scenarios in which users visit
identity (company) Web sites to find information
to fulfill a specific inquiry.
However, each case has a distinctive search
mode. Users in case 3 approach with an existence
mode; case 4 with a topical mode; while case 5
approaches with a known-item mode. Cases 6, 7,
8, and 9 represent the scenarios in which users
visit information Web sites to find information to
stay updated. Though, they are assigned different
modes of searching, which include exploratory,
existence, topical, and known-item modes respectively. Case 10 represents a scenario in which users
with a comprehensive mode of searching approach
an educational Web site to find information for
learning or researching a specific content. Each of
these 10 cases will be investigated and analyzed

to uncover similarities and differences in patterns

of user information-seeking behaviors, as well as
to identify user information needs, user search
strategies, and user search methods associated
with different user goals, modes of searching,
and Web site characteristics.

RESEARCH QUESTIONS
The study is specifically conducted within these
selected 10 cases generated from the research
analytic frame shown in Figure 5 in order to find
the answers to these research questions:

What are the common patterns of user

information-seeking behavior presented in
each study case?
What are the user search strategies, search
methods, or selected search tools commonly
found or employed in each study case?

825

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

What kinds of information do users need in

each study case in terms of the types, characteristics, formats, presentation methods,
quantity and quality of information?
What are the key factors in each study case
that help or obstruct users to accomplish the
information-seeking task?

The research findings that answer these questions will be analyzed to identify the relationships
existing among user goals and user modes of
searching with their information needs, search
strategies, and search methods. These results
will help establish the classification of cognitive
factors, as well as provide an analysis framework
for information design on Web sites.

RESEARCH METHODOLOGY
Research Methods
A qualitative research method is used in this
study to explore the similarities and differences of
user search patterns. Users information-seeking
behaviors are observed through controlled observation, through video observation combined with
protocol analysis. User profiles are also collected
through a series of questionnaires. Ten scenarios
are designed to create the 10 study cases originating from the research analytic frame to help the
participants enter the situation and the tasks they
needed to accomplish. Each scenario is embedded
with a particular mode of searching, and a different search goal resulting in the performance of a
task, ranging from open-ended to very specific
purpose and search.

826

Scenario 1 explores a commercial Web

site: Expedia.com. User goal is to make a
decision; user search mode is exploratory
searching.

Scenario 2 explores a commercial Web

site: Toyrus.com. User goal is to make a
decision; user search mode is existence
searching.
Scenario 3 explores an identity Web site:
Paris-ile-de-France.com. User goal is to
fulfill a specific inquiry; user search mode
is existence searching.
Scenario 4 explores an identity Web site:
Apple.com. User goal is to fulfill a specific
inquiry; user search mode is topical searching.
Scenario 5 explores an identity Web site:
FoodTV.com. User goal is to fulfill a specific
inquiry; user mode is known-item searching.
Scenario 6 explores an information Web
site: TVGuide.com. User goal is to stay
updated on some interesting topics; user
search mode is exploratory searching.
Scenario 7 explores an information Web
site: ABCNews.com. User goal is to stay
updated on a specific topic; user search mode
is existence searching.
Scenario 8 explores an information Web
site: DiscoveryHealth.com. User goal is to
stay updated on a specific topic; user search
mode is topical searching.
Scenario 9 explores an information Web
site: CNN.com. User goal is to stay updated
on a specific topic; user mode is known-item
searching.
Scenario 10 explores an information/education Web site: WebMD.com. User goal is
to research and learn about a specific topic;
user search mode is comprehensive searching.

Fifty participants from different cultures, all

of whom were literate in English, are observed
regarding how they search for information and try
to accomplish the tasks defined by the scenario

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

they received. The participants approached the

selected Web site with unspecified and various
modes of searching and searched for information
with goals appropriate to the selected scenario. Ten
participants are randomly selected to participate
in each scenario, with each participant doing two
cases or two different scenarios.
As a result, the research collects in total 100
observation cases, which consist of 10 cases for
each of 10 scenarios. The participants interactions (physical behaviors) on Web sites are simply
captured through a video recorder. Furthermore,
by using protocol analysis, the participants express verbally what they think while performing
tasks in order to reveal their thoughts (cognitive
behaviors) and comments, which are extremely
important for the analytical process.

Analysis Methods
Since the research data collected from participants is qualitative in nature, several methods of
qualitative analysis are used in this research to
carefully analyze various aspects of the data, in
order to obtain integrated research findings that
answer the related but different research questions
on which this research focuses. Each analysis
method used in the study delivers distinctive
analytical results answering a specific research
question. The analytical results obtained from
these different analysis methods are also crossexamined in order to accumulate further findings.
This collective analysis process helps to uncover
the pattern of relationship that exists among various user cognitive factors, as well as to identify
their substantial impact on user search behaviors
and information needs.
Thematic analysis (Boyatzis, 1998), the process used for encoding qualitative information,
is performed in this study to analyze the overall
user search behaviors including user task list and
process. In order to uncover the differences and
similarities in user search behaviors, a thematic
analysis framework with a coding scheme is

designed based on an initial observation on user

search behaviors. Participants search behaviors
are captured through video and sound recording,
then analyzed and encoded by using the coding
scheme (see Table 2).
User search behaviors are analyzed at each
Web page the user visited as cognitive behaviors
(thoughts/decisions) and physical behaviors (interactions). Each behavior is encoded using the
preset coding scheme. The result is the sequence
of user tasks performed by each user when searching for specific information on the particular Web
site as described in the scenario they received
(see Table 3).
The results, displayed as the encoded information of user search behaviors, are then further
analyzed and generalized to determine the common patterns of information-seeking tasks that
are associated with each study case (scenario)
by using a time-ordered matrix (Robson, 1993).
The time-ordered matrix is used to systematically
display the encoded information of user search
behaviors in time-ordered sequences by presenting various types of user search behaviors, both
physical and cognitive behaviors, observed in
each Web page from the start to completion of
the task (see Figure 7).
Color coding is also added to help identify
and group the same or similar tasks together.
This enables one to see the overall task list and
its sequence visually and practically in order to
compare the differences and similarities that occur
within and across different scenarios.
In addition, the results gained from thematic
analysis are eventually summarized as procedural
analysis, which presents the common process or
pattern of user search behaviors in each study
case (scenario) including search methods and task
descriptions (see Figure 8).
The encoded information of user search behaviors is also further transformed into Chernoff
Faces (Chernoff, 1973; Wainer & Thissen, 1981)
in order to further identify and compare the common patterns of search behaviors that are associ-

827

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

Table 2. The coding scheme used in thematic analysis

Thematic Analysis: Coding Scheme

Users Search Behaviors:
Cognitive Behaviors (Thoughts, Decisions)
Physical Behaviors (Interactions)
Some information is found and they want to learn
Users keep searching in the current retrieval
more, or want to know about the details of the
results.
retrieval documents.
The intended information is found. Users
primary information needs are fulfilled, but users
Users record the information they found.
are interested in finding other relevant or related
information.
The intended information is found. Users
primary information needs are fulfilled.
Users keep searching by changing search
Users are ready to use information they found to
strategy or search methods.
take further action(s).
The intended information is not found, or not
enough to take further action(s). Users primary
Users go back to the selected (bookmarked) page
needs are not fulfilled. Users need to keep
or result.
searching.
Users make a positive decision (to proceed)
about something according to information they
found.
Users make a negative decision (not to proceed)
about something according to information they
found.
Users are satisfied. All users needs are fulfilled.
Users are able to accomplish their goals based on
the information they found.
Users are somewhat satisfied, but not completely
satisfied. Users primary needs are fulfilled, and
users are able to accomplish their goals based on
the information they found. However, users still
need more information to fulfill all their needs
completely.
Users are not satisfied. Users needs are not
fulfilled. Users are unable to accomplish their
goal(s).

ated with each user goal and mode of searching.

Chernoff Faces are another coding scheme that
is, in this case, used to help identify users search
behaviors holistically with regard to how frequent

828

each behavior occurs, or which behavior occurs

more often than the others.
Chernoff Faces also help visualize the frequency of tasks performed within and across

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

Table 3. An example analysis of thematic analysis by using the pre-designed coding scheme
Information
display on
Web page
page 1
Homepage
Menu bar
Table of
contents
Search field
Recommend
features
page 2
Result Page
Table of
contents
Recommend
products
(small image +
short description)
page 3
Result Page
Small images +
short descriptions

Users key
actions

Users key
speech/thoughts

Users cognitive
and physical behaviors

Well ... I want to look

around first.
Therere lots of
categories to choose
from here, but I think I
should start searching
by Ages.

Lets see if anything

interesting is here.
Visual stimulation
... um ... it sounds
interesting.

Users keep searching in the current

retrieval results.

Some information is found, and they want

to learn more, or want to know about the
details of the retrieval documents.

Users keep searching in the current

retrieval results.

Well ... lets see

whats in it.
Well ... nothing
interesting here.
Maybe I should try
another category to see
if it has more interesting
items.

different scenarios. For example, in this coding

scheme, the face is used to represent the users information-collecting state. The bigger the face, the
more information has been collected by the user.
The eyes represent how much the user searches,
while the eyebrows represent a struggling state
(see Tables 4.1, 4.2, and Figure 9).

Some information is found, and they want

to learn more, or want to know about the
details of the retrieval documents.

The intended information is not found, or

not enough to take further action(s). Users
primary needs are not fulfilled. Users need
to keep searching.
Users keep searching by changing search
strategy.

As shown in Figure 9, Chernoff Face Analysis

reveals the patterns of users prominent tasks performed in each different scenario. For example,
users in scenario 2, as shown in the top row, need
to perform an extensive decision-making task
indicated by the gray and black hair they are all
wearing; in contrast to users in scenarios 4 and 9

829

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

Figure 7. An example use of time-ordered matrix used for further analyzing and generalizing the encoded information gained from the earlier thematic analysis by presenting user search behaviors in the
time-ordered sequences

*Participant N/n means scenario N / participant n.

who all appear with no hair signifying that they

do not perform any decision-making task at all.
The analysis also visually addresses user search
struggle or satisfaction clearly. As seen clearly
in Figure 9, all users in scenario 9 appear with
complete satisfaction while most users in scenario
4 are unsatisfied with their search.

830

In order to identify the patterns of user search

strategies and methods commonly used in each
scenario, a checklist with a sequence record
(Robson, 1993) is designed to record participants
frequency and sequence of use of various search
tools available on each Web site. The recorded
data is then further analyzed to identify the com-

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

Figure 8. An example of procedural analysis used for presenting the process of user search patterns

mon patterns of user search strategies and search

methods primarily used in each scenario, as well
as to compare the differences and similarities of
user search patterns within and across different
scenarios (see Table 5).
Similarly, in order to identify the patterns of
user information needs commonly found in each
study case, another Checklist Record (Robson,
1993) is designed to record the frequency of use
of different kinds of information available on
each Web site. Different types, characteristics,
formats, and presentation methods of information
display that are viewed by users while performing information-seeking tasks are captured by
using the Checklist Record. This process is used
to analyze and identify the main types, characteristics, formats, and presentation methods of
information needed by users to accomplish the

given task within and across different scenarios

(see Table 6).
The analysis of user profiles is collected and
built upon user data acquired through a series
of questionnaires provided to the research participants when the observation took place (see
Figure 10). The questionnaire was designed to
acquire user demographic and techno-graphic data
focusing on different aspects of user experience,
including users prior knowledge and experience
in the specific content they are searching, users
prior experience in the particular or similar Web
site interface, and users general experience in
Internet use for information-seeking purposes.
Furthermore, qualitative comparison is conducted by constructing truth tables (Ragin, 1987,
1994) to help further identify additional insights
and various factors that may reveal additional

831

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

Table 4.1. The coding scheme used in Chernoff Face Analysis

Chernoff Faces Analysis: Coding Scheme
Users cognitive states
and physical states

Chernoff Faces
coding system

Initial searching states

Some information is found, and they want to learn
more, or want to know about the details of the retrieval
documents.

Users keep searching in the current retrieval results.

Information-collecting states
The intended information is found. Users primary
information needs are fulfill, but users are interested in
finding other relevant or related information.
The intended information is found. Users primary
information needs are fulfill. Users are ready use
information they found to take further action(s).
Users record the information they found.
Struggling states
The intended information is found. Users primary
information needs are fulfill. Users are ready use
information they found to take further action(s).
Users keep searching by changing search strategy.

information about user search struggle or success,

and help confirm the results from other analytical
methods (see Table 7). This particular analysis is
important for the research since it looks across

832

all 100 cases simultaneously, in contrast to other

analyses that examine 10 observation cases of
each scenario collectively.

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

Table 4.2. The coding scheme used in Chernoff Face Analysis

Chernoff Faces Analysis: Coding Scheme

Users cognitive states
and physical states

Chernoff Faces
coding system

Decision-making states
Users make a positive decision (to proceed) about something
according to information they found.
Users go back to the previously selected or recorded
(bookmarked) pages or results, and/or compare the selected
pages or results side by side in case there are more than one
page or result selected.

Users make a negative decision (not to proceed) about

something according to information they found.

Satisfactory states
Users are satisfied. All users needs are fulfilled. Users are
able to accomplish their goals based on the information they
found.
Users are somewhat satisfied, but not completely satisfied.
Users primary needs are fulfilled, and users are able to
accomplish their goals based on the information they found.
However, users still need more information to fulfill all their
needs completely.
Users are not satisfied. Users needs are not fulfilled. Users
are unable to accomplish their goal(s).

833

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

Figure 9. Examples of Chernoff Face Analysis used to visually identify various types of user search
behaviors regarding the frequency of each behavior

Validation of Coding System

Reliability and validity of coding schemes specifically designed to use for analysis in this research
is fundamentally important and needs to be
examined before the study proceeds further. In

834

qualitative research, the observers consistency

and bias in interpreting user behaviors and using coding schemes to code events are a primary
concern. As a result, to measure reliability and
validity of the coding scheme and analysis methods, a second observer is invited to independently

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

Table 5. An example analysis of user search strategies and user search methods by using the Checklist
and Sequence Record
Checklist and Sequence Record: Showing the frequency and sequence of use of different search tools
Scenario 2: commercial Web site + making decision goal + existence searching mode

1
2** 5
Participant 2/1*
2
Participant
2/2
3
Participant
2/3
4
Participant
2/4
5
Participant
2/5
6
Participant
2/6
7
Participant
2/7
8
Participant
2/8
9
Participant
2/9
10
Participant 2/10

See more button

Index
Shortcut
Site map
Back button
Next button
7

1 1

4 6

2 3

10

14

11

6
13

15 1

10 2

4
8

0.2 4.8 0.2 0 0 0.8

Central
tendency:
mean

Advanced search function

2
3

Search field
Simple search function

Purposeful searching Auxiliary

Table or diagram w/ link

text

Related items/topics

list of items/topics
Advertising

Feature items/topics

Table of content

Menus

Exploring/Browsing

6.0

1.4 0

0.4

1.8

0 0 0 8 0.9 7.6

16.5

835

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

Table 6. An example analysis of user information needs by using the Checklist Record

Scenario 2
Commercial Web site, making decision goal, existence searching mode

Participant 2/1
Participant 2/2
Participant 2/3
Participant 2/4
Participant 2/5
Participant 2/6
Participant 2/7
Participant 2/8
Participant 2/9
Participant 2/10

Central
tendency
(mean)

836

Facts, scientific information

News, reports

Types
of Information

Opinions, reviews
recommendations

Images, icons, illustrations

Diagrams, maps
Matrix, tables

Textual descriptions

Presentation
Methods
of Information
display

Full or long text/information

Biography references

Short or brief text/information

Abstracts, summaries

Topics, keywords, headlines

Formats
of Information
display

Quick references, Bullet

points
FAQs, questions and answers
Step-by-step instructions
Comparison information
Recommendations, feature
articles
Glossary explanations
Stories, sequence information
Complete descriptions

Characteristics
of Information

Remarks:
Users
thoughts,
comments
on their
information
needs

7 15 9

23

24

25

3 14 8

22

22

22

7 11 11 22

22

22

3 15 7

23

23

24

4 11 15 26

26

25

2 19 11 30

30

YES Many users

would like
23 YES
YES to see more
and bigger
24 YES YES YES pictures,
or some
24 YES
YES interactive
displays
27 YES
YES showing the
usage of
30 YES YES YES product.

29

10

30 6

36

35

36 YES

36

16

16

12

12

27

21 6

27

26

23.8

4.5

2.8 15.4 8.4 23.7 2.8

16

23.6

21 YES

YES Some users

expect to see
18 YES YES YES the same or
similar
15
YES information
to what they
30 YES
YES would see
on the
package of the
product
when they
buy in a
store.
24.8

Most users
want to see
comparison
information
or want a
comparison
tool.

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

Figure 10. An example of analysis of user profiles

interpret and code the same video data. The scripts

encoded by both observers are then compared to
identify the degree to which both observers agree
in their interpretation and coding. This validation
process is called double coding, which is perhaps
the most used technique to attain sufficient reliability to proceed with analysis and interpretation
(Boyatzis, 1998; Miles & Huberman, 1984).
After the double coding process is completed,
the confusion matrix is constructed to show where
the two observers are different in their judgment

when coding the events. Agreement takes place

when both observers use the same code for the
same event. On the contrary, disagreement occurs
when observers use different codes to code the
same event. To read the confusion matrix, the
scores on the diagonal from top left to bottom right
indicate agreement between the two observers,
while the scores off this diagonal indicate their
disagreement (Robson, 1993) (see Figure 11).
As shown in Figure 11, the scores on the
diagonal from top left to bottom right appear-

837

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

Table 7. The construction of truth table 1

Causal conditions

Total
instances
among
100 cases

Output
code:
presence/
absence
of
instance
(P)

Output
code:
achieving
goal
search
success
(S)

1*

B
Have
visited
the Web
site
before
(return
user)?
1

n/a**

12

15

n/a**

22

0
0
0
0

0
1
0
0

0
0
1
0

1
0
0
0

6
0
3
7
= 100
cases

1
0
1
1

1
n/a**
1
0

Have prior
knowledge
and/or
experience
in content?

Utilize
different
kinds of
search
tools?

Read text
or detailed
information
thoroughly?

1*

1*

Achieve
original
goal
search
success?

*Number 1 indicates the presence of a causal condition or an output, and 0

indicates its absence. **Code n/a indicates that the output code for the particular
row is not applicable or it cannot be identified because the instance of the causal
combination on that row is absent.

838

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

Figure 11. An example construction of the confusion matrix showing the scores of agreement and disagreement between two observers in their judgment when coding the events

ing in coding A, B, C, D, and L indicate

agreement between the two observers. However,
for the coding E, there is also a score appearing
off this diagonal which indicates an event of their
disagreement. Note that in this particular case,
both observers do not assign coding F, G, H,
I, J, K, and M to any events. Therefore, there
is no score for unused coding in the matrix; this
will be different from one case to another.
Then, based on the scores on the confusion
matrix, the proportion of agreement, the proportion expected by chance, and the Cohens
Kappa are respectively calculated to measure
inter-observer agreement (see Figure 12). The
inter-observer agreement is the extent to which
two or more observers obtain the same results
when measuring the same behaviors (e.g. when

independently coding the same tape). (Robson,

1993, p. 221).
In order to assess the significance of Kappa
scores, Fliess (1981) has suggested the following
rules of thumb: the Kappa scores of 0.40 to 0.60
is considered fair; the Kappa scores of 0.60 to
0.75 is considered good; and the Kappa scores
above 0.75 is considered excellent. The results
obtained from this validation process show that
validity of the coding schemes used in this study, in
the view of inter-observer agreement, is positively
strong. The Kappa scores of seven observation
cases acquire excellent points (0.75-0.95), and
the other three cases also show good scores
(0.64-0.71).
However, the extent of agreement between
two observers who use the coding scheme to code

839

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

Figure 12. An example calculation of the proportion of agreement, the proportion expected by chance,
and the Cohens Kappa Score to measure the extent of inter-observer agreement

the same events independently is also affected by

some other factors. One primary factor may be
the observers learning curve with regard to the
coding scheme; one observer is more familiar with
the coding scheme while the other observer is new
and still learning to use the codes and/or interpret
the events. Another important factor may be the
observers lack of experience or direct contact
with the actual participants in the prior observation when the events were recorded. This occurs

840

when one observer attended the observation in

person when the events were recorded while the
other observer was absent. The observer who
had experience or direct contact with the actual
participants when the events were recorded will
be able to capture the participants emotions or
thoughts that are hard to detect through watching
video data alone. As a result, the two observers
may interpret the same user behavior differently
since the first observer also makes judgments

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

based on experience with the actual participant.

These are factors that may play an important role
in the inter-observer agreement.

ANALYSIS OF RESEARCH DATA

Patterns of User Search Behaviors
The research provides a new perspective on
design considerations for a Web site by incorporating requirements from both Web site (client) intentions and user goals. The results from
this study in which user goals and their modes
of searching were investigated simultaneously
with Web site goals to reveal common search
patterns associated with each case and significantly show that the patterns of user search
behaviors are uniquely different depending
on their goals and current modes of searching.
Even though each user performed his/her task
in isolation and in his/her own way, similar
search patterns appeared based on a shared goal

and/or the same mode of searching. Different

search patterns were associated with different
user goals and modes of searching, as well as
Web site intentions.
In this research, user search behaviors
are primarily analyzed by using the thematic
analysis (Boyatzis, 1998) with time-ordered
matrix (Robson, 1993) (see Table 3 and Figure
7), along with procedural analysis (see Figure
8), and Chernoff Faces (Chernoff, 1973, Wainer
and Thissen, 1981) (see Figure 9), to uncover the
patterns of user tasks in each scenario, while
Checklist and Sequence Record (Robson, 1993)
(see Table 5) is used to identify the types of user
search strategies and methods. The analyses (see
an example in Figure 7) show that users who
begin with the same goal will perform their
search similarly in terms of what tasks are
necessary to reach the goal. However, if they
use different modes of searching, which depend
mainly on how precisely they know what they
want, they will have different search strategies
and consequently choose different kinds of

Figure 13. User goals and user modes of searching, the main factors regulating user search behaviors and the resulting search patterns

841

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

search methods even though they begin their

search with the same goal. Therefore, based on
these research findings, user goals and modes
of searching are the main mechanisms that play
an important role in determining user behaviors
and the resulting search patterns.
While user goal is the main factor regulating
their task description, user mode of searching
provides the major impact on search strategies
and search methods. User goals determine the
different tasks they need to perform to achieve
their goal. Simultaneously, user modes of searching influence their search strategies, or the plans
of their search, determining how hard or easy the
search can be and how much time is spent on their
search, which accordingly results in selecting
different search methods based on their search
strategies (see Figure 13).
The analyses are collectively done on 10
observation cases of each of 10 scenarios which
are systematically fabricated according to the
research analytic frame previously demonstrated,
in order to uncover the patterns of similarities or
differences of user search behaviors. Based on
research findings, the participants in scenarios 1
and 2 share the same search goal. As a result, even
though they have different modes of searching
and perform their tasks on different Web sites,
the patterns of their task descriptions are very
alike. Likewise the participants in scenarios 3, 4,
and 5, or the participants in scenarios 6, 7, 8, and
9 have different modes of searching on different
Web sites but have very similar tasks. Each group
of these participants who share the same search
goal, perform their tasks similarly although they
do not share the same search modes and they visit
different Web sites.

Patterns of User Search Strategies

and Methods
When performing their search, individuals need
a search plan how to perform their search

842

and which kinds of search tools to use in order

to easily achieve their original goal. This plan is
different from person-to-person based on user
search strategy. However as stated earlier, ones
search strategy is influenced by ones mode of
searching (see Figure 13).
As a result, when visiting a Web site, individuals who arrive with different modes of searching
will form different intentions as well as plan different search strategies, and accordingly perform
their search in different ways to reach the same
goal. Search strategies range from a general or
less objective search (browse) to a more specific
or purposeful search. These are directly proportional to user modes of searching which range
from open-ended to specific search. They may
also plan a fast or slow search based on time
available and the urgency of their need. Users
who plan a slow search usually want to record
their search results by using the bookmark function or simply print out the results of retrieval
pages for later use. Search strategy may change
from time to time in accordance with modes of
searching, which are based on the current context
or situation that unfolds while they are searching
as well as in response to the search results they
find or retrieve.
While search strategy is the scheme that generally characterizes user search patterns, search
method is the procedure for how they actually
perform their search. This concerns the types of
search tools chosen for use in their search. These
search tools may include a menu bar, table of
contents, index, site map, shortcut, search engine,
and so forth. Users select from a combination of
these search tools to perform their search based
on their strategy. For example, users who are in
the mode of exploratory searching (open-ended
search), will likely plan a slow and general search
(search strategy) and explore by browsing the
menu bar and table of contents (search method).
On the other hand, users who are in the mode
of known-item searching (specific search), will

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

usually plan a fast and purposeful search (search

strategy) and comfortably use the index or shortcut
(search method) to pursue their search.
The analyses (see an example in Table 5) show
that users who begin with the same mode of
searching have similar search strategies and
choose similar methods. Based on research
findings, the participants in scenarios 1 and
6 begin their search with the same mode of
searching. As a result, even though they have
different search goals and perform their tasks
on different Web sites, they choose similar
search strategies and methods. Each group
of participants who share the same mode of
searching chooses similar search strategies and
methods although they do not share the same
search goals and they visit different Web sites.
However, note that even though the participants
in scenarios 2, 3, and 7 have different modes of
searching from the participants in scenarios 4
and 8, they also share similar search strategies
and methods.

Patterns of User Information Needs

In this research, user information needs are
primarily analyzed by using the Checklist and
Sequence Record (Robson, 1993) (see Table 6) to
identify the types, characteristics, formats, as well
as quality and quantity of information preferred
or needed by users to fulfill their original goals.
Besides having different search strategies and
methods, users also have different needs for
information that can fulfill their goals. The
findings demonstrate that user goals, modes
of searching, and prior knowledge and experience in the contents they search are collectively
the main mechanisms that play an important
role in determining their information needs.
Consequently, each user who comes to visit a
Web site with a different goal, mode of searching, and prior knowledge and experience will
need different kinds of information in order

to fulfill his/her goal. Information provided

on a Web site may be categorized based on
various aspects of information including the
characteristics of the information; formats and
presentation methods of information display;
types of information; as well as quality and
quantity of given information.
Information characteristics differ widely
including quick reference information such as
short or brief information organized and presented
using bullet points; frequently asked question
(FAQ); glossary or explanation; procedural
guideline or step-by-step instruction; comparison information; recommendation; sequential
story or report; and complete description. Based
on the research findings, user need for different
characteristics of information is influenced by
their different goals.
Types of information can be categorized into
three different groups. The first group includes
information that comes from personal or expert
opinion, critique, review, or recommendation such
as an editors choice or customers review. The
second group may include the information that
is collected from news or reports such as todays
news or weekly reports, and the last group includes
the information that presents the facts or scientific
information for any given topic or item. Similar to
user need for different information characteristics,
the research findings demonstrate that user information needs for different types of information
are influenced by different user goals.
Formats of information display range from
an abstract level, including keyword; topic or
headline; abstract; summary, to the detailed level
including brief/short text or information, full/long
text or information, reference, and bibliography.
Based on the research findings, user needs for
different formats of information display are influenced by different modes of searching. Presentation methods for information display range from
textual to visual presentation, including textual or
descriptive information (text), diagram, matrix,

843

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

table, icon, image, illustration, or combinations

of these methods. The research findings show
that information needs for different presentation
methods of information display are influenced by
the type and character of the site contents.
Quality and quantity of information range
from shallow information, which is usually not
enough to help individuals to take further actions,
to a variety of deep and detailed information.
Based on the research findings, user information
needs for different levels of quality and quantity
of given information are influenced by various
factors related to both the user and the Web site
simultaneously. These factors include user goal;
modes of searching; prior knowledge and experience in the contents they search; as well as the
characteristics of Web contents simple, everyday topics, or complicated, hard topics.

Discussion on Relevant Factors for

User Search Success
Even though the accuracy of a search engine is one
of the most recognized factors determining user
search success or failure, qualitative comparison
conducted by constructing truth tables and applying Boolean algebra method (Ragin, 1987, 1994)
demonstrates that there are other relevant factors
that play an important role to influence user search
success or struggle. These influencing factors
derive from both user profiles and behaviors as
well as Web site attributes.
The qualitative comparison method is used in
this study to examine among cases the combinations of causal conditions that help produce the
positive outcome (users achieve the goal). These
causal conditions include user prior experience in
the content and Web site interface, their behaviors
while searching, and several Web site attributes.
Two truth tables (truth table 1 [see Table 7], and
truth table 2 [see Table 10]) are constructed from
observation data, which is recorded into nominalscale and represented in binary form, to display

844

different combinations of values on the conditions

and outputs.
A presence-absence dichotomy is used in
the tables to specify what outcomes and causal
conditions are either present or absent in each
observation case. Code number 1 indicates the
presence of a condition or an output; code number
0 indicates the absence (see Tables 7 and 10). Truth
table 1, as demonstrated in Table 7, is constructed
to examine the causal conditions of user prior
experience and search behavior and identify the
combinations that significantly contribute to user
search success.
With uppercase letters indicating presence and
lowercase letters indicating absence of a particular
causal condition shown in the combination, the
data on user search success (S) from truth table
1 can be represented in the Boolean equation as
follows:
S = ABCD + ABCd + ABcD + AbCD + AbcD
+ abCD +

AbCd + aBcD + Abcd + abcD + abCd
This equation for S (search success) shows
11 primitive combinations of causal conditions
that help users to achieve their goal. In order to
simplify these primitive expressions, the concept of Boolean minimization is used. The most
fundamental of Boolean minimization rules is
(Ragin, 1987):
If two Boolean expressions differ in only one
causal condition yet produce the same outcome,
then the causal condition that distinguishes the
two expressions can be considered irrelevant and
can be removed to create a simpler, combined
expression (p. 93).
The Boolean minimization process is conducted in a bottom-up fashion until no further
stepwise reduction of Boolean expression is
possible. This process is applied to the primitive

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

Table 8. Boolean minimization process applied to the primitive expressions from truth table 1
Minimization : Step 1
ABCD
ABCD
ABCD
ABCd
ABcD
ABcD
AbCD
AbCD
AbcD
AbcD
abCD
AbCd
AbCd
aBcD

combines with ABCd

combines with ABcD
combines with AbCD
combines with AbCd
combines with aBcD
combines with AbcD
combines with AbCd
combines with abCD
combines with abcD
combines with Abcd
combines with abCd
combines with Abcd
combines with abCd
combines with abcD

Minimization : Step 2

to produce ABC
to produce ABD
to produce ACD
to produce ACd
to produce BcD
to produce AcD
to produce AbC
to produce bCD
to produce bcD
to produce Abc
to produce abC
to produce Abd
to produce bCd
to produce acD

ABC
ACD
ACD
AbC
AbC
AcD
BcD
bCD
bCD

combines with
combines with
combines with
combines with
combines with
combines with
combines with
combines with
combines with

AbC
ACd
AcD
Abc
abC
acD
bcD
bcD
bCd

to produce
to produce
to produce
to produce
to produce
to produce
to produce
to produce
to produce

AC
AC
AD
Ab
bC
cD
cD
bD
bC

Table 9. Prime implicant chart showing coverage of original terms by prime implicants

Prime Implicants

Primitive Expressions
ABCD

ABCd

AC

AD

ABcD

AbCD AbcD abCD AbCd aBcD Abcd abcD abCd

X

Ab

bC

bD

cD
ABD

X
X

X
X

Abd
Acd

X
X

845

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

expressions derived from truth table 1 as demonstrated in Table 8. With the Boolean minimization
process applied, the reduced expressions (prime
implicants) on user search success (S) from truth
table 1 can be represented in the simpler equation
as follows:
S = AC + AD + Ab +bC + bD + cD + ABD +
Abd + Acd
Then, the final step of Boolean minimization
is conducted by using the prime implicant chart
(see Table 9) to map the links between nine prime
implicants (see the second equation previously
shown) and 11 primitive expressions (see the
first equation). This process helps to eliminate
redundant prime implicants in order to produce
a logically minimal number of prime implicants
which cover as many of the primitive Boolean
expressions as possible.
Eventually, with the final process of Boolean
minimization applied, the final equation (S) from
truth table 1 demonstrates six combinations of
causal conditions that produce the positive outcome (user search success) as follows:
S = AC + AD + Ab +bC + bD + cD
This final equation significantly demonstrates
the result showing that causal condition A (users
have prior knowledge and/or experience in the
content), condition C (users utilize different
kinds of search tools), and condition D (users
read text or detailed information thoroughly) are
the important variables that help users to achieve
their goals.
Contrary to the traditional view on user experience with Web site interface (first-time versus
return users), the result shows that causal condition B (users have visited the Web site before)
is not the primary factor contributing to users
accomplishment in their search.

846

In addition, the second truth table (truth table

2) is constructed to examine the impact of various
causal conditions including user prior knowledge
in the contents they search (condition A) and
prior experience in Web interface (condition
B), combined with different causal conditions
from various Web site attributes (see Table 10).
These variables include condition E (Web site
provides different approaches to content classification), condition F (Web site has well-organized
search retrieval results), and condition G (Web
site provides search tips or examples).
With all processes of Boolean minimization
applied, the final Boolean equation (S) from truth
table 2 demonstrates six combinations of causal
conditions that produce the positive outcome (user
search success) as follows:
S = AeF + AEFg + AEfG + BeFg + bEFg +
Abefg
Significantly, this final equation derived from
truth table 2 also confirms that causal condition
A (users have prior knowledge and/or experience
in the content) is the important variable that helps
users to achieve their goals. Besides the condition
A, the Web site variables that have significant
impact on user search success include condition
E (Web sites that provide different approaches
to content classification) and condition F (Web
sites that have well-organized search retrieval
results). The result also shows that condition B
(users have visited the Web site before), and G
(Web sites provide search tips or examples) have
less impact on user search success compared with
other variables. These analytical results as well as
others are further summarized and synthesized,
in order to develop explanatory frameworks of
user search behaviors and needs, as well as to
establish classifications of substantial user factors
and analytical frameworks to evaluate information
design on Web sites.

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

Table 10. The construction of truth table 2

Causal conditions

Total
instances
among
100 cases

Output
code:
presence/
absence of
instance
(P)

Provide Have well Provide

Have prior Have
search
different organized
knowledge visited
tips or
and/or
the Web approaches search
experience site before to content (retrieval) examples?
classifiresults?
in content? (return
user)?
cation?
1*
1
1
1
1
0
1
1
1
0
0
1
1
0
1
0
1
1
0
1
0
0
1
0
0
0
1
0
0
0
0
0

1
1
1
1
0
1
1
1
0
0
1
1
0
1
0
1
1
0
0
0
0
1
0
0
1
1
0
1
0
0
0
0

1
1
1
0
1
1
1
0
0
1
1
0
1
0
1
1
0
0
0
0
1
0
1
1
1
0
0
0
1
0
0
0

1
1
0
1
1
1
0
0
1
1
1
1
0
1
1
0
0
0
1
1
0
0
0
1
0
1
0
0
0
1
0
0

Output
code:
achieving
goal
search
success (S)

1
0
1
1
1
1
0
1
1
1
0
0
1
1
0
1
0
1
1
0
1
1
0
0
0
0
0
0
0
0
1
0

Achieve
original
goal
search
success?
0
5
1
1
0
0
0
0
9
0
0
7
9
0
9
0
7
0
0
5
0
0
0
6
0
3
25
0
0
5
0
8
= 100 cases

0
1*
1
1
0
0
0
0
1
0
0
1
1
0
1
0
1
0
0
1
0
0
0
1
0
1
1
0
0
1
0
1

n/a**
1*
1
1
n/a
n/a
n/a
n/a
1
n/a
n/a
1
1
n/a
1
n/a
0
n/a
n/a
1
n/a
n/a
n/a
1
n/a
1
1
n/a
n/a
0
n/a
0

*Number 1 indicates the presence of a causal condition or an output, and 0

indicates its absence. **Code n/a indicates that the output code for the particular
row is not applicable or it cannot be identified because the instance of the causal
combination on that row is absent.

847

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

CONCLUSION

REFERENCES

This investigation demonstrates that a user-centered approach can improve information design
on Web-based media through study of various
factors, especially user cognitive factors including user goals and modes of searching, to identify
the corresponding impact of these factors on information and functional needs in terms of user
behaviors. As an attempt to solve the problems of
information-seeking tasks in Web-based media,
the research is successful in providing a new
perspective on Web site design considerations
by strongly taking a user-centered approach to
incorporate a careful consideration of actual user
needs and behaviors together with requirements
from a Web site.
By conducting extensive qualitative research
on user study in relation to search needs and
behaviors on Web sites as well as employing
various analytical methods to uncover different aspects of the research data, the study
answers the research questions. The common
patterns of user information-seeking behavior,
user search strategies and methods, as well as
user information needs presented in different
cases are revealed. These valuable findings will
be further synthesized to develop frameworks
and classifications.
Deeper understanding of these various
factors, especially user cognitive factors, may
complement the use of existing analytical
or design methods such as task analysis and
scenario-based design, by helping Web developers to recognize the important factors that
may be subtle or previously unidentified yet
substantially affect user task performances. By
recognizing these elements, Web developers
can identify the useful and appropriate functions and/or information to include in each
particular case, in order to support user needs
and task performances and eventually promote
their satisfaction.

Boyatzis, R. E. (1998). Transforming qualitative

information: Thematic analysis and code development. Thousand Oaks, CA: Sage.

848

Brinck, T., Gergle, D., & Wood, S. D. (2002).

Usability for the Web: Designing Web sites that
work. San Francisco: Morgan Kaufmann.
Chernoff, H. (1973). The use of faces to represent points in k-dimensional space graphically.
Journal of the American Statistical Association,
68, 361-368.
Fleming, J. (1998). Web navigation: Designing
the user experience. Sebastopol, CA: OReilly
& Associates.
Fliess, J. L. (1981). Statistical methods for rates
and proportions. New York: Wiley.
Ivory, M. Y., & Hearst, M. A. (2001). The state
of the art in automating usability evaluation of
user interface. ACM Computing Surveys, 33(4),
470-516.
Miles, M. B., & Huberman, A. M. (1984). Qualitative data analysis: A soursebook of new methods.
Newbury Park, CA: Sage.
Ragin, C. C. (1987). The comparative method:
Moving beyond qualitative and quantitative
strategies. Berkeley; Los Angeles: University of
California Press.
Ragin, C. C. (1994). Constructing social research:
The unity and diversity of method. Thousand
Oaks, CA: Pine Forge Press.
Robson, C. (1993). Real world research: A
resource for social scientists and practitionerresearchers. Malden, MA: Blackwell.
Rosenfeld, L., & Morville, P. (1998). Information
architecture for the World Wide Web. Sebastopol,
CA: OReilly & Associates.

A Qualitative Study in Users Information-Seeking Behaviors on Web Sites

Wainer, H., & Thissen, D. (1981). Graphical

data analysis. Annual Review of Psychology, 32,
191-241.

This work was previously published in Human Computer Interaction Research in Web Design and Evaluation, edited by P.
Zaphiris, pp. 42-77, copyright 2007 by Information Science Publishing (an imprint of IGI Global).

849

850

Chapter 2.28

Developing and Validating a

Measure of Web
Personalization Strategy
Haiyan Fan
Texas A&M University, USA
Liqiong Deng
University of West Georgia, USA

Abstract
An important theoretical undertaking in personalization research is to identify the structure of
the multidimensional construct of personalization
and to operationalize them in measurable terms.
The purpose of this study was to develop and
validate measurement scales for personalization
by identifying four distinctive personalization
archetypes and hypothesizing their respective
relationships with different cognitive and affective
outcomes. This effort was successful in several
respects. New scales for measuring personalization strategies were developed based on the definitions of personalization archetypes (architectural,
instrumental, social and commercial), which
were in turn derived from an extensive review of
multidisciplinary studies. A lab experiment with
229 student subjects was conducted to explore the

structure of the personalization construct and to

validate the instrument using four Web site stimuli
representing four types of personalization strategies. The new measures were found to have strong
psychometric properties and exhibit significant
linkages with respective cognitive and affective
outcomes as theoretically postulated.

Introduction
In e-commerce and mobile commerce, personalization has been recognized as an important
element in customer relationship and Web strategies. Personalization is commonly treated as an
independent variable that influences Web usage
outcomes such as customer experience (Novak,
Hoffman, & Yung, 2000), Web site usability
(Agarwal & Venkatesh, 2002; Palmer, 2002), and

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Developing and Validating a Measure of Web Personalization Strategy

customer churning behavior (Chen & Hitt, 2002).

However, the extant research on personalization
is constrained by the shortage of high-quality
measures for the personalization construct (Fan
& Poole, 2006). Due to the inadequacy of existing
personalization measures to capture all dimensions of personalization consumers actually value
(Chen & Hitt, 2002), empirical studies have often
failed to support the hypothesized association
between personalization and the behavioral outcomes (Chen & Hitt, 2002; Novak et al., 2000).
In responding to the need for better measures
of personalization, this study aims to develop
and validate measures for four archetypes of
personalization.
Personalization can be generally defined as
changing the functionality, interface, information access and content, or distinctiveness of a
system to increase its personal relevance to an
individual or a category of individuals (Blom,
2000, p. 313). While there exist a set of narrow,
specialized definitions of personalization that
are built around varied goals and techniques of
personalization in the literature, Bloms definition
of personalization provides an integrated view of
personalization that centers on the general goal of
increasing personal relevance without limiting the
personalization concept to particular motivations,
context or techniques (Fan & Poole, 2006).
In developing measures for personalization,
we focus on four personalization archetypes: architectural, instrumental, social and commercial
personalization, which are theorized to be four
distinctive personalization strategies that elicit
different cognitive and affective outcomes (Fan &
Poole, 2006). While the existing personalization
categories, such as Rossi et al.s three personalization categories pertaining to link, navigation structure, and navigation context, are largely concerned
with the system-level personalization mechanisms
(Rossi, Schwabe, & Mattos Guimares, 2001; Wu,
Im, Tremaine, Instone, & Turoff, 2003), Fan and
Pooles personalization archetypes provide a highlevel framework for the design of personalization

by delineating different philosophies concerning

user-centered motivations behind personalization
and their corresponding strategies of personalization. Developing valid measures for those four
personalization archetypes will not only facilitate the investigation of cognitive and affective
aspects of using the personalization system, but
also promote the accumulation of personalization
research results on the basis of a common measure
of personalization. In addition to the significant
theoretical value, developing and validating such
measures also has great practical value for both
personalization technology vendors and user
organizations.
In the rest of this article, first we review prior
literature on personalization, followed by a discussion of four personalization archetypes and proposition of hypotheses for their individual influences
on cognitive and affective aspects of Web usage.
Then, we present the method and procedure of
developing the multi-item measurement scales
of the four personalization archetypes. Next, we
discuss the research design of an empirical study
to test and validate the measurement model. After
that, the results of data analysis are reported and
interpreted. Finally, limitations of the study are
addressed along with future directions.

Review of Prior Literature

As an important social phenomenon that brings
considerable economic values (Davenport &
Beck, 2001; Pine & Gilmore, 1999), personalization has been studied extensively in e-commerce
research. Considerable research efforts have been
devoted to classifying different mechanisms for
implementing personalization. Rossi et al. (2001)
made a distinction between base information and
behavior, what the user perceives and how the user
perceives. Rossi et al.s framework mainly deals
with system-level elements such as personalization for links, navigation structure, and navigation
context. Wu et al. (2003) classified personalization

851

Developing and Validating a Measure of Web Personalization Strategy

on e-commerce Web sites into five categories:

control, content, link, customized screen design,
and anthropomorphic personalization. Wu et
al.s personalization categories fit a two-by-two
framework with implicit vs. explicit personalization on one dimension and Web content vs. Web
interface on the other dimension. Terziyan (2001)
defined three levels of personalization: service,
content, and dynamic personalization. Although
these findings have shed light on the practice of
how to do personalization, there is still a lack
of understanding regarding how personalization
can be done well.
To guide the design of personalization systems,
user-centered perspectives of personalization are
needed to develop a vision of what personalization could be, which articulates the goals of personalization and criteria for evaluating the goal
attainment by personalization. In response to
this need, we developed a high level framework
for classifying personalization. This framework
delineates fundamental assumptions and usercentered motivations underlying personalization
in the literature and relates them to design strategies for developing personalization systems. In
addition to its significance for personalization
design, the high-level framework of personalization is built around the motivations and goals of
personalization from a user perspective, and thus
provides an effective way to conceptualize and
operationalize personalization dimensions that
the user values. This has significant implications
for studying the effects of Web personalization
features on customer-related Web usage outcomes.
In the literature of Web personalization, the inadequacy of existing personalization measures
to capture customer-valued dimensions (Chen &
Wells, 1999) has led to mixed findings regarding
the relationships between Web personalization
and customer behaviors, rendering inconclusive
the effects of Web personalization (Chen & Hitt,
2002; Novak et al., 2000).
Our personalization framework differentiates
four general personalization archetypes1 extracted

852

from the personalization literature of several

fields. Representative of four distinct schools of
thought on the underlying motivations and goals
of personalization, the personalization archetypes
not only offer an analysis of possible design choices
for personalization, but also provide a structure
for defining general personalization strategies
and examining their respective influences on user
responses. The next section describes the concepts
and strategies of four personalization archetypes
in the Web site context, as well as the hypothesized relationships between different archetypes
of personalization strategy and the cognitive or
affective aspects of Web usage outcome.

Four Archetypes of
Personalization Strategy
Different schools of thought can be discerned
within the diverse personalization literature.
To capture the characteristic features of these
logically consistent approaches to thinking about
personalization, we conducted a multiparadigm
review of the literature on personalization. During
the review process, we utilized two metatriangulation techniques discussed by Lewis and Grimes
(1999) to uncover paradigmatic disparity and
complimentarily. Metatriangulation is a theorybuilding strategy for exploring the assumptions of
divergent theoretical views and gaining insights
into the multiple paradigms (Lewis & Grimes,
1999). First, we used the paradigm bracketing
technique to differentiate and articulate various
sets of assumptions of alternative paradigms of
personalization. Second, we employed the technique of paradigm bridging to identify transition
zones (Lewis & Grimes, 1999), where paradigmatic boundaries become fuzzy and new views
permeating across paradigms are synthesized.
The sample for the multiparadigm review
includes 86 journal articles, 35 books or book
sections, 13 conference papers, and 8 Web references, obtained from the electronic library

Developing and Validating a Measure of Web Personalization Strategy

databases using the keywords personalization,

variants of the same word stem, and related terms
such as customization, adaptation, individuation, consumer-centric, and one-to-one
relationship. The sample spans six general areas
in which personalization has been studied: marketing/e-commerce, computer science/cognitive
science, architecture/environmental psychology,
information science, and social sciences, including sociology, anthropology and communication.
Utilizing the paradigm bracketing technique, we
distinguished disparate views of personalization
by: 1) conducting a detailed analysis of each personalization study in the sample, 2) identifying
and categorizing the different viewpoints on the
motives or goals of personalization which match
the requisite diversity in the sample, and 3) writing
a paradigm account for each different viewpoint
resulting from step 2.
During step 2 of paradigm bracketing, we relied
on Webers ideal type theory in defining different
views of personalization in the form of personalization archetypes. This conceptual apparatus
Weber defined as the ideal type is an abstraction
of essential features of a particular social or economic phenomenon in the purest form possible
(Ringer, 1997; Rogers, 1969; Weber, 1968). The
ideal type theory is useful for studying personalization for two reasons. First, it is important to
study each of the distinct schools of thought on
personalization in its pure form so as to capture
their respective central characteristics. According
to Weber (1968), an ideal type of personalization
is an abstraction constructed from the accentuation of certain features of reality relevant to one
theoretical view of personalization. The relevant
features are put together into a consistent, logic
construct and separated from other features that
are only visible through other views of personalization (Weber, 1968). The resulting ideal type of
personalization is thus a pure type in a logical
sense where the real indistinctness and ambiguity are set aside and its relevant features are
perfect on logical grounds with the merit of

clear understandability and lack of ambiguity

(Weber, 1968). Therefore, the ideal type theory
provides a methodology for analyzing the typical or logically consistent features of different
views of personalization. Second, there is a strong
need to establish a common frame of reference
against which the current practice of personalization can be evaluated. The ideal types describe
objectively possible courses of action (Rogers,
1969) and can serve as guidelines for conducting
and evaluating personalization systems in light
of alternative approaches.
As a result, four archetypes of personalization
were distilled from the literature: the architectural,
social, instrumental and commercial personalization (Fan & Poole, 2006). Each archetype
represents a different philosophy concerning
the motivation behind personalization and what
personalization tries to accomplish. Each archetype also implies a different strategy for personalization, different means for carrying out this
strategy, and different user modeling techniques.
Finally, each archetype implies different criteria
for evaluating personalization systems.
Architectural personalization is defined as the
construction of the digital environment to create
a pleasant user space and a unique experience for
the user through arrangement and design of digital
artifacts in a way that meets the users needs and
reflects his or her style and taste. Because architectural personalization is concerned with building
digital environments, it relates particularly to
the interface aspect of the system. The goals for
personalization in this view are two-fold: (1) to
create a functional and delightful Web environment that provides aesthetic value and reflects
the users personal style; and (2) to help the user
cultivate a sense of personal and social identity
within the space (Becker, 1977).
Instrumental personalization refers to the
utilization of information systems to enhance
efficiency and personal productivity by providing, enabling and delivering useful, usable,
user-friendly tools in a way that meets the users

853

Developing and Validating a Measure of Web Personalization Strategy

situated needs. Instrumental personalization focuses on the functionality of the system. Its goal
is to support users in accomplishing their goals.
Unlike architectural personalization, in which
function and form balance each other, instrumental personalization emphasizes functionality
and usability and treats aesthetics as a secondary
consideration to be addressed once instrumental
standards are met.
Social personalization can be defined as the
personalization via the mediation of interpersonal
relationships and utilization of relational resources
to facilitate social interactions by providing a convenient platform for people to interact with others
in a way that is compatible with the individuals
desired level of communality and privacy. The
motivation behind social personalization is to
personalize by fulfilling the users needs for
socialization and a sense of belonging. The goal
is two-fold: (1) to enhance the effectiveness of
interpersonal interactions, and (2) to help generate social capital (Wellman, 2002) by providing new opportunities for strengthening social
relationships and maintaining social networks.
Applications amenable to relational personalization vary greatly in size and complexity. They can
be as simple as providing an e-mail to a friend
button to notify others of ones flight schedule
after booking tickets online or as complicated as
a conglomeration of online information portal and
activity center in a Digital City that engages
residents or visitors (Toru, 2002).
Commercial personalization is the differentiation of product, service, and information to
increase sales and to enhance customer loyalty
by segmenting customers in a way that efficiently
and knowledgeably addresses each user or group
of users needs and goals in a given context. One
of the most important human activities is the
consumption of goods and services. The motivation of commercial personalization is to fulfill
users material needs and thus contribute to their
psychic welfare (Douglas & Isherwood, 1979). It
primarily focuses on the content of the system and

854

assumes that product, service and information of

high relevance to the consumer yield a satisfying
shopping experience and loyal adherence to the
Web site, as well as the organization behind it.
There has been much evidence that commercial
personalization, by providing highly relevant
products/services/information to the customers,
promotes long-term customer relationship building and creates high customer retention called Web
site stickiness in the context of e-commerce
(Liu & Arnett, 2000; Zott, Amit, & Donlevy,
2000). The primary method to realize commercial
personalization is segmentation. Commercial
personalization is ultimately effective only to the
extent that the offerings provide value to the target
market segments by differentiating the product,
service and information provided.
After the four personalization archetypes were
identified through paradigm bracketing, we then
conducted paradigm bridging, which involves
the identification of paradigmatic connections,
similarities, and complementarities among the
four perspectives of personalization. A close
examination of the four personalization ideal
types reveals that the four types can be further
classified into a 2x2 grid (see Table 1). First, the
perspectives can be differentiated in terms of their
associations with extrinsic or intrinsic motivation.
The instrumental and commercial personalization
archetypes support extrinsic motivation. They
emphasize task performance and commercial
transactions and hence are employed to enhance
the usability and utility of information system
for specific goal/task achievement; while architectural and social personalization archetypes
attempt to support intrinsic motivation by providing self-fulfilling aesthetic and socio-emotional
values for users to obtain pleasure or enjoyment
from the activity itself.
The four personalization archetypes can also
be differentiated in terms of the basic premise of
use, whether the user primarily engages the system
as an individual or through an interaction. Both
architectural and instrumental personalization are

Developing and Validating a Measure of Web Personalization Strategy

Table 1. Personalization design paradigms

Architectural
form and function

Instrumental
functionality and usability

Social
Meaning

Commercial
content

Intrinsic Motivation

Individual

Interactional

Extrinsic Motivation

concerned with individual use of an artifact, be it

a building, an information system or a Web site.
Design emphasis is on individuals interaction
with the artifact. On the contrary, relationships
among multiple entities and the management of
the relationships are of paramount importance in
relational and commercial personalization.

Effects of Personalization
The personalization archetype scheme implies that
no single standard or approach to personalization
is the best. Each archetype employs different
criteria for evaluating how well the system succeeds in delivering the desired effect. Based on
our paradigm bridging of the four personalization
archetypes, the effects of different personalization archetypes can be differentiated in terms of
the extrinsic or intrinsic motivation they intend
to support.
Commercial and instrumental personalization,
predominantly used for information retrieval,
transaction processing, and content management,
belong to the class of productivity/utility applications. They are utilitarian-oriented, the goal of
which is to achieve specific objectives or tasks.
The idea behind instrumental personalization is
that users will find systems that are designed and

tailored to realize their particular goals or tasks as

more relevant and hence useful. For commercial
personalization, customizable or customized
products/information/services are more useful
because they meet individual needs. Because
utilitarian use is goal-driven and emphasizes
extrinsic motivation, useful content and functionality, as well as usability are given priority in
design. Instrumental personalization that focuses
on providing usable, useful and user-friendly tools
is also expected to relate to the ease of use aspect
of Web site design. Therefore, in conceptualizing
the effects of commercial and instrumental personalization, we adopt the constructs of perceived
usefulness and perceived ease of use in Davis
TAM model (Davis, 1989), which is widely used
to evaluate user perceptions of utilitarian systems
under extrinsic motivation. Although the TAM
model was originally developed to predict the
adoption of workplace technology, it has also been
successfully applied to the context of e-commerce.
Studies have demonstrated the importance of
TAM model in predicting intentions to adopt or
revisit the Web site and online shopping behaviors (Gefen, Karahanna, & Straub, 2003; Gefen
& Straub, 2000; Koufaris, 2002; Shang, Chen, &
Shen, 2005; van der Heijden, 2000, 2003). Even
for the Web site users who are highly computer

855

Developing and Validating a Measure of Web Personalization Strategy

literate, the perceived usefulness and ease-of-use

of a Web site relative to the competing Web sites
may still greatly influence the users intention to
return to that Web site (van der Heijden, 2000).
Therefore, we consider perceived usefulness
and ease-of-use as important outcome variables
of Web personalization, and formally posit the
following hypotheses:
H1: The instrumental personalization will positively influence the perceived usefulness of the
Web site.
H2: The commercial personalization will positively influence the perceived usefulness of the
Web site.
H3: The instrumental personalization will positively influence the perceived ease of use of the
Web site.
While instrumental and commercial personalization archetypes are generally utilitarian-oriented and focus on extrinsic motivation, architectural personalization and social personalization
are closely associated with intrinsic motivation.
The latter two are used primarily for creating an
attractive Web environment, a comfortable, interactive social network, and a sense of psychological
and social well-being. The design focus for this
category centers on the enjoyment and entertainment aspect of the user experience. Hence,
a balance between form and function, as well as
the meaning of using the system, is emphasized.
Architectural personalization affords an intrinsically pleasant experience of creating a personalized digital space through the arrangement and
design of digital artifacts in a way that reflects
the users individual style and taste and meets
the users needs for easy navigation, intelligent
presentation and aesthetic delight. To investigate
the effects of architectural personalization, we employ two intrinsic motivation variables, perceived
enjoyment and ease of use (Atkinson & Kydd,

856

1997). Davis et al. define perceived enjoyment

as the extent to which the activity of using the
computer is perceived to be enjoyable in its own
right, apart from any performance consequences
that may be anticipated (Davis, Bagozzi, &
Warshaw, 1992, p. 1113). Therefore, by definition, perceived enjoyment is largely concerned
with fulfillment of intrinsic motivation, which is
based on emotionally satisfying experience with
the system for its own sake but bears no relation
to task objectives external to the user-system
interaction (Van der Heijden, 2004). Perceived
ease of use has been shown to contribute to both
extrinsic and intrinsic motivations (Atkinson &
Kydd, 1997; Davis, 1989; Lee, Cheung, & Chen,
2005). On the one hand, perceived ease of use
can improve users task performance, which is
critical for extrinsic motivation (Davis, 1989),
and on the other hand, perceived ease of use can
enhance users self-efficacy, which is an important component of intrinsic motivation (Bandura,
1977, 1982; Lee et al., 2005). Due to the focus of
architectural personalization on fulfilling users
intrinsic motivation by providing aesthetically
pleasant, easy-to-navigate Web interface, we
suggest the use of architectural personalization
strategy will improve users perceived enjoyment and ease of use of the Web site. Therefore,
formally we have:
H4: The architectural personalization strategy
will positively influence the perceived enjoyment
of the Web site.
H5: The architectural personalization strategy
will positively influence the perceived ease of use
of the Web site.
The social personalization strategy personalizes ones world by creating a unique Web of
social relationships. It seeks to provide a common,
convenient platform for interpersonal communication and community building that emphasizes design on the basis of what Preece terms

Developing and Validating a Measure of Web Personalization Strategy

sociability (Preece, 2000). The intrinsic value

of sociability is grounded in human motivation
of being altruistic, interconnected, and seeking
acceptance, attachment and care in interpersonal
relationships (Argyle, 1996; Argyle & Lu, 1990).
Positive social relationships give individuals a
sense of well being and a feeling of comfort by
creating support and a sense that they are not alone
and are valued. Building a personalized network
of social relationships is a self-fulfilling socioemotional experience of satisfying the desire for
social interaction, communication, and affiliation
with others having similar interests or goals. There
is mounting evidence of strong social ties built on
social-emotional communications among participants of online communities (Fischer, Bristor, &
Gainer, 1996; Granitz & Ward, 1996). The social
personalization strategy enhances feelings of satisfaction and enjoyment by providing the social
context important for fulfilling users needs for
sociability. This suggests the following:
H6: The social personalization strategy will
positively influence the perceived enjoyment of
the Web site.

Research Method
The goal of this research is to develop and validate
measures for different dimensions of personalization corresponding to the four distilled personalization archetypes. We first distilled the four
personalization archetypes stated above from an
extensive literature review of five general areas
in which personalization has been studied: marketing/e-commerce, computer science/cognitive
science, architecture/environmental psychology,
information science, and social sciences, including sociology, anthropology and communication.
Next, we generated and pretested 20 candidate
measurement items for all four archetypes of
personalization (see Table 2).

We used a lab experiment to test the validity

and reliability of the proposed measurement model
of personalization. We used the personalization
features of actual commercial Web sites as the
experimental stimuli. Because most Web sites
tend to employ a combination of different personalization strategies, for example, Amazon.com
implements both commercial (personalized recommendations) and social personalization (Your
Amazon friends community) strategies, the lab
experiment allows us to measure participants
responses to only one personalization feature of
the Web site at a time. This helps to minimize the
variations among participants responses due to
exposure to mixed personalization features.

Experimental Design
A total of 229 undergraduate students (94 male,
135 female) enrolled in an introductory IS course
were recruited to participate in the study for the
benefit of extra course credit. Because this course
is designed for non-IS majors, the sampling frame
represents a wider range of disciplinary backgrounds than IS or business courses. Majors of
subjects included chemistry, agriculture, liberal
arts, and life sciences. This type of heterogeneity is advantageous in testing the reliability and
generalizability of the measurement instrument
(Shadish, Cook, & Campbell, 2001).
During the experiment, we showed the movie
clips of someone using the personalization features
of Web sites to the participants rather than having
them actually visit the site. We did this for two
reasons. First, having the participants view the
movie clips ensures that they are only exposed to
the personalization features that they are assigned
to. Second, this approach reduces or eliminates
the differences in participants responses resulting
from irrelevant factors such as different browsing
paths participants traversed.
To approximate actual Web browsing in the
movie as closely as possible, we used Camtasia

857

Developing and Validating a Measure of Web Personalization Strategy

Table 2. Scale items for personalization strategy

Construct
Architectural

Code

Questionnaire Item

ARCH1

Personalizing the site creates a delightful Web experience that is unique to me.

ARCH4
ARCH5

I am able to tailor the look and feel of the site to my own personal taste and
style.
Personalizing the site creates a Web environment that is aesthetically pleasing to
me.
I have a sense of control and mastery in creating my own space online.
It feels like decorating my own house when Im personalizing this site.

INSTR1

Personalizing the site makes it more useful and usable for me.

ARCH2
ARCH3

Instrumental

INSTR2
INSTR3
INSTR4
INSTR5
INSTR6
Social

SOCIA1
SOCIA2
SOCIA3

Commercial

SOCIA4
COMM1
COMM2
COMM3
COMM4
COMM5

I found personalizing the site helps me to obtain my goal for using the site more
efficiently.
I would like to see other sites that I frequently view have the functionality this
site provides.
Personalizing the site makes the interaction with the site easier for me.
The site provides many functions that I can configure to suit my own needs.
Personalizing the site helps me locate the right information/product/service I
need.
Personalizing the site helps to connect me to a community that is potentially
interesting to me.
Personalizing the site helps to fulfill my needs for socialization and communication with others.
Personalizing the site gives me a sense of community while maintaining my own
self-identity.
Personalizing the site helps to create a congenial social environment for me.
Personalizing the site enables faster checkout, transaction, or information scanning.
My concern for privacy is a major factor in determining whether I would personalize the site.
Im made aware of new products or useful information regarding sales and promotion that I didnt know before by personalizing the site.
Personalizing the site helps the businesses know me better so that they can serve
me better.
I felt that this site is knowledgeable about my needs and wants in terms of what
they can offer to me.

software to capture the entire course of Web

interactions, dubbed with synchronized human
voice to explain each mouse movement. To prime
participants, we created four scenarios for each

858

Web site describing the typical context and purpose of using the personalization features. It is
important to make the usage context and purpose
explicit to the subjects, as the conceptualization

Developing and Validating a Measure of Web Personalization Strategy

and operationalization of the four personalization strategies are based on the user motivations
and objectives underlying personalization. To
ensure the best audio quality, we tested several
volunteers and chose one with the most pleasant
voice in speaking standard American English. A
total of five AVI files were generated, including
instructions for the participants, and one for each
site. The resolution of the movie clip on screen
is 1080x720.
To optimize the effect of movie viewing, we
conducted the experiment in a state-of-the-art
instructional lab, in which participants viewed
the movie on a personal computer via a 19-inch
flat screen LCD monitor using a headphone. For
practical purposes, balanced incomplete block
design (BIBD) was adopted for the experiment.
To reduce judging fatigue and increase reliability
of evaluations, each participant was randomly
assigned to view only two of the four Web site
stimuli, with order counterbalanced across the
participants. Among the 458 data points collected
from the experiment, 390 were useable after
discarding the questionnaires with incomplete
responses and inconsistent answers to the same
question being asked twice.

Experimental Web Sites

In determining the Web personalization stimuli
for the experiment, we chose to use actual Web
sites rather than creating our own because actual
ones are representative of the tradeoffs designers
will have in creating them to meet actual business
needs and therefore will enhance generalizability
of the experimental results to other commercial
Web site settings. We identified four prominent
Web sites whose personalization strategies have
addressed personal needs of a wide base of Web
users and been advocated as exemplars of Web
personalization in the literature (Amazons book
recommendation system, Lands Ends 3-D virtual model tool, MyMSN information portal, and
Yahoo!Group). Then, we made an interpretive

analysis of those personalization strategies to

match them with the four personalization archetypes (see Table 3). The interpretive analysis
began with an examination of those personalization strategies in relation to the various personal
needs of Web users they are intended to fulfill.
Making explicit the relations between personalization strategies and types of user needs allowed
us to trace back to the motivations underlying
these personalization strategies and identify a
match between each personalization strategy
and personalization archetype. In this way, the
personalization archetypes were respectively
operationalized through the selected Web personalization stimuli.
Amazons book recommendation system was
used to operationalize commercial personalization. The virtual model tool on the Lands End
Web site was used to represent instrumental
personalization. Architectural personalization
was implemented as the customizable information portal on the MyMSN site, where users can
personalize the content and look-and-feel of the
Web page by choosing different themes and
color palates. Yahoo!Group was used to represent
social personalization. It is expected that each
representative sites personalization stimulus will
score significantly different from the other three
sites stimuli and score the highest on the personalization type it represents. In order to verify
the sufficiency of manipulation, we followed the
recommendation of using manipulation check in
the experiment (Shadish et al., 2001). This would
suggest the following criteria for checking the
manipulation of the personalization stimuli:
Criterion 1: MyMSN personalization stimulus
will score higher on architectural personalization
than the other three sites.
Criterion 2: Amazon personalization stimulus
will score higher on commercial personalization
than the other three sites.

859

Developing and Validating a Measure of Web Personalization Strategy

Table 3. Representative sites chosen in the study

Personalization
Archetype

Sample Web
Site

Operationalization
Personalization Strategy

Allowing users to: 1) customize Web

content by selecting desired information to display; 2) change the layout of
Architectural MyMSN porthe Web page by moving modularized
personalization tal Web page
content block to desired position; and
3) design Web aesthetics by choosing
desired theme and color palate.
Allowing users to: 1) create and
maintain newsgroups; 2) communicate
Social personalYahoo!Group through chat room and message board;
ization
and 3) share working space of files and
documents.
Allowing users to: 1) build a customizable 3-D virtual model configured in
Lands End
ones own body parameters; 2) examine
Instrumental
virtual model
instantaneous fitting effect by trying
personalization
tool
clothes on the model; and 3) view
clothes from different angles.

Personalization Motivation

Fulfilling users need for a personalized Web space that meets his/her
information need and reflects his/her
aesthetic taste and style.

Fulfilling users need for social

interactions in ways compatible with
his/her desired level of communality
and privacy.

Fulfilling users need for an online

clothing shopping tool that enables
user to try clothes on before purchasing.

Recommending books to customers

Amazon book
Commercial
based on their past behavior, such as
recommendaFulfilling users need for highly
personalization
past purchases, searches, and browsing
tion system
personally relevant products.
habit.

Criterion 3: Lands end personalization stimulus

will score higher on instrumental personalization
than the other three sites.
Criterion 4: Yahoo! Group personalization stimulus will score higher on social personalization
than the other three sites.

860

Data Analysis and Results

Factor analyses were conducted to assess the
reliability, and the discriminant and convergent
validities of the personalization measures. We first
performed exploratory factor analysis (EFA) on
the pooled response data of all the four person-

Developing and Validating a Measure of Web Personalization Strategy

alization stimuli, and derived four dimensions of

personalization, which were consistent with our
measurement model of personalization. Next,
confirmatory factor analysis (CFA) was conducted
on the response data of each personalization
stimulus separately, to show that the measures
of four dimensions of personalization and the
three outcome constructs held relevant across the
four sites, providing strong evidence of external
validity for the measurement model. In addition,
conducting separate analysis on four sites independently may provide site-specific knowledge
on the relative importance of the constructs
under investigation, given the different business
natures and design features of these sites (Kim,
Lee, Han, & Lee, 2002). Next, a series of planned
contrast analyses are conducted for manipulation
check purpose. Finally, we employed correlation
and regression analyses to test the hypothesized
relationships between personalization strategies
and their cognitive and affective outcomes.

Results: Personalization Strategy

Measures
We first computed reliability coefficients of the
scales using Cronbachs alpha on all 20 items.
Based on each items contribution to reliability,
we discarded two items (SOCIA1 and COMM2
and ARCH1) that showed very low corrected
item-total correlations (0.44, 0.0004, and 0.29,
respectively). Next, we removed two items (Instr5
and Instr3) that exhibited cross loading problems.
After removing items with low reliability and
cross-loading problems, the instrument consisted
of 15 items, shown in the left-most column in
Table 4.

Reliability
Table 3 presents the reliability coefficients in the
right-most column, which are calculated from
the EFA analysis of the data across four sites.

Table 4. Construct loading for personalization strategy instrument from exploratory factor analysis
Latent Construct Loading
__________________________________________________________________________
Scale Items
Instrumental Social
Architectural Commercial
Reliability Coefficient
________________________________________________________________________
INSTR1
0.613
0.8118
INSTR2
0.886
INSTR4
0.517
INSTR6
0.520
SOCIA2
0.724
0.8183
SOCIA3
0.828
SOCIA4
0.748
ARCH2
0.867
0.8129
ARCH3
0.728
ARCH4
0.708
ARCH5
0.743
COMM1
0.559
0.6334
COMM3
0.573
COMM4
0.660
COMM5
0.458
Eigenvalues
5.337
2.519
1.641
1.109

861

Developing and Validating a Measure of Web Personalization Strategy

All coefficients are larger than 0.8, except for

commercial personalization (0.6334).

also obtained acceptable goodness of fit indices (Yahoo!Group: GFI=0.8604, CFI=0.8981,

RMSEA=0.0602, chi-square/DF < 2; MyMSN:
GFI=0.8272, CFI=0.8281, RMSEA=0.0882, chisquare/DF < 2), which supports the validity of the
four measures for these three sites. Lands End, on
the other hand, did not have a good fit. One reason
could be that the instrumental personalization
features present in the experimental stimulus such
as using the avatar and virtual product presence
were not included in the measures of instrumental personalization, and thus were not properly
measured by the current instrument. This result,
however, provided us valuable feedback on how
to improve the instrument.

Convergent Validity
As our theory predicted, four components had
eigenvalues greater than 1. The four columns in
the middle of Table 3 present the factor loadings
for the four constructs of personalization from
the EFA analysis based on data pooled from all
four sites. All 15 items converge well on their
corresponding constructs, with high loadings on
the constructs they are intended to measure and
low loadings on others (factor loadings that are
less than 0.5 are not shown in the table).
Independent CFA analysis (see Table 5) on
each site showed varied indices of goodness of
fit for the measurement model. Among the four
sites, Amazon seems to demonstrate the best
fit (GFI=0.8582, CFI=0.9454, RMSEA=0.0588,
chi-square/DF < 2). Yahoo!Group and MyMSN

Discriminant Validity
Table 6 presents the factor correlations and average
variances extracted (AVE) for the four personalization constructs from the EFA analysis. The

Table 5. Summary of goodness-of-fit of personalization strategy model for the four Web sites

Websites
Amazon
Lands End
My MSN
Yahoo!Group

Confirmatory Factor Analysis

C h i DF
GFI
Square
110.7545
84
0.8582
175.3926
84
0.7911
140.8529
84
0.8272
108.3229
84
0.8604

RMSEA

CFI

NNI

NFI

0.0588
0.1112
0.0882
0.0602

0.9454
0.7857
0.8281
0.8981

0.9317
0.7322
0.7852
0.8726

0.8138
0.6700
0.6768
0.6848

Table 6. Factor correlation and AVE for personalization strategy constructs

Instrumental
Social
Architectural
Commercial

862

Instrumental
0.634
0.195
-0.443
0.404

Social

Architectural

Commercial

0.7667
-0.141
-0.207

0.7615
-0.133

0.5625

Developing and Validating a Measure of Web Personalization Strategy

diagonal elements in Table 5 are the square root

of AVE, and the off-diagonal elements are factor correlations. The four diagonal values are all
larger than 0.5 and also larger than the correlation
coefficients of their corresponding factors with
other factors, which suggests that the measures
have appropriate discriminant validity across
all four sites (Gefen & Straub, 2000). The low
factor correlations between pair-wise subscales
of personalization also indicate the distinctive
dimensionality of the four theoretical archetypes
within the personalization construct.

Results: Manipulation Check

In order to check the representativeness of each
personalization stimulus for its respective per-

sonalization strategy, a series of priori planned

contrast tests were conducted to compare the
performance of the four personalization stimuli
based on their scores on different personalization strategies. Due to the nonorthogonality of
treatment and individual effects in this study
(incomplete block design), we performed a combined intra- and inter-block analysis using SAS
Proc Mixed procedure to estimate the treatment
least-squares means and treatment contrasts. Table
8 shows the contrast coefficients, the personalization type as dependent variable, the t value, and
the significance level for each contrast test.
Contrast 1 (testing criteria 1) compared the
commercial personalization scores between
the Amazon personalization stimulus and the
other three Web sites (Lands End, My MSN, and

Figure 1. Site comparison chart

863

Developing and Validating a Measure of Web Personalization Strategy

Yahoo!Group): Contrast 2 (testing criteria 2) tested

whether the instrumental personalization score for
the Lands End stimulus is higher than the other
three Web sites scores (Amazon, My MSN, and
Yahoo!Group). In contrast 3 (testing criteria 3),
we tested whether the My MSN stimulus scores
higher than the other three Web sites (Amazon,
Lands End, and Yahoo!Group) on architectural
personalization: Finally, in contrast 4 (testing
criteria 4), we compared the social personalization
score for the Yahoo!Group stimulus with those
scores of the other three Web sites (Amazon,
Lands End, MyMSN).
As expected, the results indicate that the
Amazon stimulus scored significantly higher
than the other three Web sites on commercial
personalization; the architectural personalization
score for the My MSN stimulus was significantly
higher than those for the other three Web sites;
and the Yahoo!Group stimulus had a significantly
higher score on social personalization than the
other three Web sites. However, the Lands End
stimulus did not differ from the other Web sites
on the score of instrumental personalization.
Therefore, criteria 1, 3, and 4 were supported by
the planned contrast tests, but criteria 2 was not
met. An examination of Figure 1, which visualizes
the performance of each personalization stimulus
on each of the four personalization archetypes,
reveals that all personalization stimuli except
Lands End personalization stimulus scored significant higher on two personalization archetypes:
the personalization archetype they each represent
and instrumental personalization. The reason
may be that all personalization strategies are to
some extent instrumental in a sense that they all
provide a means to fulfill individuals personal
needs or goals, be it creating an aesthetically
appealing interface, building a personal social
network, or purchasing the products satisfying
an individuals needs. Although Lands End
personalization stimulus did not score higher
on instrumental personalization than the other
personalization stimuli, it still had a significantly

864

higher score on instrumental personalization

than on any of the other three personalization
archetypes, as shown by the results of pairwise
comparisons between its scores on instrumental
personalization and on each of the other three
personalization archetypes (all three P values
< 0.001 after Bonferroni adjustments). Hence,
this confirmed our expectation that Lands End
personalization stimulus would be perceived to
be more instrumental than architectural, social,
or commercial.

Results: Association between

personalization strategy
and outcome
In order to test H1 through H6, we conducted
three multiple regressions, each using mean personalization score as the independent variables
and one of the outcome scores as the dependent
variable. Table 7 shows the results of the multiple
regressions, in which the scores of perceived usefulness, perceived ease of use, and enjoyment of
Web site were respectively regressed on the mean
scores of four personalization dimensions. All
four multiple regression analyses were significant.
We discuss the results related to the hypotheses
testing as follows.
Hypotheses 1 and 2 suggest that instrumental
and commercial personalization will influence
perception of Web site usefulness. The results
of multiple regression 1 show that social personalization (Beta = 0.336, t = 5.562, p < 0.001)
and instrumental personalization (Beta = 0.396,
t = 5.916, p < 0.001) strategies are strong and
significant predictors of perceived usefulness of
the Web site. However, the effect of commercial
personalization was not significant. So, H1 was
supported, but H2 was rejected.
Hypotheses 3 and 5 were supported by the
results of the second multiple regression, which
show that both instrumental personalization (Beta
= 0.279, t = 3.696, p < 0.001) and architectural

Developing and Validating a Measure of Web Personalization Strategy

Table 7. Association

between personalization strategy and outcome

Multiple Regression
F Value
R Square

1
0.395
31.062**

2
0.230
14.158**

3
0.463
40.909**

Usefulness

Ease of Use

Enjoyment

0.061
0.396**
0.039
0.336**

0.116
0.279**
0.194*
0.063

0.168*
0.259**
0.425**
0.064

Beta
Dependent
Variables
Independent
Variables

Commercial
Instrumental
Architectural
Social

* P value < 0.05

** P value < 0.001

personalization (Beta = 0.194, t = 2.727, p = 0.007)

strategies have significant, positive effects on the
users perceived ease of use of the Web site.
Consistent with hypothesis 4, the results of
the third multiple regression indicate that implementing the architectural personalization (Beta =
0.425, t = 7.164, p < 0.001) strategy of a Web site
has strong, significant effect on users feeling of
enjoyment when they interact with the Web site.
We also found that instrumental (Beta = 0.259, t
= 4.109, p < 0.001) and commercial personalization (Beta = 0.168, t = 2.866, p = 0.005) strategies can influence users perceived enjoyment of
using the Web site. However, hypothesis 6 was
rejected because the effect of social personalization on perceived enjoyment of the Web site was
not significant.

Discussion
The purpose of this study was to develop and
validate measurement scales for personalization

by identifying four distinctive personalization

archetypes and hypothesizing their respective
relationships with different cognitive and affective
outcomes of usage. This effort yielded promising results in several respects. New scales for
measuring personalization strategies were developed based on the definitions of personalization
archetypes, which were derived from the review
of extensive multidisciplinary studies. A total
of 25 items initially generated based on the four
archetypes was honed down to 20 items, five for
each personalization type. A lab experiment with
229 student subjects was conducted to explore the
structure of the personalization construct and to
validate the instrument of using four Web site
stimuli, respectively representing four types of
personalization strategies. According to the results
of data analysis, the measurement scales of four
dimensions of personalization exhibited sufficient
psychometric properties in terms of reliability and
validity. Convergent and discriminant validities
were supported by both the analysis of pooled data
across all four sites and the independent analysis

865

Developing and Validating a Measure of Web Personalization Strategy

on each site separately. The study empirically supported our theoretical model of the structure of
personalization construct, which consists of four
distinct dimensions: architectural, instrumental,
social and commercial.
Because of the multidimensional nature of
the personalization construct, we should not use
only a single yardstick to measure the effectiveness of personalization strategies. We identified
two major categories of use outcomes associated
with extrinsic motivation and extrinsic motivation, respectively. To test the relationship between
personalization and extrinsic motivation, we
hypothesized that instrumental and commercial
personalization would positively influence two
TAM constructs: perceived usefulness and ease
of use. To test the relationship between personalization and intrinsic motivation, we hypothesized
that social and architectural personalization would
positively influence perceived enjoyment and ease
of use. Our hypotheses were partially supported
by the empirical results which revealed additional
interesting findings.
Among significant findings, instrumental personalization was found to have a positive influence
not only on perceived usefulness and ease of use
but also on perceived enjoyment. Instrumental
personalization facilitates perceived ease of

use as it is driven by the design philosophy that

emphasizes clarity, consistency, freedom from
ambiguity and error (Norman, 1988). Perceived
ease of use concerns the degree of free of effort
(Davis, 1989), and low attention allocation to the
system interaction, whereas enjoyment relates
to the highly aroused emotions and focal attention allocation. With the presence of ease of use,
enjoyment can be facilitated by the availability
of additional attention resources. The absence of
ease of use may hamper enjoyment because focal
attention has to be allocated to cope with the
system. This leads to our speculation that ease of
use may be the antecedent to enjoyment. Note that
instrumental personalization is the only strategy
that lands itself on all three use outcomes. To some
degree, the result adheres to the TAM prediction
that ease of use mediates usefulness.
Contrary to our hypothesis, we did not find
significant linkage between social personalization
and enjoyment. Instead, the results suggest that
social personalization is associated with perceived
usefulness. We offer two explanations for this
result. First, viewing the movie clip instead of
experiencing the actual social interaction via
the personalized communication platform may
have mitigated the social-emotional impact of
the stimulus on participants. The stimulus may

Table 8. Planned contrast of site scoring on each personalization dimensions

Contrast
1
2
3
4

Personalization
Type
Commercial
Instrumental
Architectural
Social

* P value < 0.05

** P value < 0.001

866

Website
Amazon
3
-1
-1
-1

Lands End
-1
3
-1
-1

My MSN
-1
-1
3
-1

Yahoo!Group
-1
-1
-1
3

T Value
6.40**
0.73
11.81**
17.53**

Developing and Validating a Measure of Web Personalization Strategy

have simply been too weak to elicit the expected

response. Although we are confident in the ability of the movie to simulate the visual and audio
qualities of browsing, as well as the level of detail
of the Web interaction process captured in the
movie, viewing interaction taking place may
not be a perfect surrogate for actually being in
the interaction. Secondly, the result reveals that
social personalization serves some useful goals
external to the socializing process. People engage
in social personalization not just for pure experience only, but for a purpose too. This would be
consistent with the view of relationship as social
capital (Wellman, 2002).
Finally, contrary to our expectation, we did
not find significant association of commercial
personalization with perceived usefulness, but
rather a small association with enjoyment (0.168).
This might be due to the following reasons. On
the one hand, commercial personalization usually requires users to set up personal accounts in
order to facilitate future transactions, which can
be tedious and time-consuming and hence may
not lead to perceived usefulness of the system.
On the other hand, providing personally relevant
products to the customers through commercial
personalization can lead to high levels of satisfaction and evoke a feeling of enjoyment in the
customers.

Conclusion: Theoretical and

practical Contributions
In sum, the main contribution of this research is
two-fold: theoretical and practical. Theoretically,
the study answers the call for conceptualizing the
multidimensional personalization construct. As
previous empirical studies have shown, a onedimensional scale was not sufficient to tap into
the different effects yielded by personalization
construct. We have identified four dimensions
underlying the personalization construct and
developed a measurement scale representing the

four dimensions. The instrument exhibited sufficient psychometric properties on the calibrated
Web site that are representative of respective
dimensions.
In practice, this instrument can be utilized in
at least two ways. Firstly, the instrument we developed in this research can be used as a guideline
for developing Web personalization strategies
because it provided four basic means to personalize Web sites; that is, by providing functionalities
and information specifically needed by the user
so as to enhance efficiency and productivity, by
tailoring the interface to the users own taste and
style, and by enabling interaction and connectivity specifically for the users social network. For
productivity-oriented personalization systems,
key usability issues to consider would be ease
of use, clarity, consistency, free from ambiguity, and error. The aspect of ease of use includes
both the use of the application itself and the setup
and configuration to make personalized features
functional. Consistency helps users better orient
themselves to the site and alleviates cognitive
effort. The enjoyment or entertainment-oriented
personalization applications capitalize on the
process and experience of using the systems.
They are designed to stimulate thinking and to
invoke feelings. The results are not tangible, but
the process itself is critical in creating an engaging, fulfilling user experience. The principle of
consistency may not be sufficient to invoke feelings or engage users on the site for an extended
amount of time. In addition, the instrument can
measure users perception of different personalization strategies and be used as a criterion for
evaluating the effectiveness of the implementation of personalization strategy. For example, at
the design and testing stage, Web designers and
researchers can evaluate the performance of the
site in terms of its ability to cater to users personal
needs by having the users rate the site using the
instrument. Weak scores would indicate potential
areas for improvement.

867

Developing and Validating a Measure of Web Personalization Strategy

Secondly, we empirically tested the hypothesis that different personalization strategies lead
to different cognitive and affective outcomes of
usage. As extant literature shows, measuring the
effectiveness of Web site personalization using a
monolithic method such as Return-On-Investment
or click-to-buy ratio is not sufficient to gauge the
multidimensional nature of personalization. As
Web users come to interact with the site with
different motivations, which largely dictates their
usage expectation and online behavior (Davis et
al., 1992; Venkatesh, 2000), understanding the
underlying motivations for using Web personalization features and how different motivations relate
to respective cognitive and affective outcome is
crucial for realizing the potential of Web personalization. This study is our initial attempt along this
line of research. Drawing on the existing literature
on the distinction between extrinsic and intrinsic
motivations, we empirically tested different usage
outcomes by using different Web personalization
features. Specifically, we found significant and
strong correlations between architectural personalization and perceived enjoyment and ease
of use, and between instrumental personalization
and perceived usefulness and ease of use. The empirical data also suggested significant, small-sized
correlation between social personalization and
perceived usefulness, and between commercial
personalization and perceived enjoyment. Our
initial hypotheses were partially supported and
new insights were obtained.

Limitations
Several limitations should be considered when
interpreting the results of this study. First, the
data were collected from a convenience sample of
students, which may restrict the generalizability
of the results. Although college students are representative of many young Web users and online
consumers in the real world, the participants of this

868

study are heavier Internet users and more frequent

online shoppers than other segments of the general
population. Their perceptions of personalization
may be different from other segments.
Second, having the participants watch the
movie clips instead of letting them interact with the
Web sites freely has some drawbacks. Although
this method allowed us to have a tighter control
over the experiment (e.g., to ensure that the participants are only exposed to the personalization
features that they are assigned to, and to reduce
or eliminate the differences in participants responses resulting from irrelevant factors such as
different browsing paths participants traversed),
watching other people using the Web personalization features may be different from experiencing
the features by oneself. This might introduce error into the data because the environment of the
experiment may not be ecologically representative. In addition, because the conceptualization
and operationalization of the four personalization
strategies are based on the user motivations and
objectives underlying personalization, we also
provided audio instructions to make the usage
context and purpose of the experimental personalization features explicit to the participants.
However, priming the participants to a particular
personalization regimen could produce bias in the
participants responses.
Third, the positivist research approach of this
study introduced further limitations. Positivist
research emphasizes objectivity, which separates
the researcher from the respondent. It is practiced
through standardized process, control of possible
influential variables, and pre-determined structured questions that facilitate measurement. This
approach leads respondents to reply in a restricted
range of ways rather than let them answer the
questions and elaborate on their replies in their
own ways. As a result, the positivist research
may fail to yield an authentic understanding of
the respondents opinions, attitudes, feelings, and
experiences. Therefore, the results of this study

Developing and Validating a Measure of Web Personalization Strategy

may not represent a full insight into the participants view of Web personalization features.
Finally, another limitation of this study is
concerned with the four criteria we developed to
check how well each stimulus represents its corresponding personalization archetype. Because
the focus of this study is to differentiate the four
personalization archetypes and their respective
cognitive and affective effects, the criteria examine each stimulus representativeness of its
corresponding personalization archetype relative
to the other three stimuli. Therefore, a criterion
will be supported if the stimulus of focus is shown
to represent its corresponding personalization
archetype better than the other three (i.e., the
stimulus has a higher score on its corresponding
personalization archetype than the other three).
However, meeting such a criterion may not be
sufficient to establish the stimulus of focus as a
successful operationalization of its corresponding
personalization archetype, because the criterion
evaluates the performance of stimulus using its
relative score rather than its absolute score on the
corresponding personalization archetype.
Despite the abovementioned limitations of this
study, we do believe our results provide valuable
insights on different dimensions of personalization and their influences on users cognitive and
affective responses. This research is the beginning of a rich stream of research investigating the
multidimensional nature of personalization. We
call for further studies along this line of research
using the validated instrument of personalization,
and testing it in various settings.

Future Directions
There are several possible ways of continuing this
research. First, the current study suggested ways of
improving the instrument. Revising the items for
the instrumental personalization by focusing on
the design aspects of instrumental personalization,
such as providing, enabling and delivering useful

personalization tools, is our next step toward a

more valid and stable instrument. The improved
instrument can be used for: 1) stimulus scaling,
that is, to scale Web sites of different domains
(online retailer, sports, information portal, etc.)
and industries (finance, travel, entertainment,
education, government, etc.). The purpose is to
discover patterns of Web personalization strategies within and across domain or industry, which
can be used to inform further improvement. 2)
Subject scaling, that is, to scale Web users along
the dimensions of personalization so that a personal profile based on individuals response will
be created. It would be another tool for business
to further understand their users.
In conclusion, the overarching goal in this
article was to enrich our understanding of the multidimensionality of the personalization construct,
and how different dimensions of personalization
elicit different cognitive and affective user responses. Given that personalization is becoming
an important offering of IT artifact, such research
has value for theory development as well as for
practice. Several avenues for further study are
suggested in hope of a fruitful buildup of a continuous body of personalization research.

References
Agarwal, R., & Venkatesh, V. (2002). Assessing
a firms Web presence: A heuristic evaluation
procedure for the measurement of usability. Information Systems Research, 13(2), 168-186.
Argyle, M. (1996). The social psychology of
leisure. London: Penguin.
Argyle, M., & Lu, L. (1990). The happiness of
extraverts. Personality and Individual Differences, 11, 1011-1017.
Atkinson, M., & Kydd, C. (1997). Individual characteristics associated with World Wide Web use:
An empirical study of playfulness and motivation.

869

Developing and Validating a Measure of Web Personalization Strategy

The DATA BASE for Advances in Information

Systems, 28(2), 53-62.

systems. Journal of Organizational Computing

and Electronic Commerce, 16(3), 179-202.

Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioral change. Psychological

Review, 84(2), 191-215.

Fischer, E., Bristor, J., & Gainer, B. (1996). Creating or escaping community: An exploratory study
of Internet consumers behaviors. Advances in
Consumer Research, 23, 178-182.

Bandura, A. (1982). Self-efficacy mechanism in

human agency. American Psychologist, 37(2),
122-147.
Becker, F. (1977). User participation, personalization, and environmental meaning: Three field
studies. Ithaca, NY: Cornell University.
Blom, J. (2000). Personalization--a taxonomy. In
Paper presented at the CHI 00 Conference on
Human Factors in Computer Systems.
Chen, P., & Hitt, L. (2002). Measuring switching
costs and the determinants of customer retention in Internet-enabled business: A study of the
online brokerage industry. Information Systems
Research, 13(3), 255-274.
Chen, Q., & Wells, W. D. (1999). Attitude toward
the site. Journal of Advertising Research, 39(5),
27-37.
Davenport, T., & Beck, J. (2001). The attention
economy. Boston: Harvard Business School
Press.
Davis, F. D. (1989). Perceived usefulness, perceived
ease of use, and user acceptance of information
technology. MIS Quarterly, 13(3), 319-340.
Davis, F. D., Bagozzi, R. P., & Warshaw, P. R.
(1992). Extrinsic and intrinsic motivation to use
computers in the workplace. MIS Quarterly,
13(3), 319-340.
Douglas, M., & Isherwood, B. (1979). The world
of goods. New York: Basic Books.
Fan, H., & Poole, M. S. (2006). What is personalization? Perspectives on the design and
implementation of personalization in information

870

Gefen, D., Karahanna, E., & Straub, D. W. (2003).

Trust and TAM in online shopping: An integrated
model. MIS Quarterly, 27(1), 51-90.
Gefen, D., & Straub, D. W. (2000). The relative
importance of perceived ease of use in IS adoption:
A study of e-commerce adoption. Journal of the
Association for Information Systems, 1, 1-28.
Granitz, N. A., & Ward, J. C. (1996). Virtual
community: A sociocognitive analysis. Advances
in Consumer Research, 23, 161-166.
Kim, J., Lee, J., Han, K., & Lee, M. (2002). Business as buildings: Metrics for the architectural
quality of Internet business. Information Systems
Research, 13(3), 239-254.
Koufaris, M. (2002). Applying the technology
acceptance model and flow theory to online consumer behavior. Information Systems Research,
13(2), 205-223.
Lee, M., Cheung, C., & Chen, Z. (2005). Acceptance of Internet-based learning medium: The role
of extrinsic and intrinsic motivation. Information
& Management, 42(8), 1095-1104.
Lewis, M., & Grimes, A. (1999). Metatriangulation: Building theory from multiple paradigms.
Academy of Management Review, 24, 672-690.
Liu, C., & Arnett, K. (2000). Exploring the factors
associated with Web site success in the context
of electronic commerce. Information & Management, 38, 23-33.
Norman, D. (1988). The psychology of everyday
things. New York, NY: Doubleday.

Developing and Validating a Measure of Web Personalization Strategy

Novak, T. P., Hoffman, D. L., & Yung, Y. (2000).

Measuring the customer experience in online
environments: A structural modeling approach.
Marketing Science, 19(1), 22-42.
Palmer, J. (2002). Web Site usability, design,
and performance metrics. Information Systems
Research, 13(2), 151-167.
Pine, B. J., & Gilmore, J. H. (1999). The experience economy. Boston: Harvard Business School
Press.
Preece, J. (2000). Online communities: Designing usability, supporting sociability. Chichester,
England: John Wiley & Sons.
Ringer, F. (1997). Max Webers methodology: The
unification of the cultural and social sciences.
Cambridge, MA: Harvard University Press.
Rogers, R. (1969). Max Webers ideal type theory.
New York: Philosophical Library.
Rossi, G., Schwabe, D., & Mattos Guimares, R.
(2001). Designing personalized Web applications. In Paper presented at the Proceedings
of the International WWW Conference, Hong
Kong, China.
Shadish, W. R., Cook, T. D., & Campbell, D. T.
(2001). Experimental and quasi-experimental
designs for generalized causal inference. Boston,
MA: Houghton Mifflin.
Shang, R., Chen, Y., & Shen, L. (2005). Extrinsic
versus intrinsic motivations for consumers to
shop online. Information & Management, 42,
401-413.
Terziyan, V. (2001). Architecture for mobile pcommerce: Multilevel profiling framework. In

Paper presented at the Workshop on E-business

and the Intelligent Web.
Toru, I. (2002). Digital city Kyoto. Communications of the ACM, 45(7), 76-81.
van der Heijden, H. (2000). E-TAM--a revision of
the technology acceptance model to explain Web
site revisits. Research Memorandum, 29.
van der Heijden, H. (2003). Factors influencing the
usage of Web sites: The case of a genetic portal
in The Netherlands. Information & Management,
40, 541-549.
Van der Heijden, H. (2004). User acceptance of
hedonic information systems. MIS Quarterly,
28(4), 695-704.
Venkatesh, V. (2000). Determinants of perceived
ease of use: Integrating control, intrinsic motivation, and emotion into the technology acceptance
model. Information Systems Research, 11(4),
342-365.
Weber, M. (1968). Economy and society. Berkeley,
Los Angeles: University of California Press.
Wellman, B. (2002). Designing the Internet for a
networked society. Communications of the ACM,
45(5), 91-96.
Wu, D., Im, I., Tremaine, M., Instone, K., &
Turoff, M. (2003). A framework for classifying
personalization scheme used on e-commerce Web
sites. In Paper presented at the Proceedings of the
36th Annual Hawaii International Conference on
System Science (HICSS), Big Island, HI.
Zott, C., Amit, R., & Donlevy, J. (2000). Strategies
for value creation in e-commerce: Best practice in
Europe. European Management Journal, 18(5),
463-475.

This work was previously published in the International Journal of Technology and Human Interaction, edited by B. C. Stahl,
Volume 4, Issue 4, pp. 1-28, copyright 2008 by IGI Publishing, formerly known as Idea Group Publishing (an imprint of IGI
Global).

871

872

Chapter 2.29

User Interface Formalization in

Visual Data Mining
Tiziana Catarci
University of Rome La Sapienza, Italy
Stephen Kimani
University of Rome La Sapienza, Italy
Stefano Lodi
University of Bologna, Italy

Abstract
Despite the existence of various data mining efforts that deal with user interface aspects, very
few provide a formal specification of the syntax
of the interface and the corresponding semantics.
A formal specification facilitates the description
of the system properties without being concerned
about implementation details and enables the
detection of fundamental design issues before
they manifest themselves in the implementation.
In visual data mining, a formal specification can
enable users to decide which interaction/operation
to apply to get a desired result; help users to predict
the results of their interactions/operations with the
system; and enable the development of a general
interaction model that designers/developers can
use to understand the relationships between user
interactions and their compositions. In this work,

we describe an approach for formalizing the visual

interface of a core data mining system, which has
been employed in the development of a visual data
mining system named VidaMine.

INTRODUCTION
In this day and age, data still present formidable
challenges to effective and efficient discovery of
knowledge. It should be acknowledged that a lot
of research work has been and is being done with
respect to knowledge discovery (KD). Much of
the work has concentrated on the development
and the optimization of data mining algorithms
using techniques from other fields such as artificial intelligence, statistics, and high performance
computing (Fayyad, Piatetsky-Shapiro, & Smyth,
1996b). Besides various glaring issues (such as

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

User Interface Formalization in Visual Data Mining

the need to have an overall framework that can

support the entire discovery process, supporting
human involvement in the entire process, etc),
Mannila observes that relatively little research
work has been published on the theoretical foundations of data mining (Mannila, 2000). On the
same note, although there are many data mining
efforts that deal with user interface issues, very
few efforts provide or give a precise definition
of the syntax of the user interface and the corresponding semantics.
Formal specifications enable the description
of the system properties without having to be
concerned about implementation details. The
system properties are often specified using a
precise notation. The specification can be used
to construct models of the system. Formal methods make the analyst think abstractly about the
problem at hand and the corresponding system
thereby exposing fundamental design decisions
well in advance before they manifest themselves in
the implementation of the system. While it is true
that the formal specification should not determine
the programming aspects (e.g., algorithms and
data structures), it should describe the behavior
of the system in a precise or rigorous manner.
Moreover, with a formal specification it is possible
to transform a system model while preserving
important properties of the model. In practice, a
formal approach in which each and every design
decision is proven to be a correct refinement step
is rarely performed due to the high costs involved.
However, substantial application of refinement
does considerably improve the understanding
of the design process. A formal specification
of the visual interface of a data mining system
can facilitate the gathering of information about
the most useful usage patterns, which can then
be used to guide the design and layout of user
interfaces for visual data mining. Moreover, a
formal specification of the visual interface can
facilitate automated (and/or objective) evaluation
of the usability of the user interface of a visual
data mining system.

In information visualization, various specifications/models for characterizing visualization

aspects have been proposed such as (Baudel, 2004;
Chi & Riedl,

1998; Chuah & Roth, 1996). In fact,

it does turn out that most of the efforts that are
related to our work are mainly found in information visualization and exploration efforts rather
than in core data mining. Some of the benefits
of specifications/models such as the foregoing
do apply to visual data mining as well, where
visualization tends to be a key ingredient. Consequently and borrowing from Chi et al. (1998), a
similar formal specification in visual data mining
can: enable users to decide which user interaction/operation to apply in order to get a desired
result; help users to predict the results of their interactions/operations with the visual data mining
system; and enable the development of a general
interaction model that designers/developers can
use to classify and understand the relationships
between user interactions and the composition
of interactions. In fact, such a model could help
eliminate errors caused by other imprecise or
incorrect models. In this work, we describe an
approach for formalizing the visual interface of a
core data mining system. The proposed approach
has been employed in the development of a visual
data mining system named VidaMine.

BACKGROUND
Knowledge Discovery
Knowledge discovery (KD) may be defined as
the process of identifying valid, novel, potentially
useful, and ultimately understandable models
and/or patterns in data (Fayyad, Piatetsky-Shapiro, Smyth, & Uthurusamy, 1996a; Fayyad et al.,
1996b). On the whole, the knowledge discovery
process may be defined as an interactive and
iterative non-trivial process that entails various
phases as seen in Figure 1.

873

User Interface Formalization in Visual Data Mining

Figure 1. The knowledge discovery process

The KD phases include carrying out some

initial planning (understanding the application
domain, relevant prior knowledge, and goal/s of
the user), data integration, selection of target data,
data cleaning and pre-processing, data reduction
and transformation, selection of suitable data mining techniques to support the discovery process,
and evaluation, presentation, and interpretation
of results. Through carrying out the phases, the
KD process intends to find a subset of results
that may be considered as new knowledge
(Fayyad et al., 1996a, 1996b). KD is of interest to
researchers in many research disciplines such as
machine learning, pattern recognition, databases,
statistics, artificial intelligence, expert systems,
and information visualization.

Data Mining
Data mining is a core step in the knowledge
discovery process that, under acceptable computational efficiency limitations, enumerates
models and patterns over the data (Fayyad et

874

al., 1996a, 1996b). It is worth recalling that the

knowledge discovery process is as an interactive
and iterative non-trivial process that entails: carrying out some initial planning (understanding
the application domain, relevant prior knowledge,
and goal/s of the user), data integration, selection
of target data, data cleaning and pre-processing,
data reduction and transformation, selection of
suitable data mining techniques to support the
discovery process, and evaluation, presentation,
and interpretation of results from which there
exists a subset that may be considered as new
knowledge (Fayyad et al., 1996a, 1996b). Data
mining methods include clustering, classification, regression, characterization, dependency
modeling change and deviation detection, and
pattern-based similarity matching.
The primary goals of data mining are verification and discovery. The verification goal aims
at validating some hypotheses based on specific
user needs. The user generates a series of hypothetical patterns and relationships (assumptions
or claims). The user then formulates and issues

User Interface Formalization in Visual Data Mining

queries to the system (actually to the data itself)

to verify (or disprove) the claims. The discovery
goal involves finding new patterns or discovering new knowledge. Rather than verify hypothetical patterns, the goal here is to use the data
itself to uncover or identify such patterns. Such
methods of discovery may be initiated based on
the guidance of a user to analyze a certain domain through a predetermined perspective or by
automated learning. Discovery can be predictive
or descriptive. Prediction entails foretelling
unknown or future values of the same variables
or other variables of interest whereas description
involves getting an interpretation/understanding
of the data. Classification, regression, and time
series models are primarily useful for prediction.
Clustering, association, and sequence discovery
models are primarily useful for description of the
behavior that is captured in the data.

Rules
Metaqueries
Metaquerying (Mitbander, Ong, Shen, & Zaniolo, 1996) is a data mining technique that
is especially useful in mining relational and
deductive databases. Metaqueries (or metapatterns) provide a generic description of a class of
patterns that the user may want to discover from
the underlying dataset. With metaqueries, it is
possible to mine patterns that link several tables
in the target dataset. Metaquery specification
can be carried out manually (for instance by an
expert user). Alternatively, the specification can
be automated by exploiting the schema of the
underlying dataset.
Let U be a countable domain of constants.
A database DB is (D,R1, ....,Rn) where D U is
finite, and each Ri is a relation of fixed arity a(Ri)
such that Ri Da(Ri).
A metaquery is a second-order template of
the form (Angiulli, Ben-Eliyahu-Zohary,

Ianni,
& Palopoli
, 2000):

Equation 1
T L1, ..., Lm
where T and Li are literal schemes. Each literal
scheme T or Li is of the form Q(Y1, ..., Yn) where Q
is either a predicate (second-order) variable or a
relation symbol, and each Yj (1 j n) is an ordinary
(first-order) variable. If Q is a predicate variable,
then Q(Y1, ..., Yn) is called a relation pattern of
arity n, otherwise it is called an atom of arity n.
The left-hand side T is called the consequent or
the head of the metaquery. The right-hand side
L1, ...,Lm is called the antecedent or the body of
the metaquery. Consider the relations CustCent,
ClustOutl1 and ServCent with the following
attributes: CustCent.CustID, CustCent.CentID,
ClustOut1.CustID, ClustOut1.ServID, ServCent.
ServID and ServCent.CentID. The following is an
example of a corresponding metaquery:
Equation 2
CustCent(CustID, CentID) {ClustOut1 (CustID, ServID), ServCent(ServID, CentID)}
Intuitively, given a database instance DB,
answering a metaquery MQ on DB amounts to
finding all substitutions of relation patterns
appearing in MQ by atoms having as predicate
names relations in DB such that the Horn rule
(MQ) (which is obtained by applying to MQ)
encodes a dependency between the atoms in its
head and body. The Horn rule is supposed to hold
in DB with a certain level of plausibility/relevance.
The level of plausibility is based on measures of
interestingness such as support and confidence.
The measures of support and confidence are
described in Section Support and Confidence.
Metaqueriers have been applied in the telecommunication industry, in a common-sense knowledge base, and in the chemical industry (Leng &
Shen, 1996). Metaqueries have also been applied
in analyzing time sequence data for semiconductor

875

User Interface Formalization in Visual Data Mining

process control and fault-detection (Shen, Leng,

& Chatterjee, 1995).

Association Rules
Association rules were introduced in Agrawal,
Imielinski, and Swami (1993). Association rules
represent a data mining technique that is used to
discover implications between sets of items in
the database.
Let I = I1, I2, ..., Im be a set of data items or
literals and D a set (or database) of transactions,
in which each transaction T is a set of items from I
(i.e. T I). Each transaction (T) is assigned some
unique identifier, TID.
Let X I and Y I. A transaction T is said
to contain X if X T. An association rule is an
implication of the form:
Equation 3
YX
where X Y = . The left-hand side, Y, is the
consequent or the head of the association rule
whereas the right-hand side, X, is the antecedent
or the body of the association rule.
The problem of mining association rules is
to generate all association rules with a degree of
relevance/interestingness that is greater than a
certain minimum (such as user-specified) value.
The problem of discovering all association rules
can be decomposed into two sub-problems
(Agrawal et al., 1993):
1.

2.

876

Finding all sets of items (itemsets) that have

support above the minimum support. The
measure of support is described in Section
Support and Confidence. Itemsets with
minimum support are called large itemsets
and all others small itemsets.
Using the large itemsets to generate the
desired rules. The desired rules are those

whose confidence is greater or equal to

the minimum confidence. The measure of
confidence is described in Section Support
and Confidence.
Computing the solution to the first sub-problem
naively is very expensive, and efficient techniques
to reduce its complexity have been researched
extensively (e.g., Agrawal et al., 1993; Agrawal &
Srikant, 1994). Solving the second sub-problem
is comparatively much easier, since it amounts
to computing, for every frequent itemset, the
confidence of the association rules obtained by
picking one item as the consequent of the rule
from the frequent itemset.
Association rules are valuable and readily
applicable in many areas, including marketing,
catalog design, business management, decisionmaking, add-on sales, mailing, customer segmentation, and store layout. For instance, a superstore
dealing in clothing may find that customers who
buy ties also tend to buy shirts at the same time
(same market basket). Consequently, the following would be an association rule with fairly
high measures of interestingness:
Equation 4
"shirt" "tie"

Support and Confidence

The relevance of a rule may be determined by
measures of interestingness. In general, measures
of interestingness measure the overall value of a
rule with respect to utility, novelty, certainty, and
simplicity. In the mining of metaqueries and association rules, there are two common measures
namely confidence and support.
Let D be the target dataset. The support of a
rule in D is the fraction or percentage of tuples
in D that contain the union of the antecedent and
the consequent. For instance, the association rule

User Interface Formalization in Visual Data Mining

in
Equation 3
has support supp in D if supp% of
transactions in D contain X Y . The confidence
of a rule in D is the fraction or percentage of tuples
in D containing the antecedent, that also contain
the consequent. For instance, the association rule
in
Equation 3
has confidence conf in D if conf%
of transactions in D that contain X also contain Y.
Intuitively, support indicates how frequently the
items in the rule occur together in the transactions of the database, and therefore represents the
utility of the rule, whereas confidence indicates
the strength of the implication represented by
the rule.

Clustering
Clustering is a process through which the target
dataset is divided into groups of similar objects,
such that the objects in a particular group are
dissimilar to objects in other groups. Each such
group is referred to as a cluster. Clustering is
applicable in many arenas such as in analyzing
astronomical data, in demographics, in insurance,
urban planning, and Web applications.

Classification of Clustering Methods

Traditionally, clustering methods have been
classified into a taxonomy having two broad
groups: hierarchical and partitional (Jain, Murty,
& Flynn, 1999).

Hierarchical Clustering
Hierarchical methods produce a sequence of
nested partitions. A compact way to represent
nested partitions is by a dendrogram, i.e., a tree
having single objects as leaves, showing the hierarchical relationships among the clusters. It is
therefore possible to explore the underlying dataset at various levels of granularity. Hierarchical
methods are further subdivided into agglomerative
and divisive (Jain & Dubes, 1988; Kaufman &
Rousseeuw, 1990).

Agglomerative (bottom-up) clustering starts

with singletons, i.e., with each cluster containing
exactly one point. The clustering then recursively
merges two or more most appropriate clusters.
The process goes on until a stopping criterion is
fulfilled (such as the number of clusters input by
the user). Examples of agglomerative algorithms
include CURE (clustering using representatives)
(Guha,
Rastogi, & Shim
, 1998), and CHAMELEON (Karypis, Han, & Kumar, 1999).
Divisive (top-down) clustering starts with one
single cluster of all objects, and recursively subdivides the most appropriate cluster. The process
goes on until some criterion is met. The PDDP
(principal direction divisive partitioning) (Boley,
1998) algorithm is an example of divisive algorithms. Also in this category of divisive clustering
are approaches based on the k-means algorithm
(Hartigan, 1975; Hartigan & Wong, 1979) such
as the bisecting k-means algorithm (Steinbach,
Karypis, & Kumar, 2000; Wang, Wiederhold,
Firschein, & Wei, 1998).

Partitional Clustering
Partitional methods attempt to identify clusters
directly either by iteratively relocating points
between subsets, or by associating clusters with
the areas that are densely populated with data.
Consequently, partitional methods fall into two
categories: relocation methods and density-based
methods.
Relocation methods focus on how well points
fit into their clusters. Such methods intend to
ensure that the built clusters have the proper
shapes. Relocation methods are further subdivided
into probabilistic, k-medoids, and k-means. The
probabilistic clustering model is based on the
assumption that data has been independently
drawn from a mixture model of several probability distributions. The results of probabilistic
clustering are often easy to interpret. Probabilistic
clustering algorithms include SNOB (Wallace
& Dowe, 1994), AUTOCLASS (Cheeseman &

877

User Interface Formalization in Visual Data Mining

Stutz, 1996), MCLUST (Fraley & Raftery, 1999).

In clustering methods that adopt the k-medoids
approach, a cluster is represented by one of its
points. When the medoids are selected, clusters
are considered to be subsets of points close to
respective medoids. Algorithms based on the
k-medoid approach include PAM (partitioning
around medoids) and the algorithm CLARA (clustering large applications) (Kaufman et al, 1990),
CLARANS (clustering large applications based
upon randomized search) (Ng & Han, 1994). In
k-means (Hartigan, 1975; Hartigan et al., 1979),
a cluster is represented by its centroid, which is a
componentwise arithmetic mean of points within
a cluster. Although the k-means algorithm does
not work well with a dataset that has categorical
attributes, the algorithm is an appropriate choice
for datasets with numerical attributes.
Density-based methods aim at identifying
connected components/areas in the dataset that are
dense with data. In this respect, a cluster therefore
corresponds to a connected dense component.
Density-based methods can be further divided into
two main categories: density-based connectivity
and density functions. Density-based connectivity approach reduces density to a training data
point. Algorithms that use the approach include
DBSCAN (Ester, Kriegel, Sander, & Xu, 1996),
OPTICS (ordering points to identify the clustering
structure) (Ankerst, Breunig, Kriegel, & Sander,
1999), DBCLASD (distribution based clustering
of large spatial databases) (Xu,
Ester, Kriegel, &
Sander
, 1998). Density functions approach reduces
density to a point in the attribute space. DENCLUE
(Hinneburg & Keim, 1998) is an example of an
algorithm based on density functions. In fact,
DENCLUE is a blend of density-based clustering
and grid-based preprocessing.

Miscellaneous
There exist many other clustering techniques that
do not fit well in one of the foregoing categories.

878

For instance grid-based techniques, co-occurrence techniques, etc.

Grid-based techniques work indirectly with
data by constructing summaries of data over the
attribute space subsets. They segment the space
and then aggregate appropriate segments. On the
one hand, grid-based methods often use hierarchical agglomeration as a phase in their processing.
Algorithms that use this approach include BANG
(Schikuta & Erhart,

1997),

STING (statistical
information grid-based method) (Wang, Yang, &
Muntz, 1997), and WaveCluster (Sheikholeslami,
Chatterjee, & Zhang, 1998). On the other hand,
the idea behind grid-based methods is exploited
by other types of clustering algorithms (such as
CLIQUE (clustering in quest) (Agrawal, Gehrke,
Gunopulos, & Raghavan, 1998), MAFIA (merging
of adaptive finite intervals) (Goil, Nagesh, & Choudhary, 1999; Nagesh, Goil, & Choudhary, 2001) as
an intermediate phase in their processing.
Co-occurrence techniques are meant to handle
special requirements when it comes to clustering categorical data. Algorithms ROCK (Guha,
Rastogi, & Shim
, 1999), SNN (shared nearest
neighbors) (Ertz, Steinbach, & Kumar, 2003),
and CACTUS (clustering categorical data using
summaries) (Ganti,
Gehrke, & Ramakrishnan,
1999).

The Proposed Clustering Taxonomy

and Framework
The foregoing traditional categorization of clustering methods into two broad groupshierarchical
and partitional (Jain et al., 1999)is technically
sound and relevant to various application domains.
However, such categorization does not highlight
similarities and differences between the various
definitions of a cluster that are implicit in the
methods. For instance, Wards minimum-variance method (Ward, 1963) and the PAM method
PAM (partitioning around medoids) (Kaufman
et al., 1990) are similar. However, the former is
hierarchical whereas the latter is partitional.

User Interface Formalization in Visual Data Mining

As an alternative to the foregoing traditional

approach of categorizing clustering methods,
clustering can be regarded as an optimization
problem, in which the function to be optimized
is a mathematical measure of homogeneity or
separation (Hansen & Jaumard, 1997). Such a
perspective enables one to categorize clustering
methods according to a taxonomy of homogeneity or separation functions. Therefore, such a
perspective provides recourse for categorizing
clustering methods.
Moreover, such taxonomy expresses cluster
definitions in an implicit manner. Such categorization is most likely more effective in capturing different behaviors in practice. It therefore provides
a more natural avenue for the process of selecting
a clustering algorithm, which is most suited to a
particular application or domain.
In this research work, we therefore adopt this
perspective and to the best of the research survey,

this is actually the first effort that uses a uniform

framework for selecting clustering algorithms.
More details about the same and how it has been
applied in our design and implementation efforts
will be discussed in Section Formal Specification
of the Visual Interface.

USER INTERFACE FORMALIZATION

IN VISUAL DATA MINING
Introduction to VidaMine
VidaMine1 is a visual data mining system that
exploits various visual strategies thereby offering
a consistent, uniform, flexible visual interface
that allows or enables the user not only to process
data, but also to steer, guide, or direct the entire
process of data mining (DM) (Kimani, 2002).
We adopted a user-centered user interface design,

Figure 2. An illustration of the overall outlook of the interface. Reproduced from Kimani, Lodi, Catarci,
Santucci, and Sartori (2004) with the permission of Elsevier B. V.

879

User Interface Formalization in Visual Data Mining

equipped with usability studies. As reported in

Kimani,
Catarci, & Santucci (
2003), we employed
various usability methods progressively in the
development lifecycle. The visual interface of
VidaMine offers visual interaction environments
across different mining techniques and tasks. At
present, the system offers visual environments
for mining metaqueries, performing clustering,
and mining association rules. Figure 2, which is
for the metaquery environment, is an illustration
that shows the overall outlook and feel of the
visual interface.

Formal Specification of the Visual

Interface
In developing a formal specification for the visual interface, we propose the provision of two

Figure 3. The place of the abstract and operational

specifications in VidaMine

880

specifications: an abstract formal specification

that is defined for each of the mining methods
currently supported by VidaMine and a corresponding operational specification that is defined
and exploited for implementation purposes. The
place of the two specifications with respect to the
system can be seen in Figure 3.

Abstract Formal Specification of the

Visual Interface
In this area, we started by considering the definition
an abstract syntax and a formal semantics for the
visual interface. The abstract syntax is intended
to provide a snapshot of the visual environment
in terms of static and compact structures. This
visual syntax reduces distance to mathematical
objects (such as some specific data mining functions). We intend to describe an abstract formal
specification of the visual interface for each of the
visual interaction environments corresponding to
each of the data mining methods currently supported by VidaMine aka metaqueries, clustering,
and association rules.
In order to define a semantics for a supported
data mining method, it is necessary to first define
the visual language supported by the corresponding visual interaction environment (user interface).
An abstract syntax for a visual language can be
defined in terms of multi-graphs as follows (Erwig,
1998). Let , be sets representing label types.
A directed labeled multi-graph of type (, ) is a
quintuple G(V,E, l, ,) consisting of finite sets of
nodes V and edges E where l : E V V maps
every edge to the pair of nodes it connects, : V
maps every node to its label, and : E maps
every edge to its label. A visual language of type
(, ) is a set of directed labeled multi-graphs of
type (, ). Intuitively, visual controls are the
nodes of the multi-graph, whereas the spatial
relations between visual controls are the edges of
the multi-graph. Since many of the visual controls
appearing in the user interface are standard (such
as check buttons, radio buttons, combo boxes, spin

User Interface Formalization in Visual Data Mining

boxes, edit boxes, and sliders), their geometric

properties will be ignored and simply use an
abstract value to represent each visual control.
Based on the foregoing notion of abstract syntax,
we will define the visual language for each of the
three visual interaction environments, and then
describe the corresponding semantics.

Clustering
The clustering environment in VidaMine provides
various visual widgets for specifying or selecting
parameters characterizing a clustering task. The
parameters include a fixed number of clusters or a
measure (of homogeneity, separation, or density);

Figure 4. The clustering environment: Input.

Reproduced from Kimani et al. (2004) with the
permission of Elsevier B. V.

attributes that will be directly involved in cluster

analysis; and supplementary attributes (e.g., for
labeling cases in the output). Specifying each
such parameter may also involve more specific
options/settings. The corresponding environment
is seen in Figure 4.
In clustering, it is virtually impossible to apply uniformly a single technique to uncover the
variety of structures present in multidimensional
data sets (Jain et al., 1999). In most data mining
applications, the user has to evaluate a set of techniques and then, based on experience and domain
knowledge, select possibly more than one data
mining technique to apply on the target dataset.
A visual data mining system aimed at supporting
the user during the entire mining process might
be expected to provide support to the broadest
range of clustering techniques possible. While it
is true and commendable that VidaMine has an
extensible framework, it is worth observing that
adding one extension module for every clustering
technique would render the system hard to utilize
to anyone but the most informed practitioner,
due to the variety of parameters and terminology. Therefore, in VidaMine, the approach is to
provide and support three generalizations from
the commonest types of clustering techniques.
The three clustering generalizations are:

Table 1
.
Definitions of the main sets and functions
Main Sets and Functions
Dataset S = {Oi | i = 1, 2, . . . , N}
Symmetric dissimilarity function diss : S S
R+
Classification C of S is a subset of a partition of S
Accuracy function m is a function on the set of all
classifications of S to R+

881

User Interface Formalization in Visual Data Mining

1.
2.
3.

Clustering based on homogeneity.

Clustering based on separation.
Clustering based on density.

Here are some common terms that will be

used while describing the foregoing clustering
generalizations: Consider a dataset S = {Oi | i
= 1, 2, . . . , N}, and a symmetric dissimilarity
function diss : S S R+. It is worth noting
that classification C of S refers to a subset of a
partition of S. Let an accuracy function m be a
function on the set of all classifications of S to
R+. The foregoing main sets and functions are
summarized in Table 1.
The justification for adopting the three clustering generalizations (or categorization) is highlighted in the descriptions of the generalizations.
In the sequel are the descriptions of the clustering
generalizations.
Clustering Based on Homogeneity or Separation
Traditionally, clustering methods have been classified into a taxonomy having two most general
groups: hierarchical and partitional (Jain et al.,
1999). Hierarchical methods produce a sequence
of nested partitions, whereas partitional methods
only produce one partition. Such distinction,
although relevant for many application domains,
does not highlight similarities and differences
between the various definitions of a cluster that
are implicit in the methods.
Alternatively, clustering can be regarded as
an optimization problem, in which the function
to be optimized is a mathematical measure of
homogeneity or separation (Hansen et al., 1997).
Homogeneity is a global measure of the similarity between points belonging to the same cluster,
separation is a global measure of the dissimilarity
between points belonging to different clusters.
For instance, the maximum cluster diameter
over all clusters is a measure of homogeneity,

882

whereas the minimum distance between objects

in different clusters is a measure of separation.
In such a perspective, clustering methods are
classified according to a taxonomy of homogeneity or separation functions, which expresses
implicitly cluster definitions. Such categorization
is most likely more effective in capturing different behaviors in practice. It therefore provides a
more natural avenue for the process of selecting
a clustering algorithm, which is most suited to a
particular application or domain.
It is worth noting that the actual implementation of the search for an optimal solution is left
entirely unspecified. Therefore, a suitable uniform abstraction can be given for a broad class
of methods, ranging from classical hierarchical
methods, to approximation algorithms, and a
consistent parametric user interface between the
clustering engine and the graphical user interface
can be designed.
Therefore and following up from Section The
Proposed Clustering Taxonomy and Framework,
clustering can be formally defined as an optimization problem (Hansen et al., 1997) that takes one
of the following forms:
Problem 1: Given an integer K > 1, find
the partition P of S of size K such that m(P) is
optimal.
Problem 2: Given a threshold R+, find
the partition P of S of minimum (maximum) cardinality such that m(P) (m(P) ). In other
words, sometimes m(P) is optimal when large and
at other times it is optimal when small.
In this research work, consideration is made
only for functions m ranging over the union of two
families H and S. The former family represents
homogeneity functions and the latter represents
separation functions. For a given data set S and
dissimilarity diss, define H by:

User Interface Formalization in Visual Data Mining

Equation 5
Ho(P) = {Q j : Oj C(O ) diss(Oi, Oj) | Q {max, ,
i
avg}}
Equation 6
Hc(P) = {Qi:O C h(Oi) | h Ho(P), Q {min,
i
max, , avg}}
Equation 7
H = {Qc P h(C) | h Hc(P), Q {max, ,
avg}}
and S by
Equation 8
So(P) = {Q j: Oj C(O ) diss(Oi, Oj) | Q {min, ,
i
avg}}
Equation 9
Sc(P) = {Qi : Oi C s(Oi) | s So(P), Q {min,
max, , avg}}
Equation 10
S = {QC P s(C) | s Sc(P), Q {min, max, ,
avg}}
where C(Oi) is the cluster containing object Oi.
Equation 5
defines a family of pointwise homogeneity functions, expressing that the homogeneity of object Oi can be defined as either the
maximum (i.e., worst-case) dissimilarity to other
objects in the same cluster, or the sum or average
of all such dissimilarities. Likewise, Equation

6
defines a family of clusterwise homogeneity
functions, expressing that the homogeneity of
a cluster can be defined as the maximum, minimum, sum, or average of pointwise homogeneity

of all its objects. Finally, Equation

7
defines a
family of partitionwise homogeneity functions;
the homogeneity of a partition can be defined as
either the minimum (i.e., worst-case) clusterwise
homogeneity over all its clusters, or the sum or
average of all such homogeneities. Equation 8Equation 10
provide analogous definitions for
the separation function. Note, however, that the
quantifier expressing worst-case pointwise or
partitionwise separation is the minimum instead
of the maximum, and that the quantifiers defining
the separation of Oi extend to every object not in
its cluster.
Equation 5
-
Equation 10
induce a simple
taxonomy with four levels into which functions
m are classified. At the first level, homogeneity is
separated from separation. Then, classes at lower
levels in the taxonomy are separated according
to the objectwise, clusterwise, or partitionwise
quantifier.
Clustering Based on Density Estimation
In this section, diss is assumed to be a distance
function and S to be a subset of a metric space
(X, diss). By elementary intuition, clusters can
be regarded as regions of the object space where
objects are located most frequently. Such simple
analogy leads to approaches to clustering based on
statistical techniques of non-parametric density
estimation (Ankerst et al., 1999; Ester et al., 1996;
Hinneburg et al., 1998; Schikuta, 1996; Silverman,
1986). The goal of density estimation is to fit to a
data set S a density function of type X R+. The
implemented system supports clustering based
on an important family of estimates, known as
kernel estimators (Silverman, 1986). Functions in
such family are defined modulo two parameters,
the window width h, and the kernel function .
The value of the estimate at x X is obtained
by summing, over all data objects, a quantity
modeling the influence of the object. Influence
is computed by transforming distance, scaled by
a factor 1/h, using :

883

User Interface Formalization in Visual Data Mining

Equation 11

h , ( x ) =

diss (Oi , x)
1 N
(
)

Nh i =1
h

Since the estimated density at x should be

large when the number of data objects which are
located near x is large, far data objects should have
little influence. Thus is usually a probability
distribution function, which decreases monotonically with distance from zero. Commonly
used kernel functions are the Gaussian kernel,
1
1
1
exp( x 2 ) , and the Uniform kernel,
2
2
2
(sign(x+1) sign (x 1).
The parameter h controls the amount of
smoothing of the estimate, i.e. the sensitivity of
the estimate to local structure in the data.
Once a density estimate is computed for a
given kernel function, clusters can be defined in
the following way. Let A(Oi) be a neighborhood
of Oi, and let h,(Oi) be the data object (if it exists) nearest to Oi in A(Oi) having density greater
than Oi:

Equation 13
P = {C : (xo S)C = {x S : xo ~ x}}
The method can be further enhanced by
introducing a notion of noise objects (Ester et
al., 1996): Density at a noise object is less than a
specified threshold parameter . Noise objects are
not part of any cluster, thus the method generates
a classification instead of a partition:
Equation 14
C = {C : (xo S)C = {x S : xo ~ x h, (x) }}
Problems 1 and 2 can be considered meaningful defining m(C) as the minimum density of
an object, over all objects in all clusters of C.
Abstract Syntax and Semantics of
Clustering
The visual language is defined by the following:
Equation 15

Equation 12

= R+ {min, max, , avg} {1, 2} {Homogeneity, Separation, Density}

diss(h,(Oi),Oi) = O min
{diss(Oi,Oj) : h,(Oj)>
j A ( Oi )
(O
)}
h,
i

Equation 16

The tree roots in the forest induced by are

objects whose neighborhood does not contain
objects having greater density. Therefore, the
tree roots may be considered as representatives
of the modes of the density estimate, and the
descendants of a root are part of a single cluster.
Formally, induces a forest whose connected
components are clusters: If is the equivalence
relation generated by , then the clustering P is
the set of its equivalence classes:

884

=
Equation 17
V = {ProbRad, NClusSpin, HomSlid, SepSlid,
DenSlid, AccurRad, HomObjwQCom, HomCluswQCom, HomPartwQCom, SepObjwQCom,
SepCluswQCom, SepPartwQCom, KernCom,
SmoothSlid}
F i n a l l y, l e t f o r b r e v i t y h p q =
v(HomPartwQCom), hcq = v(HomCluswQCom),

User Interface Formalization in Visual Data Mining

h o q = v ( H o m O b j w Q C o m) , s p q =
v(SepPartwQCom), scq = v(SepCluswQCom),
soq = v(SepObjwQCom). The classification C
of the dataset S is defined by:
Equation 18
C=
Phom(S) if v(accurRad) = Homogeneity,
Psep(S) if v (accurRad) = Separation,
Cden(S) if v (accurRad) = Density
and the following hold

if v(ProbRad) = 2
where
Equation 23
mhom(P) = hpq

msep(P)= spq
CP

Equation 25

|Phom(S)| = v(NClusSpin)
|Psep(S)| = v(NClusSpin)
|Cden(S)| = v(NClusSpin)

mden (C) = min

Equation 20
mhom(Phom(S)) v(HomSlid)
msep(Psep(S)) v(SepSlid)
mden(Cden(S)) v(DenSlid)
if v(ProbRad) = 2
Equation 21
P : | P |= v(NClusSpin) mhom (P) mhom (Phom (S))
P : | P |= v(NClusSpin) msep (P) msep (Psep (S))
C : | C |= v(NClusSpin) mden (C) mden (Cden(S))
if v(ProbRad) = 1
Equation 22
P : mhom(P) v(HomSlid) | P | | Phom (S) |
P : msep(P) v(SepSlid) | P | | Psep (S) |
C : mden(C) v(DenSlid) | C | | Cden (S) |

i:Oi C

hoq diss(Oi, Oj)

j:O j C ( Oi )

Equation 24

Equation 19

if v(ProbRad) = 1

hcq

CP

scq

i:Oi C

i:Oi C

soq

j:O j C ( Oi )

diss(Oi, Oj)

v ( SmoothSlid ), v ( KernCom )

(Oi )}

Metaqueries: In the metaquery environment of

VidaMine, the user can specify patterns/relationships between or among data tables in an intuitive
visual manner. The interface provides hooks
and chains through which users can visually
specify the relationships. By simply linking two
attributes, the users indicate to the system that
they are interested in metarules that have the two
attributes related. Therefore, VidaMine enables
users to visually construct metaqueries of interest. The left-hand side of Figure 2 shows the part
of the metaquery environment that supports the
foregoing specification.
Abstract Syntax and Semantics of
Metaquerying
Here a simple abstract syntax for the metaquerying visual interaction environment is defined,
together with its semantics, that is, the set of rules
discovered by the system as instantiations of the
metaqueries corresponding to the visual state
of the interface when the torch icon is clicked
(see Figure 2). For consistency with the visual
interaction environments for the other mining
techniques, the metaquerying visual environment

885

User Interface Formalization in Visual Data Mining

represents metaqueries using relation schemes

instead of literal schemes, that is, using named
attributes to denote table columns, instead of the
usual logical variables.
The metaqueries, which are added to the pool
when the add pattern button is clicked, are all
constructible metaqueries, given the named and
unnamed relations (i.e., named X) in the target
space. Note that the links between handles which
are represented in the target space do not have
a counterpart in the IF. . .THEN expressions
in the pool. In fact, such expressions only show
the relation schemes, and not the literal schemes,
that compose the head and body of a rule. Therefore, only the content of the target space defines
multiple occurrences of variables in a rule. Since
a user might click add pattern several times
with different configurations of the target space,
before clicking the torch button, for simplicity
only one such configuration is considered in the
semantics.
In the target space, rectangular frames enclosing attribute or relation names, and lines representing links connecting the frames (ignoring
the handles) are nodes of the multi-graph. A node
representing a frame is labeled by the relation
name or attribute name it represents (i.e., the
name appearing inside the frame). Attribute and
relation frames may be adjacent, and attribute
frames may be connected by lines. Therefore, for
relations, attributes and links, two visual relations
need to be represented: adjacent and intersecting.
Adjacency will be assumed antisymmetric: A
frame is adjacent to another if and only if the two
frames share a horizontal edge, and the first frame
is located above the second in the display.
In the following, let U and R be universes of
attribute names and relation names, respectively,
in the database. Let V = {X1, . . . ,Xi, . . .} be a
countably infinite set of variable symbols, and W
= {P1, . . . ,Pi, . . .} be a countably infinite set of
predicate variable symbols. V and W are assumed
disjoint and both disjoint from R. Let also : V
V, : V W be injective functions. In the sequel,

886

and will be used to construct the literals of

the rules from the frames.
The language is defined by the following:
Equation 26
= R+ U R {"X"}
Equation 27
= {adjacent, intersecting}
Equation 28
V {ConfSlid, SuppSlid}
The set IRS of rules returned by metaquerying
is defined as follows.
Equation 29
IRS = {r RS : conf(r) v(ConfSlid) supp(r)
v(SuppSlid)}
Equation 30
RS = {r:()(mq MQ) r = (mq)}
Equation 31
MQ =

{h

hL

bL { h}

Equation 32
L = {P(X1, ..., Xm) : (n) P = pred(n) isrel(n)
(i m)(n')Xi = (n') inschema (n', n)}
where
Equation 33
isadj (n,n') (e)l(e) = (n,n') (e) = adjacent

User Interface Formalization in Visual Data Mining

Equation 34
intersects (n,n') (e)l(e) = (n,n') (e) = intersecting
Equation 35
(n) if v(n) = " X ",
pred(n)=
v(n) otherwise
Equation 36

one metaquery for each literal, having the literal

as head and the remaining literals as body. The
rules RS instantiating MQ are defined by means
of a class of substitutions. In Equation 30, is any
substitution which, given a metaquery mq, consistently replaces exactly every predicate variable
occurring in mq with a relation name. Finally, the
rule set IRS is defined as the set of all rules in RS
that satisfy the support and confidence specified
by the sliders SuppSlid and ConfSlid.

Equation 37

Association Rules: Let be a set of items, and

T be a database of transactions, that is, a list of
elements of 2. Association rules (Agrawal et al.,
1993) are implications of the form:

isrel(n) (n') isadj (n', n)

Equation 40

Equation 38

I1, ..., Im I

islink (n) (n')(n")(n1)(n2) isrel(n1) isrel

(n2) isconn(n1,n2) isconn (n', n1) isconn (n",
n2) intersects (n, n') intersects (n, n")

where {I, I1, . . . , Im} . If r is the foregoing

association rule, and t = { I1, . . . , Im}, then r
has confidence conf(r) and support supp(r) in T
defined by:

isconn = isadje

Equation 39
inschema (n, n') (islink(n) (n") isconn (n', n")
intersects (n, n") (islink (n) isconn(n', n))
and isadje is the equivalence relation generated
by isadj , that is, the smallest equivalence relation
containing isadj. Therefore, an equivalence class
contains nodes corresponding to frames gathered
in one relation scheme in the target space.
L is the set of literals defined by the visual
configuration (Equation 32).
In each literal, P is the relation name enclosed
in a frame, or a distinct predicate variable, if the
name is X (Equation 35 and Equation 37). Every
variable corresponds to an attribute frame that
is connected to the relation frame, which names
the literal, or corresponds to a link that intersects
such an attribute frame (Equation 39). The set MQ
of metaqueries is obtained from L by generating

Equation 41
conf (r) =

{t T :{I } t ' t}
{t T : t ' t}

Equation 42
supp(r) =

{t T :{I } t ' t}
T

In the association rule environment of VidaMine, there is the provision of market baskets.
As seen in Figure 5, the association rule environment offers two baskets, the IF basket and the
THEN basket. The IF basket represents items in
the antecedent part of an association rule and the
THEN basket represents items in the consequent
part of the rule. Users may drag and drop items
from the target dataset into the relevant baskets.

887

User Interface Formalization in Visual Data Mining

Figure 5. Association rule environment: Input.

Reproduced from Kimani et al. (2004) with the
permission of Elsevier B. V.

the target space to the IF or THEN basket,

and later combined to form the rule and put into
a pool (of rules). The reader may refer to the
association rule visual environment in Figure 5,
and where the pool of rules is formed below
the baskets.
It is assumed that the abstract graph contains
one item node for every item occurring in the
IF. . . THEN rule, labeled with the items
name.
Assuming also nodes representing confidence
and support (see Section Abstract Syntax and
Semantics of Metaquerying), the set of association rules returned by the system is:
Equation 43
IAR = {r AR : conf (r) v (ConfSlid) supp(r)
v (SuppSlid)}
Equation 44
AR = {I1, ..., Im I : VTHEN = {n} I = v(n) (n
VIF)(i m)Ii = v(n)}
where VIF and VTHEN are the sets of item nodes in
the rules IF and THEN part, respectively.

When through with the visual construction of a

particular association rule, the users may empty
both baskets into a pool of association rules that
are of interest to them, and may go ahead and
construct another rule.
Abstract Syntax and Semantics of Association
Rules
The VidaMine systems support for association rules permits the specification of syntactic
constraints (Agrawal et al., 1993), that is, listing
items, which must appear in the antecedent or
the consequent. Such items are dragged from

888

Operational Specification of the Visual

Interface
As part of implementation, our research effort
also considers the provision of an operational
specification. This specification entails defining
a concrete syntax of mining tasks and an XMLbased communication protocol. The concrete
syntax of mining tasks is a high-level syntax of
each of the tasks proposed in our task analysis
and usability tests (Kimani et al., 2003). This
syntax describes the click-streams that are allowed to occur during the users interaction
with the interface. The syntax reduces distance
to the specification of procedural objects. The
XML-based protocol serves as a two-way link for

User Interface Formalization in Visual Data Mining

exchanging information between any applications

that deal with mining aspects. It is interesting to
note that the protocol idea can be exploited not
only to import and incorporate new components
into the system framework, but also to export the
systems components to external systems.
We start by describing the syntax of mining tasks. We then use the clustering subtask to
demonstrate how the communication protocol
has been defined.
Concrete Syntax of Tasks
Specifying a concrete, high-level syntax of mining
tasks can serve at least three purposes. First, it
highlights the most useful usage patterns, which
guided the interfaces layout and design. Second,
it permits an objective evaluation of the results
of usability tests. Finally, it provides a design
specification, which can be refined to yield the
complete parsing structure of the interface at the
level of information passed to and returned by
the mining engines. The abstraction level will be
such that terminal symbols represent interactions
that can be regarded as elementary in the current

Figure 6. Mining task

MiningTask ClusteringTask | MQTask | ARTask
Clustering Task ClusteringInputTask PerformClustering

ClusteringOutputObservingInteracting
MQTask MQInputTask PerformMQ RuleOutputObservingInteracting
ARTask ARInputTask PerformAR RuleOutputObervingInteracting

Figure 7. Syntax of clustering

ClusteringInputTask ClusteringTargetDataSubTask
ClusteringParametersSubTask
ClusteringTargetDataSubTask

SpecifyConstraints AssignAttributes

ForwardingTargetRelation |

SpecifyConstraints AssignAttributes

ForwardingTargetRelation

ClusteringTargetDataSubTask
SpecifyConstraints SpecifyConstraint |
Specif yConstraint Specif yConstraints
AssignAttributes AssignAttribute|

AssignAttribute AssignAttributes
ClusteringParametersSubTask Specify Parameter |

Specify Parameter

ClusteringParameterSubTask
SpecifyParameter SelectAccuracyMeasure | SelectCaseLabel |

SelectMiningAttribute
SelectAccuracyMeasure SelectHomogeneity | SelectSeparation |

SelectDensity
SelectHomogeneity PushHomogeneityRadioButton

SetHomogeneity
SelectSeparation PushSeparationRadioButtonSe
tSeparation
SelectDensity PushDensityRadioButton SetDensity
SetHomogeneity SetHomogQuantif | SelectHomFixedNum |

SelectHomogThreshold |

SetSeparQuantif SetSeparation |

SelectSepFixedNum SetSeparation |

SelectHomogThreshold SetHomogeneity
SetSeparation SetSeparQuantif | SelectSepFixedNum |

SelectSeparThreshold |

SetSeparQuantif SetSeparation |

SelectSepFixedNum SetSeparation |

SelectSeparThreshold SetSeparation
SetDensity SetKernel | SetSmoothing | SelectDenFixedNum

| SelectensThreshold | SetKernel SetDensity
|

SetSmoothing SetDensity | SelectDenFixedNum SetDensity |

SelectDensThreshold SetDensity

889

User Interface Formalization in Visual Data Mining

state of technology (e.g., clicking buttons, setting a

sliders position, selecting an item in a combo box,
setting the value of a spin box, etc.). The syntax
will be defined using standard grammar notation,
where terminal symbols are enclosed in double
quotes. For brevity, only the first occurrence of
every grammar rule, which is shared by two or
more tasks has been written.
Figure 6 describes the mining task. Each
subtask (clustering, metaquerying, association
rules) comprises an input task, clicking the torch
button to run the discovery algorithms, and an
output task.
In the sequel we briefly describe the concrete
syntax of clustering input and which can be seen in
Figure 7. The task is divided into two consecutive
parts, namely specifying target data and parameters, matching the upper and lower sections of
the clustering interface. Specifying constraints
means performing a sequence of manipulations
of the controls in the top part of the specification
phase, e.g., setting the number of attributes of a
given type, choosing a relation, etc. (see Figure 4).
Assigning attributes corresponds to dragging and
dropping attributes from the resource relations to
the X relation, which is subsequently dragged
into the target space. Selecting an accuracy
measure leads to selecting either homogeneity,
separation, or density, by pushing the respective
radio buttons (see Figure 4). Subsequently (see
Figure 7), the user may select quantifiers, or the
kernel function and the amount of smoothing.
Selecting the problem type gives access to setting the number of clusters or the threshold for
the accuracy function.
An XML-Based Communication Protocol
The large amount of recently proposed mining
techniques, and the need to easily select the most
effective method for the multivariate data set
at hand, calls for the design of extensible data

890

mining systems. Exchanging XML documents

allows for the addition of new mining engines
or new methods for existing engines as plug-ins.
Such a protocol would serve as a two-way link for
exchanging information between any applications
that deal with mining aspects. It is interesting to
realize that this idea can be exploited not only to
import and incorporate new components into the
system framework, but also to export the systems
components to external systems. In the VidaMine
system, communication between the GUI and
the mining engines is established by means of
XML document streams. The communication
protocol is specified by XSchema for the selection of methods and passing parameters, and for
passing output results to the GUI. Since PMML,
the predictive model markup language (DMG),
has already set an XML standard for representing
mining results in the data mining community,
our specification is based on, and in some parts
extends, PMML.
Example of XML-Based Protocol for
Clustering Task
Assuming the target dataset in Figure 4, the
generated XML document for method selection
is listed in Figure 8. Each attribute listed in the
X relation matches exactly one corresponding
DataField in DataDictionary and one MiningField in MiningSchema. In this research
work, active corresponds to the checked
state whereas supplementary corresponds to
the unchecked state of the attributes check
box in the target space. It is worth observing that
checked attributes are those that have been
chosen to directly participate in cluster analysis.
Unchecked attributes are those that will not
directly participate in cluster analysis but will be
included in the results for other purposes (such
as for labeling). AccuracyFunction allows for
selecting one of Homogeneity, Separation, and

User Interface Formalization in Visual Data Mining

Density. Within each, the problems form can be

selected by FixedNumberOfClusters (problem
1 of Section Clustering Based on Homogeneity
or Separation) or FixedThreshold (problem 2
of Section Clustering Based on Homogeneity or
Separation). The Density element includes
the type of density estimate used. Currently, the
only allowed type is Kernel. Other types can be
easily added as alternatives to Kernel. Figure
10 is a view of the XSchema, which specifies
documents containing the selection of a method
and parameters for clustering. It is worth noting
that the XSchema in Figure 10 does not allow for
specifying the algorithm used to solve the problem.

Figure 8. An example XML document for clustering input

Thus, for given problem definition and form of

the accuracy function, the clustering engine may
implement an exact or approximate algorithm.
In the current example, the selection of min for
objectwise, clusterwise, and partitionwise quantifiers is equivalent to the worst-case split clustering
problem (Hansen et al., 1997). For such problem,
an exact solution can be obtained applying the
well-known single link hierarchical algorithm to
the target dataset, halting at the topmost partition
whose worst-case split value is less than the value
of the split function (0.03 in the example). The

Figure 9. An example XML document for clustering output

891

User Interface Formalization in Visual Data Mining

document containing clustering result/output,

whose view can be seen in Figure 9, represents
the solution to the clustering problem in Figure
4. For brevity, the XSchema for the clustering
output is omitted, which is an extension of the
corresponding XSchema in PMML. The Cluster elements describe five clusters named 1, 2,
3, 4, and 5. Each Cluster element contains the
clusterwise accuracy function that was selected
for clustering, and its value at the cluster. (The
split of the cluster, in the example.) The engine
may include any other accuracy function in the
result document, specifying its value at the cluster,

Figure 10. XSchema for specifying clustering

input

and the clusterwise parameters defining it. (In

the example, the engine returned the clusters
Homogeneity computed as its diameter.) Finally, ClusteringArray assigns each object in
the target dataset to its cluster.
Mapping Between Clustering Task and XML
Documents
It is important to define a mapping between the
clustering and the XML-based communication
protocol that was just defined. When the formal
specification of the interface is defined through an
abstract syntax and semantics, a mapping between
user actions and XML documents describing the
actual parameters passed to the engines can be
defined by means of a recursive function. In the
VidaMine system, a mapping function M is defined by recursion on the grammar rules describing
the clustering task of Figure 7 to XML documents
conforming to the XSchema of Figure 10.
Although for brevity we have omitted its
formalization, it should however be noted that M
basically is a mapping satisfying a list of equalities of the forms:
Equation 45
(term)M xml(value1, ..., valuep)
Equation 46
(nt1, nt2)M = (EXT(nt1)M (nt2)M)
In Equation 45, term is a terminal symbol, xml
is a function constructing an appropriate XML
element corresponding to term, value is the value
of a visual control. In
Equation 46, nt is a nonterminal symbol representing the manipulation of
a visual control, EXT is a function that extends
the XML tree in its second parameter using the
elements in the first parameter. EXT also provides
appropriate defaults when the users actions do
not affect some of the controls.

892

User Interface Formalization in Visual Data Mining

Comparison with Data Mining

Systems
SQL Server 2005, developed by Microsoft Corporation, includes several data mining tools: association rules, clustering, decision trees, linear
and logistic regression, naive Bayes, and neural
networks. Its interface is based on menus, panes,
and tabs. Three panes present the user with an
explorer tree of folders related to analytical tasks
(data sources, multidimensional cubes, mining
structures, etc.), a properties visualization tool,
and tabs for model settings, visualization, evaluation, and for prediction. Parameters for the mining
algorithms can be accessed from pop-up menus.
The visualization tab supports all mining models
with several types of visualization each, with
some active graphical elements. Visual features,
such as color, can be associated to quantitative
features of models, such as relative frequency.
Overall, the interface provides a powerful set of
functionalities, but requires training to be used,
in particular for structuring the mining task. No
formal model is known for the graphical part of
the interface.
Clementine was originally developed by Integral Solutions Ltd (ISL) and later distributed as
a part of the SPSS system, after SPSS purchased
ISL on December 31, 1998. Clementine supports
many prominent mining techniques including
clustering, association rules, sequential patterns,
factor analysis, and neural networks. Users construct a map (a stream) of their data mining
project/model by selecting icons called nodes
that represent steps in the data mining process.
Clementine allows users to adjust/refine their
streams and rerun the system on the refined
model. Although its visual interface illustrates
the flow of control and data, allowing the user to
better understand and follow the mining process,
no formal syntax and semantics of the interface
as a whole is reported.
Enterprise Miner, developed by the SAS
Institute, provides a large number of data min-

ing algorithms, including decision trees, neural

networks, regression, radial basis functions, selforganizing maps, and clustering. Enterprise Miner
has an interesting visual interface, allowing for
powerful support of data transformation, visualization tools for multidimensional analysis, and
extensive parameter options for the algorithms.
The user drags and drops icons from a task bar
into a work area where the icons can be connected. Each connection represents the flow of
information from one node to its successor, and
each icon type represents a type of processing (for
example, imputation of missing data, declaration
of variable types and target variables, regression,
etc.) on the data from the predecessor node. Mining algorithms are viewed as a special type of
processing. The visual metaphor of nodes and
edges is intuitive; however, most of the complex
settings of each node type cannot be operated
upon graphically, but must be accessed through a
type-specific tab-based interface. The effort does
not report any formal description of the graphical
interface available.
Intelligent Miner is a data mining software
developed by IBM, supporting association
rules, clustering, classification trees, and neural
networks. There are two versions of Intelligent
Miner, Intelligent Miner for Data and Intelligent
Miner Modeling/Scoring/Visualization. Intelligent Miner for Data is a standalone program. Its
interface is based on explorer-like panes and dropdown menus introducing the various data mining
techniques, parameter settings, and visualization
windows. Intelligent Miner Modeling/Scoring/
Visualization is a suite of three distinct tools,
for the extraction mining models, for prediction,
and the visualization of models. The product is
integrated into the IBM DB2 Universal Database,
which provides all metadata support, for example
by storing the mining models and their visualizations. The interface of the modeling component is
textual and consists of stored procedures for data
mining algorithms callable from DB2s command
windows, whereas the visualization component

893

User Interface Formalization in Visual Data Mining

has a graphical display with active elements

and extensive sorting options. The interface of
Intelligent Miner for Data has no known formal
specification, as the interface of Intelligent Miner
Modeling/Scoring/Visualization. (For the latter, a
formal specification is relevant only to the visualization component, which is however, the same
supplied with the standalone version.)

OPEN ISSUES AND

FUTURE TRENDS
Despite the remarkable efforts and developments
that have taken place in the field of visual data
mining, the field still faces quite a number of
challenges and also exhibits various issues as
highlighted in the sequel:

894

Data mining methods normally can handle

or cope with large amounts of data. Since
visual data mining systems integrate data
mining techniques and visualization techniques, visualization techniques have to face
up to the challenge regarding large quantities
of data. Bearing in mind the limitations of
the display space, visualizations have to
seek effective mappings so as to display all
relevant data. Moreover, visualizations have
to use efficient techniques to load relevant
data in the main memory.
Most of the existing visual data mining
systems have not been subjected to thorough
usability evaluations. In order to determine
the usefulness of visual data mining systems,
the importance of usability studies cannot
be overemphasized. On the same note, it is
worth acknowledging that the lack of precise
theory or a universal/standard framework
for developing visual data mining systems
may not only be partly responsible for the
current situation, whereby each visual data
mining system seems to be an ad-hoc development, but may also to some extent at least

complicate the usability issue as far as the

visual data mining field is concerned. In that
respect, Grinstein, Hoffman, Laskowski,
and Picketts work on benchmarking and
evaluation in visualizations for data mining
(Grinstein et al., 2002) might be instructive.
It is a challenging activity to find the methods, techniques, and corresponding tools
that are suitable for a specific visual mining
task, or a particular type of data. One way
forward would be to evaluate the disparate
frameworks to determine their effectiveness
and to verify their applicability in different
application domains (Simoff, 2001).
Visual support often necessitates the usage
of graphical elements. The user is expected
to interact with these graphics in real time.
When the user interacts with the visualization, there should be some kind of immediate
feedback from the graphic. Bearing in mind
the fact that the memory demands on the
computer that are posed by graphics, may
decrease the performance of the system.
Consequently, changes introduced on the
visual environment may not immediately
reflect the changes occurring in the information realm in real time and vice versa. It is
no trivial task determining a good balance
between interactivity and performance.

In terms of our future research activities, we

are planning to support the output of descriptive
statistics, principally using pie charts and histograms, in the clustering environment. It also is
worth noting that a set of the system features are
under an engineering process in order to become
part of the commercial suite D2I, distributed by
Inspiring Software (http://www.cisat-group.com).
The D2I (Data to Information) environment will
serve as a generic suite for handling the data to
be subjected to mining. Even as we consider the
foregoing research directions, it is also important
to be aware of various trends in data mining. One

User Interface Formalization in Visual Data Mining

of the current trends is distributed data mining,

where the target data is located in different places
and/or in different physical locations. There is
also a growing interest in ubiquitous data mining, where the user may carry out data mining
from ubiquitous devices (such as PDAs, cellular
phones, etc) rather than from the conventional
computer. Another area that has been drawing
a lot of attention is hypertext and hypermedia
data mining. Efforts in the foregoing type of data
mining would normally entail one or more of the
following three aspects: content mining, structure
mining, and usage mining. It is also interesting to
observe a growing interest in audio data mining,
where audio signals are used to represent data
trends/patterns or other specific data features.

CONCLUSION
Despite the existence of many data mining efforts that deal with user interface aspects, there
is relatively little work going on or that has been
published on the specification of the syntax of such
user interface and the corresponding semantics.
In general, a formal specification can bring about
many benefits such as facilitating the description
of the system properties without having to be concerned about implementation details and enabling
the detection of fundamental design issues before
they manifest themselves in the implementation.
In visual data mining, where visualization is often
a key ingredient, a formal specification can be
rewarding. For instance, it can enable users to
decide which interaction/operation to apply to get
a particular output, it can help users to predict the
output of their interactions/operations with the
system, and it can facilitate the development of
interaction models. In this work, we have proposed
and described an approach for specifying such a
formal specification in the process of developing
a visual data mining system, VidaMine.

References
Agrawal, R., & Srikant, R. (1994). Fast algorithms
for mining association rules in large databases.
In J.

B. Bocca, M. Jarke, & C. Zaniolo (Ed.), Proceedings of the 20th International Conference on
Very Large Data Bases (VLDB). San Francisco:
Morgan Kaufmann Publishers.
Agrawal, R., Gehrke, J., Gunopulos, D., &
Raghavan, P. (1998). Automatic subspace clustering of high dimensional data for data mining
applications. In A.

Tiwary & M. Franklin, (Ed.),

ACM SIGMOD Record, Proceedings of the 1998
ACM SIGMOD International Conference on
New York:
Management of Data, 27(2), 94-105.
ACM Press.
Agrawal, R., Imielinski, T., & Swami, A. N. (1993).
Mining association rules between sets of items
in large databases. In P. Buneman & S. Jajodia
(Ed.), ACM SIGMOD Record, Proceedings of the
1993 ACM SIGMOD International Conference
New
on Management of Data, 22(2), 207-216.
York
: ACM Press.
Angiulli, F., Ben-Eliyahu-Zohary, R., Ianni, G.,
& Palopoli, L. (2000).
Computational properties
of metaquerying problems. Proceedings of the
19th ACM SIGMOD-SIGACT-SIGART Symposium
on Principles of Database Systems (pp. 237-244).
New York
: ACM Press.
Ankerst, M., Breunig, M., Kriegel, H. P., & Sander,
J. (1999). OPTICS: Ordering points to identify
the clustering structure. ACM SIGMOD Record,
Proceedings of the 1999 ACM SIGMOD International Conference on Management of Data,
New York
: ACM Press.
28(2), 49-60.
Baudel, T. (2004). Browsing through an information visualization design space. Extended Abstracts on Human factors in Computing Systems
New York
: ACM Press.
(CHI04) (pp. 765-766).

895

User Interface Formalization in Visual Data Mining

Boley, D. L. (1998). Principal direction divisive

partitioning. Data Mining and Knowledge Discovery, 2(4), 325-344.
Cheeseman, P., & Stutz, J. (1996). Bayesian classification (AutoClass): Theory and results. In U.
Fayyad, G. Piatetsky-Shapiro, P. Smyth, & R.
Uthurusamy (Ed.), Advances in knowledge dis
Park, CA
: AAAI
covery and data mining. Menlo
Press/MIT Press.
Chi, E., & Riedl, J. (1998). An

operator interaction
framework for visualization systems. In
G. Wills
& J. Dill (Ed.), Proceedings IEEE Symposium on
Information Visualization (InfoVis 98) (pp. 63-70).
Los Alamitos, CA
: IEEE Computer Society.
Chuah, M. C., & Roth, S. F. (1996). On the semantics of interactive visualizations. Proceedings
IEEE Symposium on Information Visualization 96
(pp. 29-36). Los Alamitos, CA
: IEEE Computer
Society.
DMG (The Data Mining Group). PMML: Predictive Model Markup Language. Retrieved March
15, 2006, from http://www.dmg.org
Ertz, L., Steinbach, M., & Kumar, V. (2003).
Finding clusters of different sizes, shapes, and
densities in noisy, high dimensional data. In
D. Barbar & C. Kamath (Ed.), Proceedings of
the 3rd SIAM International Conference on Data
Mining. SIAM. Retrieved from http://www.siam.
org/meetings/sdm03/index.htm
Erwig, M. (1998). Abstract syntax and semantics
of visual languages. Journal of Visual Languages
and Computing, 9(5), 461-483.
Ester, M., Kriegel, H. P., Sander, J., & Xu, X.
(1996). A density-based algorithm for discovering clusters in large spatial databases with noise.
In E. Simoudis, J. Han, & U. M. Fayyad (Ed.),
Proceedings of the 2nd International Conference
on Knowledge Discovery and Data Mining (KDD96) (pp. 226-231). AAAI Press.

896

Fayyad, U., Piatetsky-Shapiro, G., & Smyth, P.

(1996b). Knowledge discovery and data mining:
Towards a unifying framework. In E. Simoudis,
J. Han, & U. M. Fayyad (Ed.), Proceedings of
the 2nd International Conference on Knowledge

AAAI Press.
Discovery and Data Mining.
Fayyad, U., Piatetsky-Shapiro, G., Smyth, P.,
& Uthurusamy, R. (Ed.). (1996a). Advances in

knowledge discovery and data mining. Menlo

Park, CA
: AAAI Press/MIT Press.
Fraley, C., & Raftery, A. (1999). MCLUST: Software for model-based cluster and discriminant
analysis (Tech. Rep. 342). Department of Statistics, University of Washington.
Ganti, V., Gehrke, J., & Ramakrishnan, R. (1999).
CACTUS: Clustering categorical data using summaries. In S. Chaudhuri & D. Madigan (Ed.),
Proceedings of the 5th ACM SIGKDD International
Conference on Knowledge Discovery and Data
Mining (pp. 73-83). New York: ACM Press.
Goil, S., Nagesh, H., & Choudhary, A. (1999). MAFIA: Efficient and scalable subspace clustering for
very large data sets (Tech. Rep. CPDC-TR-9906010). Evanston, IL: Northwestern University.
Grinstein, G. G., Hoffman, P. E., Laskowski, S. J.,
& Pickett, R. M. (2002). Benchmark development
for the evaluation of visualization for data mining. In U. Fayyad, G. G. Grinstein, & A. Wierse
(Eds.), Information visualization in data mining
and knowledge discovery. San Francisco: Morgan
Kaufmann Publishers.
Guha, S., Rastogi, R., & Shim, K. (1999). ROCK:

A robust clustering algorithm for categorical attributes. In M. Kitsuregawa, L. Maciaszek, M.

Papazoglou, & C. Pu (Ed.), Proceedings of the 15th
International Conference on Data Engineering
(ICDE) (pp. 512-521). Los Alamitos, CA: IEEE
Computer Society.

User Interface Formalization in Visual Data Mining

Guha, S., Rastogi, R., & Shim, K. (1998). CURE:

An efficient clustering algorithm for large databases. Proceedings of the 1998 ACM SIGMOD
International Conference on Management of Data
(pp. 73-84). New York: ACM Press.
Hansen, P., & Jaumard, B. (1997). Cluster analysis
and mathematical programming. Mathematical
Programming 79, 191-215.
Hartigan, J. A. (1975). Clustering algorithms.
New York: John Wiley & Sons.
Hartigan, J. A., & Wong, M. (1979). Algorithm
AS136: A k-means clustering algorithm. Applied
Statistics 28, 100-108.
Hinneburg, A., & Keim, D. A. (1998). An efficient approach to clustering in large multimedia
databases with noise. In R. Agrawal, P. Stolorz,
& G. Piatetsky-Shapiro (Ed.), Proceedings of
the 4th International Conference on Knowledge
Discovery and Data Mining (pp. 58-65). Menlo
Park, CA: AAAI Press.
Jain, A. K., & Dubes, R. C. (1988). Algorithms

Cliffs, NJ: Pren

for clustering data. Englewood
tice-Hall.
Jain, A. K., Murty, M. N., & Flynn, P. J. (1999).
Data clustering: A review. ACM Computing Surveys, 31(3), 264-323.
Karypis, G., Han, E. H., & Kumar, V. (1999). Chameleon: Hierarchical clustering using dynamic
modeling. IEEE Computer, 32(8), 68-75.
Kaufman, L., & Rousseeuw, P. J. (1990). Finding
groups in data-introduction to cluster analysis.
New York: John Wiley & Sons.

International: Universal Access in HCI: Inclusive

Design in the Information Society. Lawrence

Erlbaum Associates.
Kimani, S., Lodi, S., Catarci, T., Santucci, G., &
Sartori, C. (2004). VidaMine:

A visual data mining environment. Journal of Visual Languages

and Computing, 15(1), 37-67.
Leng, B., & Shen, W. M. (1996). A metapatternbased automated discovery loop for integrated
data mining: Unsupervised learning of relational
patterns. IEEE Transactions on Knowledge and
Data Engineering, 8(6), 898-910.
Mannila, H. (2000). Theoretical frameworks for
data mining. SIGKDD Explorations Newsletter,
1(2), 30-32. New York: ACM Press.
Mitbander, B., Ong, K., Shen, W. M., & Zaniolo,
C. (1996). Metaqueries for data mining. In U.
Fayyad, G. Piatetsky-Shapiro, P. Smyth, & R.
Uthurusamy (Eds.), Advances in knowledge dis
Park,
covery and data mining (Ch. 15). Menlo
CA
: AAAI Press/MIT Press.
Nagesh, H., Goil, S., & Choudhary, A. (2001).
Adaptive grids for clustering massive data sets.
Proceedings of the SIAM International Conference on Data Mining. Retrieved from http://www.
siam.org/meetings/sdm01/pdf/sdm01_07.pdf
Ng, R. T., & Han, J. (1994). Efficient and effective
clustering methods for spatial data mining. In J. B.
Bocca, M. Jarke, & C. Zaniolo (Ed.), Proceedings
of the International Conference on Very Large
Data Bases (VLDB) (pp. 144-155). San Francisco:
Morgan Kaufmann Publishers.

Kimani, S. (2002). An effective visual data

mining environment. Doctoral Posters at the
International Conference on Very Large Data
Bases (VLDB).

Schikuta, E. (1996). Grid-clustering: An efficient

hierarchical clustering method for very large data
sets. Proceedings of the 13th International ConferLos
ence on Pattern Recognition (pp. 101-105).
Alamitos, CA: IEEE Computer Society.

Kimani, S., Catarci, T., & Santucci, G. (2003).

Visual data mining: An experience with the users. In C. Stephanidis (Ed.), Proceedings of HCI

Schikuta, E., & Erhart, M. (1997).

The BANGclustering system: Grid-based data analysis. In
X. Liu, P. Cohen, & M. Berthold (Ed.), Advances

897

User Interface Formalization in Visual Data Mining

in intelligent data analysis. Reasoning about

Data: Second International Symposium, IDA-97,
Proceedings, Lecture

Notes in Computer Science,

1280. Berlin; Heidelberg, Germany: Springer.
Sheikholeslami, G., Chatterjee, S., & Zhang, A.
(1998). WaveCluster: A multi-resolution clustering
approach for very large spatial databases. In A.
Gupta, O. Shmueli, & J. Widom (Ed.), Proceedings of the International Conference on Very
Morgan
Large Data Bases (VLDB) (pp. 428-439).
Kaufmann Publishers.
Shen, W. M., Leng, B., & Chatterjee, A. (1995).
Applying the metapattern mechanism to time
sequence analysis (Tech. Rep. ISI/RR-95-398).
USC Information Sciences Institute.
Silverman, B. W. (1986). Density estimation for
statistics and data analysis. London: Chapman
and Hall.
Simoff, S. J. (2001). Towards the development of
environments for designing visualisation support for visual data mining. In
S. J. Simoff, M.
Noirhomme-Fraiture, & M. H. Bhlen (Eds.),
Proceedings of the ECML/PKDD Workshop on
Visual Data Mining. Retrieved from http://wwwstaff.it.uts.edu.au/~simeon/vdm_pkdd2001/
Steinbach, M., Karypis, G., & Kumar, V. (2000).
A comparison of document clustering techniques.
Proceedings of the ACM SIGKDD Workshop on
Text Mining. Retrieved from http://www.cs.cmu.
edu/~dunja/KDDpapers/
Wallace, C. S., & Dowe, D. L. (1994). Intrinsic
classification by MML: The snob program. In C.

Zhang, J. Debenham, & D. Lukose (Ed.), Proceedings of the Australian Joint Conference on Artificial Intelligence (pp. 37-44). World Scientific.
Wang, J. Z., Wiederhold, G., Firschein, O., &
Wei, S. X. (1998). Content-based image indexing
and searching using Daubechies wavelets. International Journal on Digital Libraries (IJODL),
1(4), 311-328.
Wang, W., Yang, J., & Muntz, R. R. (1997).
STING: A statistical information grid approach
to spatial data mining. In M.

Jarke, M. J. Carey,
K. R. Dittrich, F. H. Lochovsky, P. Loucopoulos,
& M. A. Jeusfeld (Ed.), Proceedings of the 23rd
International Conference on Very Large Data

Francisco:
Bases (VLDB) (pp. 186-195). San
Morgan Kaufmann Publishers.
Ward, J. H. (1963). Hierarchical grouping to optimize an objective function. Journal of the American Statistical Association 58(301), 236-244.
Xu, X., Ester, M., Kriegel, H. P., & Sander, J.
(1998). A distribution-based clustering algorithm
for mining in large spatial databases. Proceedings of the International Conference on Data
Los Alamitos,
Engineering (ICDE) (pp. 324-331).
CA: IEEE Computer Society.

Endnote

VidaMine is an acronym for VIsual DAta

MINing Environment.

This work was previously published in Visual Languages for Interactive Computing: Definitions and Formalizations, edited
by F. Ferri, pp. 247-272, copyright 2008 by Information Science Reference, formerly known as Idea Group Reference (an
imprint of IGI Global).

898

899

Chapter 2.30

Understanding the Nature of

Task Analysis in Web Design
Rod Farmer
The University of Melbourne, Australia
Paul Gruba
The University of Melbourne, Australia

Abstract

Introduction

Designing usable Web-based interfaces challenges practitioners to carefully consider end-user

behaviour and requirements. Unfortunately, in
meeting this challenge, human-computer interaction (HCI), task analysis is often poorly understood and applied during Web design activities.
Rather than purely evaluating usability against
prescriptive guidelines, we argue that designing for Web-based interaction requires a more
holistic and descriptive approach. This chapter
provides an overview of cognitive and postcognitive HCI task analysis frameworks, and their
respective abilities to capture a systemic view of
stakeholder requirements. As such, this chapter
provides a valuable resource for researchers and
practitioners alike.

Although improved system design results when

researchers and developers understand how users
use technology (Raeithel & Velichkovsky, 1996),
understanding individual user traits, such as motivation and other contextual factors that guide
user participation during computer-mediated
activities can be deceptively complex. Simply
asking users what they want and how they use a
system is further complicated by the fact that users are often incapable of vividly and objectively
describing their experiences with the system
(Sommerville, 2004). Expertise, sociocultural,
and organisational policy factors may impact
perception of purpose, meaning, and context,
and hence influence the quality of user feedback
(Gasson, 1999). Therefore, determining whether a

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Understanding the Nature of Task Analysis in Web Design

system is fit-for-purpose for a particular end-user

population can be extremely challenging.
As developing fit-for-purpose systems is a
principal concern of human-computer interaction
(HCI), the system design process must ensure that
end-user requirements are validated against those
who have a vested interest in its use (stakeholders).
Therefore, choosing the right HCI framework for
eliciting, analysing, and modelling stakeholder
requirements is critical for ensuring overall system
quality (Farmer, Gruba, & Hughes, 2004). The
process of seeking to understand the human nature
of these requirements is referred to in HCI as task
analysis. There is a wealth of frameworks, models, methodologies, and tools that can be applied
to assist in this process. However, choosing the
most appropriate approach is dependent upon
several factors, including: the domain, context of
use, and available resources.
Task analysis is arguably the most important
aspect of HCI as it provides the analyst, researcher,
or developer with insights into the nature of human behaviour. A major benefit of conducting task
analysis throughout the software development life
cycle (SDLC) is its communicative power and ability to elicit and elucidate requirements throughout
each phase of development via a set of formalised
attributes and notations. Unfortunately, comparing and choosing the right task analysis approach
during system design is frequently hampered by
the lack of universal notations and user attributes
that can be applied across frameworks (Balbo,
Ozkan, & Paris, 2004).
The aim of this chapter is to provide a critical
overview of task analysis in HCI and its application to Web design. Specifically, the chapter will
discuss the cognitivist origins of task analysis,
and the recent shift towards more ecologically
valid approaches. We discuss several leading approaches within each paradigm and describe their
general applicability to Web design. We conclude
with an integrative approach to task analysis that
attempts to bridge the divide between cognitive
and postcognitivist perspectives.

900

Task Analysis in HCI

The term task analysis is commonly used to
denote a wide range of activities and processes
that attempt to either describe, equate, or predict
human performance during task-based interaction
(Diaper, 2004). A direct corollary of early cognitive psychological research concerning cognition
and procedural knowledge (Kirwan & Ainsworth,
1992; Miller, 1953, 1962), task analysis has been
applied successfully to numerous fields of research, including:

Interactive system design (Newman & Lamming, 1998)

Safety critical systems design and evaluation
(Patern & Santoro, 2002)
Cognitive engineering (Rasmussen, Pejtersen, & Goodstein, 1994; Vicente, 1999)
Computer-assisted language learning (Corbel, Gruba, & Enright, 2002; Farmer &
Hughes, 2005a, 2005c)
Multi-modal interaction (Farmer, 2005)
Intelligent learning object classification
(Farmer & Hughes, 2005b)
Social intimacy (Vetere et al., 2005)
Web design (Dix, 2005)

It is therefore not surprising to see an increasingly divergent array of theoretical perspectives

emerging on the nature of human-machine interaction. Observing that methodologies in HCI have
already reached a sufficient level of sophistication
and application, Whittaker, Terveen, and Nardi
(2000) have argued that it is time to address existing problems, rather than develop additional idiosyncratic models and notations. Indeed, renewed
focus has recently been applied to the problem of
integrating and grouping task analysis theories
and techniques, which at first glance may appear
fundamentally incommensurate (Farmer, 2006;
Wild, Johnson, & Johnson, 2003).
The primary aim of task analysis is to produce
a reliable procedural description of human praxis.

Understanding the Nature of Task Analysis in Web Design

Table 1. Types of task analysis (MITRE, 2003)

Types of Task Analysis
Cognitive task analysis
Knowledge elicitation
Computational cognitive modelling
Task analysis
Computational task simulation

However, unlike mere procedural functions,

tasks incorporate the notion of purpose (Preece,
Rogers, Sharp, Benyon, Holland, & Carry, 1994,
p. 411), hence goals and planning. Many task
analysis techniques are built on the belief that
systems are reducible to their constituent parts,
promoting a systematic, linear decomposition of
human praxis (Watts & Monk, 1998). However,
systems can never be logically decomposed into
subsystems without losing some implicit value,
whether operational (knowledge) or functional
(capability) (Latour, 1987). This said, linear reductionist descriptions of work may be appropriate
for developing highly constrained, well-defined
environments, such as interactive voice response
(IVR) systems that typically restrict users to
inhabiting a single state. Such cognitive environments therefore promote a strong adherence to a
narrow view of user behaviour (Robinson, 1993).
Table 1 shows a variety of task analysis approaches
typically employed within HCI.
Disregarding the actual methods applied
within these frameworks, we can further collapse these approaches onto two conceptual axes:
descriptive/analytical and cognitive/system. The
descriptive/analytical axis reflects whether the
purpose of the analytical framework is to describe
or to empirically analyse interaction behaviour.
The cognitive/system axis reflects whether the
framework is cognitively oriented, concerned with

System evaluation methods

Descriptive analysis
Human reliability analysis
Cognitive-oriented methods
System-oriented methods

modelling mental processes during work activity,

or system oriented, concerned with modelling
how individuals relate cognitively to their social,
organisational, and environmental work contexts.
As these axes represent continuums, more finegrained categories of task analysis now emerge.
Low-level analytic/cognitive approaches such
as goals, operators, methods and selection rules
(GOMS) and hierarchical task analysis (HTA)
tend to treat task interaction as a series of linearly
sequenced actions from which we can derive the
sense of the activity, and thus effective system design. Higher-level system/descriptive approaches,
such as soft systems methodology (Checkland,
1999), tend to take a more holistic perspective,
focusing on situated activity rather than actions.
The growing importance of modelling situation
awareness during HCI is representative of a paradigmatic shift away from individual, information
processing accounts of interaction, towards more
socially relevant, tool-mediated representations
of activity.

Gathering Requirements via Task

Analysis
Requirements engineering (elicitation, analysis,
and specification) is essential to developing quality Web-based systems as it (1) helps to elicit
possible user groups and their level of involve-

901

Understanding the Nature of Task Analysis in Web Design

ment; (2) provides insight into user and system

requirements; and (3) places the development
effort within a contextual environment (Balbo
et al., 2004; Farmer et al., 2004). Schach (1999)
provides a case-based analysis of software development projects, highlighting that the origin of
80% of all software defects can be traced back
to earlier requirements engineering and design
phases. In addition, the cost of fixing a defect
during the maintenance phase, as opposed to
these earlier phases, is an order of magnitude. As
both requirements engineering and task analysis
serve to determine which features of a system will
render it fit-for-purpose, it is necessary to ensure
they are conducted appropriately throughout the
SLDC (see Table 2).
According to Mager (1991), there are four
distinct fields of work analysis: (1) performance,
(2) critical incident, (3) task, and (4) goal analysis.
Performance analysis attempts to describe aspects

of human activity that are directly related to understanding and improving human performance.
Traditionally, this area of research has focused
on two aspects of work: operator awareness
and capability modelling; and human-system
simulation environments. Operator modelling
is for the most part a behavioural activity, using
techniques such as time analysis and operator
function models (Kirwan & Ainsworth, 1992).
Human-system simulation environments, on the
other hand, require sophisticated, computationaltask simulation tools and theories to evaluate
human decision making, including recognitionprimed decisions and naturalistic decision making
(Zsambok & Klein, 1997).
Critical incident analysis focuses on developing causal models of relationships within
complex systems to prevent accidents or error
states in safety-critical environments (Shrayne,
Westerman, Crawshaw, Hockey, & Sauer, 1998).

Table 2. Purpose of task analysis various stages of the SDLC

Phase
Discovery
and definition

Design

Development
and Deployment
Evaluation
and prediction

902

Purpose
To elicit, analyse, and specify functional and nonfunctional stakeholder requirements within the context of use and the existing limitations and constraints upon
the activity
To generate high-level, coarse-grained descriptions of the main tasks and objectives that are relevant to the user(s)
Traditionally, the role of this phase has been to develop the user-interface design, however it should also include high-level, system architecture specification,
documentation and additional resource design. These activities should all occur
concurrently.
To define and model the generic functionalities of the system, especially consistency and learnability (affordance, usability) when the system is to be deployed
across several platforms
To analyse the implemented functionality of the system in terms of efficiency
To automatically generate parts of the architecture and subsystem components
related to the functional stakeholder requirements
Through the use of specific task modelling notations, to produce a series of design
tests which evaluate user performance and ensure that the final product is fit-forpurpose

Understanding the Nature of Task Analysis in Web Design

As such, this work requires a higher-degree of

probabilistic modelling and stochastic analysis
than the other fields. More than any other work
analysis related field, critical incident analysis
attempts to reduce complex activities to discrete,
measurable events.
Perhaps the most difficult category to define,
however, is task analysis. Researchers and practitioners have failed to agree upon a universal
definition of task, including an appropriate
unit of analysis (Carroll, 2002). As a result, the
term task analysis has been used to describe all
four categories of work analysis, especially goal
analysis. Accordingly, contention has arisen
concerning the structure of tasks, and their relationship to goals and planning (Nardi, 1996;
Suchman, 1987).

What is a Task?
Figure 1 depicts a traditional structural view of
task in HCI where it is seen as a conscious act
(Kirwan & Ainsworth, 1992; Watts & Monk,
1998). Task-as-activity is comprised of some goal
that must be achieved through mediated interaction via agents and artefacts of the environment
(Flor & Hutchins, 1991; Hutchins, 1995). Planning
is implied as goals reflect the system state to be
achieved through effecting some combination of
actions (events). Unfortunately, here we run into

Figure 1. Traditional structural view of task

trouble. Preece et al. (1994) conceive activities

as devoid of control structures and not requiring
thought. While this may be true of some activities where procedural knowledge has reached a
certain level of automaticity through habituation,
it cannot be true of all interactive systems, such
as Web environments.
Goals require constant monitoring and evaluation (Suchman, 1987), which implies conscious
decision making. Indeed, it is more logical to
assume that tasks are constructed and maintained
within a frame of reference that includes planning
(Filkes, 1982). Planning requires knowledge of
when and how to initiate tasks. We therefore question the correctness of this traditional view of Task,
and propose Figure 2 as an alternative. Here the
term task is substituted for activity. As activities
are motivated social constructs (Leontev, 1978),
it is appropriate that they appear superordinate to
goals, individual goals being a subset of collective goals. As activities exist at the social level,
they are less temporal than subjective individual
acts. Planning is therefore rooted firmly within
specific communities of practice, and less open
to change and violent fluctuations in behaviour.
Given their degree of permanence, we classify
conscious decision making conducted at this level,
meta-planning. Furthermore, Suchman (1987)
argues that individuals plan reflexively rather
than deliberatively, responding to immediate
changes in environmental conditions. As such,
planning at the goal level is likely to be affected
by micro changes during an activity. We classify
this within-activity planning, micro-planning.
Figure 2 suggests that tasks are subordinate to
goals, thus inverting the traditional view. Moreover, Figure 2 distinguishes between two types
of task: conscious and unconscious.
The mediating variable in task classification
is habituation. The more one practices an event,
the fewer attentional (cognitive) resources need to
be assigned to performing that task. Conversely,
when an error or breakdown in information flow
occurs, these tasks are raised into consciousness

903

Understanding the Nature of Task Analysis in Web Design

Figure 2. Activity-oriented structural view of task

and associated with an existing or newly formed

goal (Engestrm, 1987). Raising unconscious
actions into conscious goal-oriented behaviour
has been shown to significantly impact task performance (Farmer, 2005).

Divisions within HCI on Task

Analysis
The notion of Task has historically been dependent upon the predominant theoretical models
of the time. According to Goguen (1996), there
are today two dominant theoretical perspectives
within HCI: the cognitive (cognitive science and
experimental psychology) and the postcognitive
(sociology and anthropology). These perspectives
exist along a continuum polarised by the split between the natural and social sciences (Raeithel &
Velichkovsky, 1996). Selecting between cognitive
and ecological theories of human praxis can have
considerable impact upon system development.
For instance, the view that participation within

904

an activity is not isolated, yet inexorably bound

to external relationships and sociocultural conditions poses serious questions for how we model
both functional and nonfunctional requirements
during online system design.
Cognitive approaches to task analysis primarily focus on aspects of user performance
(Diaper, 2004), often neglecting sociocultural
phenomenon essential to meaningful interaction. Postcognitivist, or ecological, approaches
underscore the importance of context in evaluating human praxis. Postcognitive approaches hold
that meaningful interaction can only be derived
from careful consideration of the environmental,
sociocultural, and historical contexts in which
an activity occurs. Moreover, activity itself is
distributed among various cognitive systems
within the environment, and not just at the user
interface, as interaction requires the accessing
of information and/or anticipation of events that
may be beyond the users experience. Hence,
measuring the meaningfulness of interaction is

Understanding the Nature of Task Analysis in Web Design

not merely an empirical measure of performance

(human or technological), rather an evaluation of
the appropriateness of an activity in relation to a
set of established goals. Furthermore, meaning
during online activity is negotiated through the
user interface, and not at the user interface.
Contrary to the more behaviouristic views of
cognitivist approaches, postcognitive perspectives
are driven primarily by sociological considerations, including how work is shared, organised,
and completed within communities of practice
(Wegner, 2003). This shift is also indicative of
increased research into how we can better capture
the context of interaction; develop greater understanding of user behaviour in situ ; and how
tool mediation impacts knowledge acquisition and
management (Kaptelinin, 1996b).
The cognitivist dialogue maintains an assumption that derived task models can be universally
applied across different user groups, implying a
greater degree of similarity between individuals
than should otherwise be accorded (Preece et
al., 1994). Carroll (1997) suggests that designing systems upon the notion of universally applicable task models implies certain cognitive
equivalences between humans and computers, yet
maintains that artefacts lack self-awareness, and
should never be thought to possess knowledge of
their behaviour. Although some may suggest that
artefacts possess certain cognitive affordances,
this is often a misinterpretation of the impact of
social process, such as cultural conventions, on
object use (Norman, 1999).

Approaches to Task Analysis

in HCI
As previously highlighted, the two primary task
analysis movements within the field of HCI are
cognitive (behaviourist) and postcognitive (sociocultural). Methods employed within the cognitivist paradigm promote action-oriented analyses
of interaction (Kaptelinin, Nardi, & Macaulay,

1999). Postcognitive frameworks represent not so

much an opposing view, but rather an argument
for greater emphasis on modelling human activity
as it occurs in situ. Postcognitivist methodologies
favour goal-oriented or object-oriented analyses.
To better understand how these paradigms differ from one another, we now consider related
methodologies.

Behavioural Task Analysis

Behavioural, or cognitive task analysis frameworks focus primarily on how human behaviour
is determined by the users internalisation and
transformation of input, and how this information
results in output as action. Cognitive approaches to
HCI mostly focus on the user-computer interface,
as it forms the critical point of coherence between
two discrete models of processing that of the
human and the computer. While this is true of
most underlying cognitive task models, differing
purposes of use have resulted in considerable
variation between approaches.

Cognitive Task Models

There are various levels of behavioural task
analysis approaches within HCI. These include
low-level cognitive methodologies, such as
HTA (Annett, 2004; Annett & Duncan, 1967;
Shepherd, 2001), and higher-level, meta-cognitive frameworks, such as scenario-based design
(Carroll, 1996, 1997, 2000, 2002; Rosson & Carroll, 2002).
Behavioural frameworks are structured upon
specific task models, structured descriptions of
the interactive task to be performed by the user(s).
It encompasses the notion of task and as such
represents the specific set of actions that the user
may undertake in a goal-directed activity. Task
models may also include a task hierarchy, indicating task sequencing and any additional constraints
upon usage (Limbourg & Vanderdonckt, 2004).
Although most vary substantially in their depth

905

Understanding the Nature of Task Analysis in Web Design

of analysis, degree of formalism, and purpose, all

task models comprise a set of common objectives
(Table 3).
According to Table 3, these common objectives
imply that task models can be used for a variety of
purposes, from requirements engineering through
to system evaluation, representing varying levels
of abstraction and serving different purposes
within the SDLC (Balbo et al., 2004).
The underlying theme within behaviourist
frameworks is that users have an overarching
system goal they wish to achieve, which through
intentionality is transformed into action. Decision
making occurs through comparison of intention
and perceived system behaviour (feedback) (Norman, 1999). As such, interaction is modelled as
a closed information-processing loop input,
transformation, and output. These types of analysis are effective for modelling and evaluating the
primitive tasks that determine user behaviour in
well-structured activities. Here, we briefly describe a set of highly influential, cognitive task
analysis frameworks. With the possible exception
of cognitive work analysis (CWA), we argue that
these approaches fail to sufficiently support decision making and problem solving in unstructured
and dynamic environments.

Goals, Operators, Methods and

Selection Rules
GOMS-based approaches have received considerable attention in the HCI literature (John & Kieras,
1996a, 1996b), as it can be used quantitatively to
predict efficient user interaction with a system
design (task performance), and qualitatively as
a means of modelling low-level tasks (Kieras,
2004). GOMS-based approaches are also effective at predicting future task performance.
However, models of interaction derived from
a GOMS analysis tend to be abstract, rather
than specific. Moreover, GOMS as initially
conceived does not account for motivational
factors involved in computer-mediated activities,
nor is it well suited for modelling social and collaborative work practices.
The unit of analysis in GOMS is the task. Like
HTA, GOMS describes task-based interaction
as a hierarchically ordered sequence of actions,
aligned with specific goals (and subgoals) (Card,
Moran, & Newell, 1983). Briefly, GOMS views
tasks as comprising:
1.

Goals: Desired outcomes within some

computer-mediated activity;

Table 3. Common objectives of task models (Bomsdorf & Szwillus, 1998)

Objective
To inform
Evaluation of human performance
System design support
System generation

906

Purpose
To inform users and designers about potential problems. Common themes
involve needs analysis, usability, and affordance.
To establish the impact of new or existing tools upon task performance in
some work practice
To provide a detailed and structured conceptual model of the task, including behavioural (time, errors, feedback, etc.) and structural modelling
(functionality, visibility, etc.)
To help develop prototypes, user interfaces, and elements of the system
architecture

Understanding the Nature of Task Analysis in Web Design

2.
3.

4.

Operators: Actions that must be executed

in order to achieve some goal;
Methods: The set of operators that are
required to achieve some larger (system)
goal; and
Selection rules: Logic rules used to choose
between competing methods given some
system feedback.

Within GOMS, input is received from the

system, which is processed and associated with
internalised representations of the system (mental models). This internal system information is
aligned with immediate and overarching goals,
and subsequently associated with competing possible actions. Selection rules serve as transformation functions, helping to determine which actions
possess the greatest utility, and thus provide the
most effective means of interaction. Output is the
corollary of some cognitive utility evaluation,
representing the most effective path to achieving
the desired goal.
For GOMS, generating meaning from interaction occurs at the man-machine interface
and is highly determined by subjective internal
representations of the activity. Unfortunately,
GOMS is highly prescriptive and fails to consider distributed models of interaction, which
consider environmental artefacts, organisational
behaviour, and social conventions that reflect
historical decision-making processes and cultural
conventions (Hutchins, 1995).

Hierarchical Task Analysis

HTA is an action-oriented, decompositional
framework primarily concerned with the training of people in system design and use. Initially
developed for analysing nonrepetitive operator
tasks (Annett, 2004; Annett & Duncan, 1967),
HTA grew out of the need for a more expressive
model for describing mental processes during
man-machine interaction. HTAs theoretical

foundations lie in system theory and information processing, seeing task performance as the
interaction between human and machine, the latter becoming increasingly complex as computers
and automation develop[ed] (Annett, 2004, p.
68). With relation to task performance, the term
analysis in HTA refers to the process of problem
identification and structuring. HTA is an effective process for proposing empirical solutions to
existing specified problems (Annett, 2004).
In HTA, goal refers to an expected outcome,
or system state. Goals and subgoals, may be active
or latent, arising when the need presents itself to
achieve some expected outcome. HTA frequently
models this goal-oriented behaviour as tree structures, such as decision trees. Goals are established
and acquired through an information processing
cycle similar to that in our discussion of GOMS
(Annett & Duncan, 1967; Kieras, 2004). As with
goals, actions in HTA may also be decomposed
into nested actions, each maintaining their direct
relationship to an established system state. Being
nested, actions are available at both the current
node of activity, and at their super node (parent
task). Therefore, according to HTA, user behaviour can be reduced to a cycle of monitoring for
new input, deciding upon available alternatives
and controlling subsequent behaviour.
Although an attractive framework, there are a
number of problems HTA practitioners typically
encounter. Firstly, when a parent goal becomes
active, all subgoals (and their related actions) become active as well. However, it is seldom the case
that a user is simultaneously aware of current and
future goals. Rather, user behaviour is typically
more anticipatory and reflexive (Endsley, 2000;
Suchman, 1987). According to HTA, subgoals
and their actions are maintained as conscious
constructs available to the user. Again, this is
unlikely to be the case, as many procedural actions
essential for effective task completion are habituated or unconscious even when the associated goal
is being actioned (Whittaker et al., 2000).

907

Understanding the Nature of Task Analysis in Web Design

GroupWare Task Analysis

Developed by van der Veer, Lenting, and Bergevoet (1996), GroupWare task analysis (GTA)
attempts to incorporate cooperative strategies
within cognitive task models. Within GTA,
complex tasks are decomposed into low-level
unit tasks (Limbourg & Vanderdonckt, 2004).
However, by assigning roles to both tasks and
agents in the environment, GTA differs from
HTA in its ability to model aspects of collaboration and cooperation. As roles change within the
organization and across activities, traits that link
specific tasks to activity structure, agents, and artefacts can be identified. This process contributes
a social dimension to task performance analysis.
Unfortunately, GTA currently lacks sufficient
formalism, and therefore expressive power, to
handle complex environments. Additionally,
GTA is presently only effective when evaluating
existing systems. It is therefore not appropriate
for exploratory modelling activities.

Cognitive Task Analysis

Cognitive task analysis (CTA) refers to a host
of techniques that can be applied to investigating expert system usage (Gordon & Gill, 1997).
Examples include naturalistic decision making,
recognition-primed decisions, and schema theory
(Zsambok & Klein, 1997). CTA techniques address many of the weakness inherent to HTA
and GOMS by explicitly modelling procedural
knowledge and complex information-processing
structures (Hollnagel, 2003). Decision ladders
capture the central characteristic of information
processing, yet incorporate short circuits, or
experience-based patterns of action that permit
experienced individuals to move between goal
structures in response to changing action and planning resources. CTA is therefore highly suited to
modelling expert behaviour. However, analysing
sociocultural features of an activity is not sufficiently developed or formalised in CTA. Although

908

HTA and GOMS may be cast as derivatives of the

CTA approach, CTA is inherently more adaptable,
as seen through (1) its advocacy of methodological plurality; (2) promotion of task heterarchies
over hierarchies; and (3) its consideration of both
cognitive and environmental artefacts within an
activity. As CTA models activities in terms of
cognitive architectures, it is less applicable to
investigating exploratory, descriptive, and illstructured social activities.

Cognitive Work Analysis

CWA is a work-centred framework that represents
a major advancement upon traditional CTA approaches. It is designed specifically for analysing
work environments that require significant human
behavioural adaptation to new and existing tasks
(Rasmussen et al., 1994; Vicente, 1999; 2004).
It is built upon a meta-cognitive task model that
integrates both cognitive and ecological practices
within a holistic framework for work practice
analysis.
CWA differs from other cognitive approaches
in that it does not merely focus on normative
modelling (how HCI should proceed); descriptive modelling (how HCI currently occurs); or
formative modelling (how specific actions are
sequenced during an activity). Rather, the aim
of CWA is three fold:
1
2.
3.

Identify the properties of the work environment

Define likely work boundaries
Evaluate boundary effects on HCI

In CWA, actor-system interaction is always

situated within some activity; hence, it considers not just the tasks actors perform, but also
the situation in which they are undertaken. In
conducting a work analysis, activity is broken
down into five levels of analysis: (1) work domain
analysis; (2) control task analysis; (3) strategies
analysis; (4) social-organisational analysis; and

Understanding the Nature of Task Analysis in Web Design

(5) work competencies analysis. In practice, these

five levels are used to constrain possible humancomputer design considerations. Although prescriptive methods are formalised and integrated
within CWA, the framework itself advocates a
plurality of analysis techniques, so long as they
fulfil a practical purpose. As conceptual levels
are interdependent (soft boundaries), analysis
progresses from one layer to another in an iterative fashion all layers activated to some lesser
or greater extent at any point in time. Moreover,
in applying CWA, there is no explicit ordering of
analysis activities (Vicente, 1999). This renders
the approach extremely flexible and reduces the
learning curve required to apply its principles
in practice.
Work domain analysis seeks to establish the
structural limitations of the work domain. This
stage identifies the type of users expected to
participate in the activity, the activity problem,
and additional logical and physical constraints
upon the system. Control task analysis represents
the set of abstract tasks that control the general
flow of information within an activity. Strategies
analysis investigates how individuals or groups
overcome a problem when presented with a set

of alternatives. This decision-making process is

similar to that of selection rules in GOMS. However, unlike GOMS, this process considers prior
experiences, spatio-temporal constraints upon
operation, motivation, aptitude, and other similar
criteria. Social-organisational analysis seeks to
establish how tasks are shared between actors
and artefacts in a cooperative and collaborative
manner. Finally, work competencies analysis
examines the degree of training and experience
actors bare upon some activity.

Summary of Approaches
To determine which approach may best be used
within a particular Web design project, Table 4
provides a brief overview of the cognitive approaches previously mentioned and demonstrates
their shared and unique attributes.

Ecological Task Analysis

Until recently, task analysis in HCI has focused
almost exclusively upon deriving reductionist
cognitive representations of individual tasks (Hollan, Hutchins, & Kirsch, 2001). Although HCI is

Table 4. Overview of cognitive task models

Criteria
Expressive power
Complexity
Collaboration
Timing
Roles and responsibilities
Evaluation
Requires Training
Scalable
Social Orientation

Approaches
GOMS
HTA
LOW
LOW
LOW
LOW
NO
NO
YES
YES
NO
NO
YES
YES
LOW
LOW-MED
NO
NO
NO
NO

GTA
MEDIUM
MEDIUM
YES
NO
YES
YES
MED-HIGH
NO
NO

CTA
HIGH
HIGH
NO
NO
NO
YES
HIGH
YES
NO

CWA
HIGH
HIGH
YES
NO
YES
YES
HIGH
YES
YES

909

Understanding the Nature of Task Analysis in Web Design

currently moving towards more ecologically valid

models of human praxis, postcognitivist theorists
have yet to agree upon the most promising alternatives (Kaptelinin, 1996a). Resolving this issue
depends on two factors:
1.
2.

Establishing a common unit of analysis

Constructing integrative frameworks that
model both cognitive and social phenomena

The need for integrative frameworks is especially relevant to Web designers who must not
only support existing social practices through a
novel medium, but also the cognitive demands
imposed by semiotic constraints when delivering
information via the Web (Smart, Rice, & Wood,
2000). Table 5 outlines the required properties of
an integrative framework.

At the heart of the postcognitive movement

lies the notion of situation awareness. Situated
awareness represents an understanding of social
norms and rules, or a state of knowledge about
the context, situation, and environment in which
an activity takes place (Erickson & Kellogg,
2001). Situation awareness does not describe,
per se, the process of acquiring knowledge, but
rather is the end result of invoking any number
of cognitive and meta-cognitive processes during
active participation in a practice (Endsley, 2000).
Furthermore, situation awareness is distributed
among individuals, and hence, actionable and
relevant only within specific communities of
practice (Gasson, 1999). Situation awareness
requires individuals to maintain an awareness of
the continuously unfolding characteristics of an
activity. Summarising Endsley (2000), situation
awareness entails:

Table 5. Postcognitive task analysis criteria

Property

Description

Formalised notation
system

Common, accessible notation that facilitates conceptual modelling and communication of knowledge between interested parties (Balbo et al., 2004; Erickson &
Kellogg, 2001)

Methodological flexibility

The ability to adopt existing methodologies rather than constantly reinvent the
wheel (Whittaker et al., 2000)

Cost-effective practices

Cost-effective practices that encourage the use task analysis during system design,
thus increasing return-on-investment (Stanton, 2004)

Practical methodologies with sufficient expressive power to support novice and

expert usage (Balbo et al., 2004)
An ability to be applied across the SDLC, including integrative mechanisms that
Integration within formal
facilitate software engineering practices (Farmer, 2006; Farmer et al., 2004; Flor
software engineering
& Hutchins, 1991; Kazman, Abowd, Bass, & Clements; 1996; Rosson & Carroll,
frameworks
2002)
Expressive power

Reuse

The ability to support that HCI claims reuse (Carroll, 1996; Sutcliffe, 2000)

Ecological validity

Research findings and technology must support existing social processes and cultural practices (Nardi, 1996)

910

Understanding the Nature of Task Analysis in Web Design

1.
2.

3.

Perceiving relevant contextual information

Understanding how the roles and relationships between actors impacts information
presentation and communication
An ability to use social cues to assist in
planning for future events

Albers (1998) states that the quality of an

individuals situation awareness is governed by
their ability to comprehend, integrate, and reconstruct new information. This three-step process
involves extracting the relevant basic information
from the task/activity, integrating the information
into the individuals conceptual understanding of
the problem, subsequently generating new rules
and beliefs so as to extrapolate the information
into future problem-solving situations. As situations change and new information is presented,
so must the individuals awareness of emerging
conditions that are likely to impact upon their
effective participation in an activity. The problem of situation awareness is highly relevant to
developers of Enterprise Content Management
systems where the most critical factor impacting
successful deployment and integration is user
adoption. Postcognitive task analysis frameworks
that maintain the centrality of situation awareness are more likely to succeed at meeting this
challenge.
Goals, experiences, expectations, and motivations are interdependent constructs that produce
localised expectations about situated action, and
as such influence situation awareness (Endsley,
2000). Goals help to orientate an individual across
a series of tasks, including error recovery. For
example, when a user encounters an error during
interaction with a Web-based form, a breakdown
in planning occurs. In assessing available feedback, the user must evaluate available actions
against their goals and state of knowledge about
the activity. In overcoming the breakdown or
conflict, emergent behaviour may not only result

in a reprioritisation of individual goals, but may

also lead to new interaction strategies.
Expectations are influential in the division
of attention during HCI, acting as a selection
mechanism, reducing demand on cognitive resources to efficiently respond to new information.
Experience is also influential in the development
of automaticity. For instance, expert system
users are expected to demonstrate better skills
compared to novices, as they are likely to have
developed more efficient strategies and greater
responsiveness to overcoming breakdowns. The
difficulties novices face may be a result of their
lack of habituated actions during HCI, rendering
the task more cognitively demanding. Situation
awareness among novice users typically increases
through gradual exposure to, and experience with,
the system. However, increased situation awareness does not imply good task performance. It is a
probabilistic construct, making good performance
more likely, yet not assured.
The notion of situation awareness is found
within several postcognitive HCI frameworks.
Here, we present a broad comparison of a few
notable frameworks. We examine how each of
the competing frameworks conceives of an appropriate unit of analysis, and how this unit of
analysis impacts each framework.

Situated Action Models

The situated action view of HCI treats knowledge
as a subjective construct. Knowledge is inexorably
linked to interpersonal communication and contextual activity. Knowledge is not only subjective,
but specifically relevant to a particular situation in
which some work practice occurs (Gasson, 1999).
Nardi (1996) notes that situated action perspectives
are more focused on context-sensitive practices
than on the cognitive properties of the artefacts
with which individuals interact.
The unit of analysis in situated action is the
motivated activity within a relevant community

911

Understanding the Nature of Task Analysis in Web Design

of practice (Lave, 1988), supplanting the largely

cognitivist unit of analysis, task. Within each of
the ecological perspectives treated in this chapter,
the notion of task is redefined or appropriated to
mean action. Subjugating the importance of task
to activity promotes the influence of actor roles,
relationships, and sociocultural cues within a
particular setting.
One of the key characteristics of the situated
action model is its adherence to reflexivity in
system design. Experience, cultural and historic
beliefs and anticipatory planning on behalf of the
individual are therefore de-emphasised within
this perspective. Rather, situated action focuses
on the immediacy of interaction, the fluctuating
conditions of the situation, and the learning opportunities provided by an individuals creative
response(s) to alternative paths through the activity (Lave, 1988; Nardi, 1996; Suchman, 1987). An
activitys structure is not something that precedes
interaction but emerges directly out of the immediacy of the situation. Planning and goals are
produced in the course of action, and thus are
work products, rather than characteristics that
orient the activity (Bardram, 1998). This view
implies that people are basically opportunistic.
Planning and goals either emerge concurrently
out of participation in the activity itself, or by
conscious reflection upon previous interaction.
As situated action subscribes to the view that
analysts should only be concerned with recording
observable behaviour (Endsley, 2000), we believe
that it is not appropriate for analysing exploratory
domains, where it is crucial to first identify the
contextual problem under investigation and subsequently determine the expected outcomes of
the activity. Similarly, we feel that the approach
is not sufficiently descriptive to examine large
collaborative work activities where knowledge
and planning are often spread across actors. In
large organisations, activity is seldom seen as a
knee-jerk reaction to current events.

912

Distributed Cognition
Distributed cognition is concerned with how
knowledge is propagated and transformed by
agents within activity. An agent is any cognitive
artefact of the system, be it human, machine, or
other work product. The unit of analysis is the
cognitive system. Distributed cognition relaxes the
assumption that the individual is the best or only
useful unit of analysis and thus extends the reach
of what is considered cognitive to both systems
that are smaller and larger than the individual
(Hollan et al., 2001; Hutchins, 1995). The cognitive system in distributed cognition is thus more
akin to the term complex cognitive system.
Goals, according to distributed cognition,
are not merely maintained within the mind of
the subject (individual or group), but rather embedded within the cognitive system. Distributed
cognition posits that artefacts may themselves
possess goals. The cognitive system can only be
understood when we know the contributions of
individual agents, their shared contributions and
collaboration strategies, and the nature agent behaviour in the environment. In contrast to situated
action, distributed cognition incorporates culture,
context and history, but from within an embedded
perspective (Hollan et al., 2001).
There are striking similarities between the
distributed cognition and activity theory (Nardi,
1996). Both are activity-centric: they recognise
activity as a hierarchical, goal-oriented structure;
align physical, verbal, and nonverbal actions
with specific goals; and they distinguish between
conscious and unconscious actions. Additionally, neither framework prescribes a particular
set of methodological practices. Nevertheless,
there are two notable differences between the
approaches:

Activity theory is essentially human-centric,

individuals motivating activity.

Understanding the Nature of Task Analysis in Web Design

In activity theory, artefacts are not seen as

goal-oriented, cogent entities.

In contrast to activity theory, distributed

cognition holds that in cooperative environments, individuals maintain only partial models
of the problem. Hence, for distributed cognition,
complex systems rather than individuals are the
appropriate unit of analysis.
Latour (1987) provides theoretical support
for this last argument, stating that part of what
is technological is social, and what is social is
technological. As relationships in actor-networks
constitute human-human and human-artefact
interaction, a change in content structure, user
interface, pedagogy, or number of participants in
an activity has the potential to change the nature
of an activity. While not implying that an artefact
is cogent, actor-network theory suggests how a
work product may impact social and cognitive
processes (Tatnall, 2002).

Activity Theory
Activity theory is a descriptive conceptual
framework that has emerged primarily from contributions by Vygotsky (1986), Leontev (1978),
and Engestrm (1987). Activity theory serves to
describe the different forms of human praxis and
developmental processes involved in HCI. Activ-

ity theory represents a truly ecological approach

to task analysis, providing a broad theoretical
framework for examining collaborative activities
(Stanton, 2004). As the name suggests, the unit
of analysis in activity theory is the activity itself.
Individual and social processes are interwoven in
the generation of contextualised meaning. Similar
to distributed cognition, individual knowledge and
meaning cannot be separated from the context of
an activity.
Activity theory (depicted in Figure 3) is an
object-oriented, task analysis framework. The
importance of goals is present, however they are
subordinate to motivated outcomes, or objective
purpose (object) (Whittaker et al., 2000). Unlike
situated action, interaction is not opportunistic,
but purposeful. Purpose is context specific as
activities define context, and context defines an
activity. Therefore, the act of doing in one context
cannot be considered congruous with the act of
doing in another (Lave & Wegner, 1991). This is
in direct contrast to cognitive task models such
as GOMS.
One of the principal tenets of activity theory
is tool mediation. Activities are mediated by individuals carrying out their tasks through the use
of available artefacts in the environment. When
an artefact is used to mediate interaction, it is
said to become a tool. A tool can be any material
or mental process used to transform or convey

Figure 3. Activity theory framework

913

Understanding the Nature of Task Analysis in Web Design

information (Kaptelinin, 1996b). As tools exist

only through purposeful activity, they contain
cultural and historical residue. They are therefore
capable of imparting external sociocultural beliefs
and historic conventions upon an activity.
Activity theory posits a hierarchical structure
of activity, consisting of three primary elements:
activity, object-oriented actions, and operations
(Leontev, 1978). As shown in Figure 3, activity is
motivated by the subject (individual or group). An
activity is comprised of mediated, object-oriented
actions. Each action is associated with one or more
parent and subgoals. This implies that actions are
polymotivated; each action (task) may be applied
to accomplish several goals. This is important as
it provides a way of supporting alternative paths
or strategies within an activity for obtaining the
purposeful outcome. Accordingly, we cannot
focus purely on the action itself, as its purpose or
role is ambiguous. Lastly, operations are habitual
behaviours or unconscious actions.
According to activity theory, there is constant
interaction between these three elements of human praxis. This dynamicism is critical to our
understanding of human behaviour in interactive
environments. However, despite the advantages of
applying an activity theoretic approach to interactive system design, there is as yet no systematic
way of interfacing the approach with rigorous
software engineering methodologies (Turner,
Turner, & Horton, 1999).

Scenario-Based Design
Scenario-based design (SBD) promotes the use
of scenarios (or structured narratives) in HCI as a
way of understanding human activity and creating
computer systems (Carroll, 2000). Today, they are
widely used across disciplines. In an attempt to
render software development more social, Carroll
(1996, 1997) argues that activity theory can be
applied effectively to SBD. While Carroll does

914

not describe how this can be accomplished, suggestions are found elsewhere (Carroll, 2000; Go
& Carroll, 2004; Kazman et al., 1996; Rosson
& Carroll, 2002). Admittedly, natural language
narratives are not the most scientific notation used
within HCI or software engineering; however, it is
often by these means that technical information is
conveyed throughout the development effort (Carroll, 1996). Moreover, scenarios provide a flexible
mechanism for integrating real life accounts of
activity with the more empirically discrete views
employed within software engineering.
During discovery and definition (see Table
1), scenarios can be used speculatively at the
start of the SDLC to document expected user
behaviour or to describe hypothetical activities.
In the absence of an existing system, this process
improves ecological validity during requirements
engineering (Stanton, 2004). To assist with requirements engineering, Kaptelinin et al. (1999)
suggest that Activity Checklists can assist practitioners to focus on salient aspects of an activity,
thereby constraining the process of requirements
engineering during SBD.
Scenarios represent purposeful interaction
within a system, and are inherently goal-oriented.
Scenarios are always particular to a specific situation, and thus situated within a particular frame
of research. Because scenarios are actor-driven,
describing interaction from the perspective of
at least one individual, they can be effectively
integrated with an activity theory framework.
Finally, as scenarios use a common notation, natural language, all stakeholders in the development
process can easily communicate requirements,
experiences, and other opinions/beliefs without
requiring extensive training. Consequently,
scenarios serve as an excellent lingua franca for
communication between all project stakeholders; a
primary goal of any good task analysis framework
(Balbo et al., 2004).

Understanding the Nature of Task Analysis in Web Design

Table 6. Overview of postcognitive task models

Approaches
Criteria
Expressive power
Complexity
Collaboration
Timing
Roles and responsibilities
Evaluation
Requires training
Scalable
Social orientation

Situated
Action

Distributed
Cognition

Activity
Theory

LOW
LOW
NO
YES
NO

LOW
LOW
NO
YES
NO

MEDIUM
MEDIUM
YES
NO
YES

ScenarioBased
Design
HIGH
HIGH
NO
NO
NO

YES
LOW
NO
NO

YES
LOW-MED
NO
NO

YES
MED-HIGH
NO
NO

YES
HIGH
YES
NO

Summary of Approaches
Table 6 provides a brief overview of the postcognitive approaches described previously and demonstrates their shared and unique attributes.

Towards Situated Task

Analysis in Web Design
In this chapter we have outlined the limitations
associated with both cognitive and postcognitive
approaches to task analysis. Cognitive approaches
suffer from their inability to model dynamic
social processes; to consider aspects of interaction which are not reducible to discrete cognitive
events; and to elicit and describe user goals and
their limited extensibility and transferability
to new domains. Postcognitive approaches on
the other hand suffer from a lack of tools and
techniques that bridge theoretical frameworks

and practical design and evaluation methods.

Although promoting increased reuse of HCI
knowledge, postcognitive frameworks are less
computationally powered, thus possessing a significant barrier to their integration into standard
Web design practices as described in Vora (1998).
The challenge therefore is to promote integrative,
task analysis frameworks that support both cognitive and ecological perspectives (Czerwinski &
Larson, 2002). This new breed of task analysis
frameworks is referred to as situated task analysis
(Farmer & Hughes, 2005b).
Situated task analysis incorporates the view
that in natural settings, problems are frequently
ill defined, requiring communication, collaboration, negotiation, and prioritisation. This broadens
the field of regard from the individual to include
organisational and social contexts. This situated
view can be seen as an example of macrocognitive architecture, describing cognitive functions
as they are performed in natural settings (Klein,

915

Understanding the Nature of Task Analysis in Web Design

Ross, Moon, Klein, Hoffman, & Hollnagel, 2003).

This is in opposition to the traditional microcognitive view of task analysis that focuses on the
building blocks of cognition. A detailed account
of a situated task analysis framework: cognition,
activity, social organisation, environment (CASE)
is provided in Farmer (2006).

A Simple Approach to Situated Task

Analysis in Web Design
A major obstacle in developing Web sites that
are both usable and fit-for-purpose is capturing
stakeholder requirements and communicating
them to other project team members, especially
designers. User requirements are often qualitatively produced, resulting in rich descriptions of
functional and operational requirements. While
this assists with understanding the nature of
activity, it does not always suit developers who
are required to describe the system in terms of
discrete, verifiable behaviour. In the following,
we describe an iterative five-step process that
facilitates communication and integration of both
static and dynamic views within the requirements engineering and design phases of Web
site development.

Step One: Requirements

Engineering and Goal Generation
The first step in our process is requirements
engineering. Scenarios are ideal in this situation as they are written in natural language and
are therefore highly accessible and require little
practitioner training (Go & Carroll, 2004).
Figure 4 describes the iterative process of eliciting user requirements using activity checklists
(Kaptelinin et al., 1999) and scenarios. Scenario
descriptions are iteratively refined until they are
stable, the outcome being a set of scenarios reflecting multiple viewpoints of interaction. Using
an initial Activity Checklist, containing likely
questions and statements related to the purpose
of the activity, and a minimal-use scenario, we
initiate the requirements elicitation process. As
shown in Figure 4, our first aim is to elicit initial
traits or themes that emerge out of the users
perception of the activity. Users may be additionally asked to develop their own use scenario to
increase participation and contextuality during
the development process. This process continues,
redefining the scenarios and checklists, until we
are able to group these traits into sets of themes
or categories. At each stage, the new scenarios are

Figure 4. Concept/goal generation during requirements engineering

916

Understanding the Nature of Task Analysis in Web Design

validated against the users themselves in order to

maintain internal consistency.
In a similar manner, once no further traits
emerge from our analysis, we proceed by reducing these initial categories to a core set of salient
concepts. These concepts can be considered
high-level objectives within the activity or goals.
These concept-goals can be considered fairly
stable and transferable between users, as they
will have been derived from multiple perspectives. Finally, we arrive at a set of scenarios that
represent the key concept-goals of our activity.
These scenarios can be linked together to provide
a chain or hierarchical view of the scenarios relative to a particular requirement. The corollary of
this process becomes increasingly important as
we move towards the initial design phase.
In Figure 5 we see more clearly this process
of scenario refinement and linkage. Our initial
minimal use scenario (R1.1) is iteratively refined
to reveal particular traits of the activity (R1.2).
As this ultimately is an exploratory process, each
stage can be seen as a form of goal refinement.
We notice in Figure 5 that within this process of

refinement, conflicts between scenario views (i.e.,

between users or between users and developers)
result in an update of the Activity Checklist.
In this way, we are able to capture the context
of use and apply this knowledge to subsequent
elicitation cycles. As we move towards conceptbased scenarios in our analysis, we move away
from action-oriented perspectives towards more
holistic, goal-oriented views of activity.

Step Two: Determine Salient

Features of Interaction
Having constructed a core set of requirements
containing both low and high-level views of
interaction, the next step is to determine which
concepts are most critical to the success of the
Web application. Although we already possess
relatively stable concept-goals with which to
initiate our design process, some concepts may be
more salient than others. Techniques for determining salient cognitive and social features during
interaction are described in Farmer (2006).

Figure 5. Iterative goal-driven scenario development

917

Understanding the Nature of Task Analysis in Web Design

Step Three: From Scenarios to UML

Use Cases
Upon establishing the most salient concept-goals
with the system, the next step is to transform
these rich textual descriptions into static, structural views, amenable to inclusion in the design
process. This can be achieved using unified
modeling language (UML), use case diagram
notation (Rumbaugh, Jacobson, & Booch, 1999).
We start by modelling the high-level use cases.
These typically correspond to the concept-goal
scenarios previously elicited, and hence provide a
goal-oriented system view. Subsequently, we drill
down into each of these use cases, decomposing
them into their subsystems and determining their
structural constraints.
Use case decomposition, as shown in Figure
6, is the process of iteratively decomposing an
activity into its constituent parts. This process
has already been achieved via the bottom-up
process of concept-goal generation. Therefore,
use case decomposition may easily be achieved

Figure 6. Example of use case decomposition

918

by associating each concept-goal with its related

categories and traits. This information is present
within the previously constructed scenario chains.
Designers are therefore able to isolate requirements at particular levels of granularity as well
as investigate the structural relationships between
use cases (implements, depends, generalises) via
the rich descriptions provided in the scenarios.
Figure 6 provides an example of a use case
decomposition for an online shopping activity.
In this instance, the user has only four conceptgoals. These are search catalogue, view shopping
cart, buy book, and change profile. Common
concepts that are shared among requirements,
such as login, can be modelled via a dependency
relationship. Modelling the relationship between
a concept and a category (see Figure 4) can be
achieved through inheritance, or implementation.
In addition to modelling relationships, designers
may choose to isolate particular concept areas,
such as social organisation, to further narrow the
design perspective.

Understanding the Nature of Task Analysis in Web Design

Step Four: From Use Cases to

Design
After constructing the UML use cases at various levels of granularity (from goals to actions),
the next step involves transforming these formal
decompositions into design elements. One way
to manage this complex process is to establish
distinct viewpoints, modelling the system from
various perspectives (Rumbaugh et al., 1999).
Here, we only consider the structural and behavioural system views. The major distinction
between these views is that the former emphasises
object-oriented interaction, whereas the latter
describes event-driven interaction.
The structural view provides a physical description of how the Web site will be made up
of various components, including relationships
between those components. This is primarily
achieved using UML class diagrams that provide
an object-oriented view of the system. While
primarily promoting a static view of the system,
object methods and attributes can also be shown.
Using the use cases generated previously from
the concept scenarios, it is possible to rapidly
construct a high-level representation of the site

structure, objects acting as placeholders for each

concept goals.
Having modelled at a conceptual level the key
user requirements as UML classes or objects, we
use the link structure provided during the initial
requirements engineering phase to determine
additional objects, methods, and attributes of the
system. For instance, we can model categorybased scenarios in UML class notation via aggregate relations (is-part-of) to our concept-based
objects (Figure 7). Additional techniques, such
as noun extraction (noun=object, verb=method,
adjective/adverb=attribute), may be used to extract structural information from the scenarios
themselves (Schach, 1999).
Figure 7 provides a simple demonstration of
this process. In this instance, the class objects
Login and Search represent complex objects associated with login management and searching a
Web site. We note these relationships could have
been modelled alternatively as methods (login
and search are verbs) within other classes, or as
interfaces that could be implemented by other
objects, as shown in Figure 6.
This process can be extended to the trait-based
scenarios to establish additional object attributes.

Figure 7. UML Class diagram representation of concepts

919

Understanding the Nature of Task Analysis in Web Design

This process has several benefits: it provides a

rapid means of determining initial site structure
within a goal-oriented framework; it permits
designers to link design decisions back to initial
requirements, as well as examining requirements
from various stakeholder points of view and at
various levels of granularity; and designers may
link additional design-centric scenarios to each
requirement, providing details such as event sequencing and information flow, therefore improving traceability between the SDLC phases.
In examining methods and attributes more
closely using our scenarios, we are likely to become more focused on event-driven behaviour.
In developing a behavioural system view, we
are interested in modelling the flow of system
information through event or function-driven
interaction. We typically see functional flows
(Appendix A) and sequence diagrams used at this
point in the initial design phase. These techniques
constrain interaction to a finite-state representation, limiting users to predetermined paths
through the system. Where multiple paths may
initially exist, functional flow analysis and task
models may be used to determine optimal paths
through the system in order to reduce variability
and improve site effectiveness and efficiency. At
this point, our design analysis has moved away
from modelling goals to modelling actions and
evaluating task performance.
Finally, these event-driven views may be abstracted to form UML dialog models or activity
diagrams, providing a means of aggregating task
models or flow analyses. These dialog models
may then be linked into our initial scenario/requirements chain to provide additional system
information. This process ensures a multi-view
representation of requirements that are linked
forwards and backwards to design. Moreover, it
ensures that both goal-oriented and action-oriented perspectives of interaction are conveyed
during development.

920

Step Five: Start Over

We recall that task analysis is not a once-off process (Balbo et al., 2004). This is especially true
of interactive systems and environments, such as
the Web. It is important to note that most people
actually learn by doing (Lave & Wegner, 1991).
Rather than thinking of a sites design as merely
being usable, we should consider how the site or
application, as a tool, facilitates participation in
an activity.
Developing effective Web sites requires consideration of both novice and expert users. Most
users, however, do not arrive at a site as experts.
They reason, use past experiences, and explore
through interaction. Their ability to gain expertise in these virtual environments is partially
governed by how well the design supports and
extends the users current capabilities. This can
only be achieved when there is sufficient coherence between the context of use and the virtual
environments design. Taking a user-centred approach to Web design, including an appropriate
task analysis framework and set of methodological
practices, ensures that developers will never be
too far away from their users. Ultimately, it is
they who will judge the relative merits of your
site/application.

Conclusion
Web design is a complex activity at the best of
times. Not only do designers frequently encounter technological limitations imposed by a novel
communication medium, but also they are highly
isolated from their users. Moreover, arguably
more than with any other technological medium,
the Web application target audience is extremely
heterogeneous. Therefore, the ambiguous and
diverse nature of Web application use imposes
critical limitations on Web design practices.
Shadowed by the importance of developing

Understanding the Nature of Task Analysis in Web Design

quality, fit-for-purpose systems, this chapter has

highlighted the considerable benefits to be gained
by incorporating task analysis in Web design.
Specifically, this chapter has described how task
analysis is not a single process that can be applied
indiscriminately of context and use. It is a highly
stylised and domain-dependent activity that occasionally suffers from conflicting theoretical and
methodological approaches.
We have argued that task analysis is not a
silver bullet, rather a means of getting to know
your users and communicating this information
through product development and evaluation.
Rather than describing a core set of prescriptive
guidelines for conducting task analysis in Web
design, this chapter has set out to inform the
reader via a critical examination and principled
discussion of various approaches to task analysis.
Consequently, our hope is that the reader as practitioner will be both more aware and appreciative
of the role task analysis can and should play in
Web design.

Acknowledgments
We would like to thank Baden Hughes and Sandrine Balbo for their collaboration and comments
on related work. Special thanks go to Sandrine
Balbo for her presentation on task models in
HCI, which spawned an insightful and lively
afternoon of discussions that stimulated aspects
of our discussion here.

References
Albers, M. J. (1998). Goal-driven task analysis:
Improving situation awareness for complex
problem solving. In Proceedings of the 16th
Annual International Conference on Computer

Quebec, Canada:
Documentation (pp. 234-242).
ACM Publishing.

Annett, J. (2004). Hierarchical task analysis. In

D. Diaper & N. Stanton (Eds.), The handbook of
task analysis for human-computer interaction
(pp. 67-82). Mahwah, NJ: Lawrence Erlbaum
Associates.
Annett, J., & Duncan, K. (1967). Task analysis
and training design. Occupational Psychology,
41, 211-227.
Balbo, S., Ozkan, N., & Paris, C. (2004). Choosing
the right task-modelling notation: A taxonomy.
In D. Diaper & N. Stanton (Eds.), The handbook
of task analysis for human-computer interaction
(pp. 435-465). Mahwah, NJ: Lawrence Erlbaum
Associates.
Bardram, J. E. (1998). Plans as situated action:
An activity theory approach to workflow systems.
In J. A. Hughes, W. Prinz, T. Rodden, & K. I.
Schmidt (Eds.), Proceedings of the 5th European
Conference on Computer Supported Cooperative
Work (pp. 17-32). Dordrecht, The Netherlands:
Kluwer Academic Publishers.
Bomsdorf, B., & Szwillus, G. (1998). From task
to dialogue: Task-based user interface design.
SIGCHI Bulletin, 30(4), 40-42.
Bonaceto, C., & Burns, K. (2003). Mapping the
mountains: A survey of cognitive engineering
methods and uses. In R. Hoffman (Ed.), Expertise
out of context: Proceedings of the 6th Conference
on Naturalistic Decision Making. Hillsdale, NJ:
Lawrence Erlbaum Associates.
Card, S. K., Moran, T. P., & Newell, A. L. (1983).
The psychology of human computer interaction.
Hillsdale, NJ: Erlbaum.
Carroll, J. M. (1996). Becoming social: Expanding
scenario-based approaches in HCI. Behaviour &
Information Technology, 15(4), 266-275.
Carroll, J. M. (1997). Human-computer interaction:
Psychology as a science of design. International
Journal of Human-Computer Studies, 46, 501522.

921

Understanding the Nature of Task Analysis in Web Design

Carroll, J. M. (2000). Five reasons for scenariobased design. Interacting with Computers, 13,
43-60.
Carroll, J. M. (2002). Making use is more than
a matter of task analysis. Interacting with
Computers, 14, 619-627.
Checkland, P. (1999). Soft systems methodology
in action. New York: John Wiley & Sons Ltd.
Corbel, C., Gruba, P., & Enright, H. (2002). Taking
the Web to task. Sydney, Australia: National Centre
for English Language Teaching and Research,
Macquarie University.
Czerwinski, M. P., & Larson, K. (2002). Cognition
and the Web: Moving from theory to Web
design. In J. Ratner (Ed.), Human factors and
Mahwah,
NJ:
Web development (pp. 147-165).
Erlbaum.
Diaper, D. (2004). Understanding task analysis for
human-computer interaction. In D. Diaper & N.
Stanton (Eds.), The handbook of task analysis for
human-computer interaction (pp. 5-49). Mahwah,
NJ: Lawrence Erlbaum Associates.
Dix, A. (2005). Human-computer interaction
and Web design. In R.W. Proctor & K.-P. L. Vu
(Eds.), Handbook of human factors in Web design
(pp. 28-47). Mahwah, NJ: Lawrence Erlbaum
Associates.
Endsley, M. R. (2000). Theoretical underpinnings
of situation awareness: A critical review. In M.
R. Endsley & D. J. Garland (Eds.), Situation
awareness analysis and measurement (pp. 3-33).
Mahwah, NJ: Lawrence Erlbaum Associates.
Engestrm, Y. (1987). Learning by expanding.
Helsinki, The Netherlands: Orienta-Konsultit.
Erickson, T., & Kellogg, W. A. (2001). Social

translucence: Designing systems that support

social processes. In J. M. Carroll (Ed.), Humancomputer interaction in the new millennium (pp.
324-336). London: Addison-Wesley.

922

Farmer, R. A. (2005). Multimodal speech recognition errors and second language acquisition: An
activity theoretic account. In Proceedings of the
6th Conference on Using Technology in Foreign
Language Teaching. Compiegne, France: Universit de Technologie Compigne.
Farmer, R. A. (2006). Situated task analysis in
learner-centered CALL. In P. Zaphiris & G.
Zacharia (Eds.), User-centered computer-assisted
language learning (pp. 43-73). Hershey, PA: Idea
Group Publishing.
Farmer, R. A., Gruba, P., & Hughes, B. (2004).
Towards principles for CALL software quality
improvement. In J. Colpaert, W. Decoo, M.
Simons, & S. Van Beuren (Eds.), CALL and
Research Methodologies: Proceedings of the 11th
International Conference on CALL (pp. 103-113).
Belgium: Universiteit Antwerpen.
Farmer, R. A., & Hughes, B. (2005a). A situated
learning perspective on learning object design. In
P. Goodyear, D. G. Sampson, D. Yang, Kinshuk,
T. Okamoto, R. Hartley, et al. (Eds.), Proceedings
of the 5th International Conference on Advanced
Learning Technologies, Kaohsiung, Taiwan (pp.
72-74). CA: IEEE Computer Society Press.
Farmer, R. A., & Hughes, B. (2005b). A classification-based framework for learning object
assembly. In P. Goodyear, D. G. Sampson, D.
Yang, Kinshuk, T. Okamoto, R. Hartley et al.
(Eds.), Proceedings of the 5th International Conference on Advanced Learning Technologies (pp.
4-6). Kaohsiung, Taiwan. CA: IEEE Computer
Society Press.
Farmer, R. A., & Hughes, B. (2005c). CASE: A
framework for evaluating learner-computer interaction in Computer-Assisted Language Learning.
In B. Plimmer & R. Amor (Eds.), Proceedings
of CHINZ 2005 Making CHI Natural, Auckland, New Zealand (pp. 67-74). New York: ACM
Press.

Understanding the Nature of Task Analysis in Web Design

Filkes, R. E. (1982). A commitment-based framework for describing cooperative work. Cognitive

Science, 6, 331-347.
Flor, N., & Hutchins, E. (1991). Analysing
distributed cognition in software teams: A case
study of team programming during perfective
software maintenance. In J. KoenemannBelliveau, T. Moher, & S. Roberson (Eds.),
Proceedings of the 4th Annual Workshop on
Empirical Studies of Programmers (pp. 36-59).
Norwood, NJ: Ablex Publishing.
Gasson, S. (1999). A social action model of situated
information systems design. The DATA BASE for
Advanced in Information Systems, 30(2), 82-97.
Go, K., & Carroll, J. M. (2004). Scenario-based
task analysis. In D. Diaper & N. Stanton (Eds.), The
handbook of task analysis for human-computer
interaction (pp. 117-134). Mahwah, NJ: Lawrence
Erlbaum Associates.
Goguen, J. A. (1996). Formality and informality
in requirements engineering. In Proceedings of
the 2nd International Conference on Requirements
Engineering (ICRE96) (pp. 102-109). Silver
Spring, MD: IEEE Computer Society Press.
Gordon, S. E., & Gill, R. T. (1997). Cognitive task
analysis. In C. E. Zsambok & G. Klein (Eds.),
Naturalistic decision making (pp. 130-140).
Mahwah, NJ: Lawrence Erlbaum Associates.
Hollan, J., Hutchins, E., & Kirsch, D. (2001).
Distributed cognition: Toward a new foundation
for human-computer interaction research. In J.
M. Carroll (Ed.), Human-computer interaction
in the new millennium (pp. 77-94). London:
Addison-Wesley.
Hollnagel, E. (Ed.). (2003). Handbook of cognitive
task design. Mahwah, NJ: Lawrence Erlbaum
Associates.
Hutchins, E. (1995). Cognition in the wild.
Cambridge, MA: MIT Press.

John, B. E., & Kieras, D. E. (1996a). Using GOMS

for user interface design and evaluation: Which
technique? ACM Transactions on ComputerHuman Interaction, 3, 287-319.
John, B. E., & Kieras, D. E. (1996b). The GOMS
family of user interface analysis techniques:
Comparison and contrast. ACM Transactions on
Computer-Human Interaction, 3, 320-351.
Kaptelinin, V. (1996a). Activity theor y:
Implications for human-computer interaction.
In B. A. Nardi (Ed.), Context and consciousness:
Activity theory and human-computer interaction
(pp. 103-116). Cambridge, MA: MIT Press.
Kaptelinin, V. (1996b). Computer-mediated
activity: Functional organs in social and
developmental contexts. In B. A. Nardi (Ed.),
Context and consciousness: Activity theory
and human-computer interaction (pp. 45-69).
Cambridge, MA: MIT Press.
Kaptelinin, V., Nardi, B., & Macaulay, C. (1999,
July/August). The activity checklist: A tool for
representing the space of context. Interactions,
27-39.
Kazman, R., Abowd, G., Bass, L., & Clements,
P. (1996, November). Scenario-based analysis of
software architecture. IEEE Software, 47-53.
Kieras, D. (2004). GOMS models for task analysis.
In D. Diaper & N. Stanton (Eds.), The handbook
of task analysis for human-computer interaction
(pp. 83-116). Mahwah, NJ: Lawrence Erlbaum
Associates.
Kirwan, B., & Ainswoth, L. K. (1992). A guide to
task analysis. London: Taylor & Francis.
Klein, G., Ross, K. G., Moon, B. M., Klein, D.
E., Hoffman, R. R., Hollnagel, E. (2003, May/
June). Macrocognition. IEEE Intelligent Systems,
18(3)81-84.

923

Understanding the Nature of Task Analysis in Web Design

Latour, B. (1987). Science in action: How to follow

engineers and scientists through society. Milton
Keynes, UK: Open University Press.

Newman, W. M., & Lamming, M. G. (1998).

Interactive system design. Essex, UK: AddisonWesley.

Lave, J. (1988). Cognition in practice: Mind,

mathematics and culture in everyday life.
Cambridge, UK: Cambridge University Press.

Norman, D. A. (1999, May/June). Affordance,

conventions and design. Interactions, 6(3), 3843.

Lave, J., & Wegner, I. (1991). Situated learning:

Legitimate peripheral participation. Cambridge,
UK: Cambridge University Press.

Patern, F., & Santoro, C. (2002). Integrated

support based on task models for the design,
evaluation, and documentation of interactive
safety-critical systems: A case study in the air
traffic control domain. International Journal of
Systems Science, 33(6), 513-527.

Leontev, A. N. (1978). Activity, consciousness,

and personality. Englewood Cliffs, NJ: Prentice
Hall.
Limbourg, Q., & Vanderdonckt, J. (2004).
Comparing task models for user interface design.
In D. Diaper & N. Stanton (Eds.), The handbook
of task analysis for human-computer interaction
(pp. 135-154). Mahwah, NJ: Lawrence Erlbaum
Associates.
Mager, R. F. (1991). Goal analysis. London:
Kogan Page.
Miller, R. B. (1953). A method for man-machine
task analysis (TR 53-137). Wright Patterson
Air force Base, OH: Wright Air Development
Center.
Miller, R. B. (1962). Task description and analysis.
In R. M. Gagne (Ed.), Psychological principles
187-228)
. New York:
in system development (pp.
Hold, Rinehart, & Winston.
MITRE. (2003). Mental models for naturalistic
decision making. Centre for Air Force Command
and Control. Retrieved October 12, 2005, from
http://mentalmodels.mitre.org/
Nardi, B. A. (1996). Studying context: A
comparison of activity theory, situated action
models, and distributed cognition. In B. A. Nardi
(Ed.), Context and consciousness: activity theory
and human-computer interaction (pp. 69-102).
Cambridge, MA: MIT Press.

924

Preece, J., Rogers, Y., Sharp, H., Benyon, D.,

Holland, S., & Carey, T. (Eds.). (1994). Humancomputer interaction. London: AddisonWesley.
Raeithel, A., & Velichkovsky, B. M. (1996). Joint
attention and co-construction of tasks: New ways
to foster user-designer collaboration. In B. A.
Nardi (Ed.), Context and consciousness: Activity
theory and human-computer interaction (pp.
199-234). Cambridge, MA: MIT Press.
Rasmussen, J., Pejtersen, A. M., & Goodstein, L.
P. (1994). Cognitive systems engineering. New
York: John Wiley & Sons.
Robinson, M. (1993). Design for unanticipated
use. In G. D. Michelis, C. Simone, & K. Schmidt
(Eds.), Third European Conference on Computer
Supported Cooperative Work (187-202). Dordrecht, The Netherlands: Kluwer.
Rosson, M. B., & Carroll, J. M. (2002). Usability
engineering: Scenario-based development of
human-computer interaction. San Diego, CA:
Academic Press.
Rumbaugh, J., Jacobson, I., & Booch, G. (1999).
The unified modelling language: User guide.
Upper Saddle River, NJ: Addison-Wesley.

Understanding the Nature of Task Analysis in Web Design

Schach, S. R. (1999). Classical and objectoriented software engineering (4th ed.). Singapore:
McGraw-Hill International.
Shepherd, A. (2001). Hierarchical task analysis.
London: Taylor & Francis.
Shrayne, N. M., Westerman, S. J., Crawshaw, C.,
Hockey, G. R. J., & Sauer, J. (1998). Task analysis
for the investigation of human error in safetycritical software design: A convergent methods
approach. Ergonomics, 41(11), 1719-1736.
Smart, K. L., Rice, J. C., & Wood, L. E. (2000).
Meeting the needs of users: Toward a semiotics
of the web. In Proceedings of the 18th Annual
ACM International Conference on Computer
Documentation: Technology & Teamwork (pp.
593-605). Piscataway, NJ: IEEE Educational
Activities Department.
Sommerville, I. (2004). Software engineering
(7th ed.). New York: Addison-Wesley.
Stanton, N. (2004). The psychology of task analysis
today. In D. Diaper & N. Stanton (Eds.), The
handbook of task analysis for human-computer
interaction (pp. 569-584). Mahwah, NJ: Lawrence
Erlbaum Associates.
Suchman, L. A. (1987). Plans and situated actions:
The problem of human-computer communication.
New York: Cambridge University Press.
Sutcliffe, A. (2000). On the effective use and
reuse of HCI knowledge. ACM Transactions on
Computer-Human Interaction, 7(2), 197-221.
Tatnall, A. (2002). Actor-network theory as a
socio-technical approach to information systems
research. In S. Clarke, E. Coakes, M. G. Hunter,
& A. Wenn (Eds.), Socio-technical and human
cognition elements of information systems (pp.
266-283). Hershey, PA: Idea Group.
Turner, P., Turner, S., & Horton, J. (1999).
From description to requirements: An activity

theoretic perspective. In Proceedings of the

International ACM SIGGROUP Conference on
Phoenix,
Supporting Group Work (pp. 286-295).
AZ: ACM.
Van Der Veer, G. C., Lenting, B. F., & Bergevoet, B.
A. J. (1996).
Groupware task analysis: Modelling
complexity. Acta Psychological, 91, 297-322.
Vetere, F., Gibbs, M. Kjeldskov, J., Howard, S.,
Mueller, F., Pedell, S., et al. (2005). Mediating
intimacy: Designing technologies to support
strong-tie relationships. In Proceedings of CHI
2005 (pp. 471-490). Portland, OR: ACM Press.
Vicente, K. J. (1999). Cognitive work analysis:
Toward safe, productive, and healthy computerbased work. Mahwah, NJ: Lawrence Erlbaum
Associates.
Vicente, K. J. (2004). The human factor
revolutionizing the way people live with
technology. New York: Routledge
Vora, P. R. (1998). Designing for the Web: A
survey. Interactions, 5(3), 13-30.
Vygotsky, L. S. (1978). Mind and society.
Cambridge, MA: Harvard University Press.
Vygotsky, L. S. (1986). Though and language.
Cambridge, MA: Harvard University Press.
Watts, L. A., & Monk, A. F. (1998). Reasoning
about tasks, activities and technology to support
collaboration. Ergonomics, 4(11), 1583-1606.
Wegner, E. (2003). Communities of practice:
Learning, meaning and identity. Cambridge, UK:
Cambridge University Press.
Whittaker, S., Terveen, L., & Nardi, B. A.
(2000). Lets stop pushing the envelope and start
addressing it: A reference task agenda for HCI.
Human-Computer Interaction, 15, 75-106.
Wild, P. J., Johnson, H., & Johnson, P. (2003).
Understanding task grouping strategies. In S.

925

Understanding the Nature of Task Analysis in Web Design

McDonald, Y. Waern, & G. Cockton (Eds.), People

Bath University,
and computers XIV (pp. 3-21).
UK: Springer-Verlag.

926

Zsambok, C., & Klein, G. (Eds.).

(1997).
Naturalistic decision making. Mahwah, NJ:
Lawrence Erlbaum Associates.

High

Medium
Method

Low

Cognitive Task Analysis

Cognitive and Task-Knowledge Structures (TKS)
HTA
CognitiveOriented task GOMS
Analysis
Cognitive Work Analysis (CWA)
Applied Cognitive Work Analysis
Interviews
Think-Aloud Protocols
Mock-ups, Prototypes, Storyboards
Knowledge
Wizard of Oz Technique
Elicitation and
Cluster Analysis/Repertory Grids
Structuring
Cognitive Ladders
Cognitive walk-throughs
Functional Flow Analysis
Timeline Analysis
Performance
Operator Function Model (OFM)
Analysis
Perceptual Control Theory (PCT)
Approach
Activity Theory
Descriptive,
Situated Action
Goal-Oriented
Distributed Cognition
Analysis
Scenario-Based Design
Human ReliFailure Modes and Effects Analysis
ability and
Work Safety Analysis
Critical InciGeneric Error Modelling Systems
dent Analysis
Category

Category

Key (applicability)
Concept Require- Task Interface Work- Training Problem
Definiments Design Develop- load
Develop- Investigation
Analysis
ment
Estimament
tion
tion

Understanding the Nature of Task Analysis in Web Design

Appendix

Task analysis matrix adapted from a survey conducted by Bonaceto and Burns (2003).

This work was previously published in Human Computer Interaction Research in Web Design and Evaluation, edited by P.
Zaphiris, pp. 78-107, copyright 2007 by Information Science Publishing (an imprint of IGI Global).

927

928

Chapter 2.31

Designing for Tasks in

Ubiquitous Computing:
Challenges and Considerations
Stephen Kimani
Jomo Kenyatta University of Agriculture and Technology, Kenya
Silvia Gabrielli
University of Rome La Sapienza, Italy
Tiziana Catarci
University of Rome La Sapienza, Italy
Alan Dix
Lancaster University, UK

Abstract
The traditional desktop computing paradigm has
had major successes. It also should be noted that we
are in a day and age where many good computer
and device users are increasingly finding themselves being required to perform their activities
not in offices/desktops but in real-world settings.
Ubiquitous computing can make possible in the
real-world setting what would have otherwise
been impossible through desktop computing.
However, there is a world of difference between
the real-world and the desktop settings. The move

from the desktop to the real-world settings raises

various issues when we consider the nature of tasks
that the ubiquitous devices/applications would be
expected to support and the real-world context in
which they will be used. A careful study of the
nature of tasks in ubiquitous computing can make
some design requirements in the development
of ubiquitous applications more evident. This
chapter proposes ubiquitous application design
and evaluation considerations emerging from a
deeper understanding of the nature of tasks in
ubiquitous computing.

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Designing for Tasks in Ubiquitous Computing

Introduction
It is worth acknowledging that the traditional desktop computing paradigm has had major successes.
On the same note, it should be observed that we
are in a day and age where many people have become good computer and device users. However,
these users are increasingly finding themselves
performing or being required to (or having to)
perform their activities not in offices and desktops
but in the real world settings. In describing the
situation, Kristoffersen and Ljungberg indicate
that the hands of such users are often used to
manipulate physical objects, as opposed to users
in the traditional office setting, whose hands are
safely and ergonomically placed on the keyboard.
(Kristoffersen & Ljungberg, 1999). It is interesting
to observe how ubiquitous computing can come
in handy toward making possible in the natural
setting what would have otherwise been impossible through the desktop computing paradigm.
It is therefore not uncommon to encounter a user
who carries out one or many parallel activities
from virtually anywhere at anytime while at the
same time interacting with other user(s) and/or
device(s). (Bertini et al., 2003).
However, it is worth noting that there is a world
of difference between the real world setting and
the desktop setting. As we consider the move
from desktop computing (fixed user interfaces)
to the real world settings, various issues and
demands arise when we consider the nature of
tasks the ubiquitous devices/applications (and thus
ubiquitous user interfaces) would be expected to
support and the real world context in which they
will be used.
Consequently, it does turn out that a careful
study of the nature of tasks in ubiquitous computing can make some requirements in the design and
evaluation of ubiquitous applications become more
evident, which forms the basis of this chapter. In
particular, we will describe the nature of tasks
in ubiquitous computing, and then propose and

describe ubiquitous application user interface

design and evaluation considerations emerging
from a deeper understanding of the nature of
tasks in ubiquitous computing.
The rest of the chapter is organized as follows;
it first provides some background knowledge.
It then gives an overview of the nature of tasks
in ubiquitous computing. After that we propose
and describe ubiquitous application design and
evaluation considerations respectively based on
the foregoing. We then highlight some open issues
and conclude the chapter.

Background Knowledge
In this section, we describe some of the key concepts relevant to the chapter. In particular, we
describe ubiquitous computing. It should be noted
that in the history of computing, the requirement
to take into consideration the real world context
has arguably never been more critical and pressing
than in this day and age of ubiquitous computing.
After describing ubiquitous computing, we then
focus the description on the concept of context.

Ubiquitous Computing
Weiser coined the term ubiquitous computing (ubicomp) and gave a vision of people and
environments augmented with computational
resources that provide information and services
when and where desired (Weiser, 1991). Dix et al.
define ubicomp as: Any computing activity that
permits human interaction away from a single
workstation (Dix et al., 2004). Since then, there
have been tremendous advances in mobile and
wireless technologies toward supporting the envisioned ubiquitous and continuous computation
and, consequently, ubiquitous applications that
are intended to exploit the foregoing technologies
have emerged and are constantly pervading our
life. Abowd et al. in (Abowd et al., 2000) observe

929

Designing for Tasks in Ubiquitous Computing

that ubicomp applications are characterized by

the following:

Natural interfaces: Supporting interaction

techniques that permit humans to interact
with computing machines through the use
of more natural interaction paradigms (e.g.,
speech, gesture, pen writing).
Context-awareness: Ubicomp applications
are expected to exploit the whole set of
computing and telecommunication technologies that operate taking into account
the context.
Automatic capture of live experiences:
Ubicomp applications often adopt or provide
techniques that enable the user to record elements of their live experiences (e.g., photos,
video, audio) and the management of the
same.

Context
Context has been defined as any information
that can be used to characterize the situation of
an entity. (Dey, 2000), where an entity refers
to a person, place, or object that is considered
relevant to the interaction between a user and an
application, including the user and applications
themselves. (Dey, 2000). Context entails aspects
such as location, infrastructure, user, environment, entities, and time. The infrastructure could
include technical resources such as server and network capabilities and connections, applications,
and so forth. User includes user data/profile, usage
patterns, and so forth. The environment refers
to the physical condition of the setting an could
include light, temperature, and so on. Entities refer
to people, devices and objects. Time could include
date, time of the day, season, and so on. Abowd
et al. provide in (Abowd et al., 2000) a review of
ubicomp research and summarize context in the
form of five W s:

930

Who: As human beings, we tailor our activities and recall events from the past based
on the presence of other people.
What: Perceiving and interpreting human
activity is a difficult problem. Nevertheless,
interaction with continuously worn, contextdriven devices will likely need to incorporate
interpretations of human activity to be able
to provide useful information.
Where: In many ways, the where component of context has been explored more than
the others. Of particular interest is coupling
notions of where with other contextual
information, such as when.
When: With the exception of using time as an
index into a captured record or summarizing
how long a person has been at a particular
location, most context-driven applications
are unaware of the passage of time. Of
particular interest is the understanding of
relative changes in time as an aid for interpreting human activity. Additionally, when
a baseline of behavior can be established,
action that violates a perceived pattern would
be of particular interest.
Why: Even more challenging than perceiving what a person is doing is understanding
why that person is doing it.

Nature of Tasks
The interaction of the user with the ubiquitous
device/application can be viewed in at least two
dimensions:

user-ubiquitous application interaction

dimension and
user-ubiquitous device dimension.

User-ubiquitous application interaction dimension entails tasks in which the user is primarily
interacting with the ubiquitous application and

Designing for Tasks in Ubiquitous Computing

the device I/O modalities in order to access services such as support services (e.g., emergencies,
help/service), information services (e.g., gathering/recording information, accessing/retrieving
information, sharing information, communicating) and entertainment services (e.g., games,
music, videos). User-ubiquitous device dimension
categorizes tasks that entail the actual handling
of the device (such as holding the device, wearing
the device, attending to the device).
There are situations whereby interaction with
the ubiquitous application, though important, is
not the primary task but rather a secondary/supplementary task. In such a case, such ubiquitous
devices/applications would be used to provide
support/assistance and gather/make available
some resources (such as information) on behalf
of a user who is engaged in another primary task
in the real environment/setting. In fact, the tension between the primary tasks of a ubiquitous
user and the users interaction with the ubiquitous
device/application can be seen in the literature
(e.g., Pascoe et al., 2000). Notwithstanding the
foregoing, there also are situations in which
interaction with the ubiquitous application is the
primary contributor to the users accomplishment
of the primary task; interacting with the ubiquitous device/application can be viewed as directly
carrying out the primary task. In this case, the
use of the ubiquitous device/application tends to
be more intimately connected with what the user
is really doing (or intends to achieve) in his/her
embodied/physical self. For instance, where the
user is using the ubiquitous application to inform
him/her about the location he/she is in. However,
it is also worth pointing out that at different time
granularities, the primary task and secondary task
may swap in ubiquitous computing. The foregoing
situations raise challenges and that would need to
be taken into consideration when designing and
evaluating (developing) ubiquitous application
user interfaces.

Some ubiquitous interactions are low-intention or incidental interactions--that is one where

the user is not focused on the interaction being
supported by the ubiquitous system and may not
even be directly aware of the support (Dix et al.,
2004, Dix, 2002). In such cases, the primary task
or purposeful task needs to be understood in order
to interpret sensor data. However, the supported
task also needs to be understood in order to be
able to administer appropriate support. This
distinction is likely to be useful in other context
sensitive situations. For example, if a user has
recently consulted an online diary related to a
coming birthday and a short while later starts to
initiate a phone call, then it may be appropriate
for the number of the person whose birthday is
coming to be at the top of the list of suggested
numbers. The former task, looking up the diary
entry, is the initial purposeful task and needs to
be interpreted in order to determine which people
are important or relevant. The latter task, phoning, is the supported task and the knowledge of
which people are currently significant is used to
support this task.
While simpler task models assume a level
of pre-planning or training, it is expected that
most ubiquitous interactions are worked out at
the moment (see situated action below) or maybe
triggered from things in the environment (Dix
et al., 2004b). Furthermore, being in a particular
place or meeting a particular person may prompt
tasks or activities that were not particularly waiting to be done, but were either very low priority but suggested by having the right resources
available. Whilst more goal-oriented models of
tasks assume one gathers resources to perform a
task, in many real-world situations this gathering
of resources is the hard or expensive thing, and
it is worth doing activities using the resources
and location available, even preparatory just in
case activities.

931

Designing for Tasks in Ubiquitous Computing

Design Considerations
It is worth noting that the user of a ubiquitous
device often has to focus on more than one task
because s/he might have to interact with the
device itself (which is itself a task) while probably performing another task in the real world
setting (where this could be the primary task or
the secondary task). On one hand, interaction
with the ubiquitous device/application to some
extent requires users innate resources (such as
attention). On the other hand, the latter task often
too does require the users physical, visual, and
cognitive involvement/resources (such as hands,
visual attention, mental focus). The users physical, visual, and cognitive involvement/resources
are therefore likely to get constrained. Ideally, the
ubiquitous application (including interactions with
the device) should support the user in carrying
out that which is the primary task without supporting the user in tampering with the primary
task. We should minimize distracting the user
from the primary task or disrupting the users
primary task, unless the disruption/distraction
is of genuine (and great) value or of critical importance. In the words of Holland and Morse: It
is important that the critical focus of the users
attention be directed towards the primary task
at hand (Holland & Morse, 2001). In adopting
ways to meet the requirement, it is also critical
to consider the status of a users attention in the
timing of the tasks on the ubiquitous device. Borrowing from a research effort on guidelines for
using agents and direct manipulation (Horvitz,
1999), it is important to consider the costs and
benefits of deferring action to a time when action
will be less distracting. Where necessary, the
ubiquitous application should enable/allow the
user to temporarily halt a task on the device and
to resume the interrupted task.
One of the challenges with a new or innovative
technology/application is that its users may try
to use it in situations or ways the designers and

932

developers had never thought of. This is true in

mobile computing (Gorlenko & Merrick, 2003).
There is therefore a sense in which the user may
perform tasks on the device (and otherwise) in
unpredictable and opportunistic ways. Taking
into account all possible scenarios of use for a
product is a non-trivial challenge to the ubiquitous application analysts and designers. It is
also worth observing that the variability of the
environment/natural setting may affect the course
of a task. Therefore, analysts and designers may
also need to account for such variability in the
task analysis (Gorlenko et al., 2003).
The model human processor model (Card et
al., 1983) has been a benchmark for a lot of work
in HCI. The model is a simplified view of the
human processing while interacting with computers. It focuses on the internal cognition driven
by the cooperation of the perceptual system, the
motor system, and the cognitive system. Each
of the systems maintains its own processing
and memory. However, as the role and domain
of the computers (and devices) have widened,
researchers and designers have been considering
theories and approaches that take into account
the relationship between the internal cognition
and the outside world (Dix et al., 2004). Among
these, researchers are exploring the following
three main understandings of cognition for possible application in ubiquitous computing; they
are activity theory model, situated action model,
distributed cognition model (Abowd et al., 2002),
and even their variants.

Activity Theory
The activity theory model provides a broad conceptual framework for describing the structure,
development, and context of computer-supported
activities. It was developed by the Russian psychologists Vygotsky, Rubinshtein, Leontev and
others (Kaptelinin et al., 1995; Leontev, 1978).
Activity theory is comprised of a set of basic

Designing for Tasks in Ubiquitous Computing

principles that constitute a general conceptual

system, rather than a highly predictive theory.
The principles include the hierarchical structure
of activity, object-orientedness, internalization/
externalization, tool mediation, and development.
It should be noted that the principles should be
considered as an integrated system, because they
are associated with various aspects of the whole
activity. In activity theory, the unit of analysis is an
activity. The activity is directed at an object which
motivates the activity, giving it a specific direction.
An activity is made up of goal-directed actions that
must be undertaken to fulfill the object. Different
actions may be undertaken to meet the same goal.
Actions are conscious and they are implemented
through automatic operations. Operations do not
have their own goals, but rather they provide an
adjustment of actions to suit current situations.
Therefore, the constituents of activity are not
fixed, but can dynamically change (or adjust) as
situations/conditions change. This principle is of
great interest in ubiquitous computing, since it is
desired that the ubiquitous application appropriately adapt to the changing conditions/context.
In the context of activity theory, the principle of
object-orientedness states that human beings live
in a reality that is objective in a broad sense; the
things that constitute this reality have not only the
properties that are considered objective according to natural sciences, but also socially/culturally defined properties as well. The principle of
object-orientedness is very relevant to ubiquitous
computing since the ubicomp to a great extent
leads to situations where the user directly interacts
with other people while (at the same time) using
the ubicomp device or application. Therefore the
social and cultural issues become even more crucial. An example is societys perspective regarding
a persons speaking on a cellphone while directly
interacting with another person. Internalization
is the transformation of external activities into
internal ones. Externalization transforms internal activities into external ones. Activity theory
emphasizes that internal activities cannot be

understood if they are analyzed separately from

external activities, because they transform into
each other. The external activities in this case can
be closely associated with the contextual aspects
in ubiquitous computing. For instance, the way the
activity of speaking on the cellphone is designed
could be better informed by considering the
contextual aspects such as the simultaneous but
direct interaction with another person, the noise
level in the locality, and so forth. Activity theory
emphasizes that human activity is generally mediated by tools. Tools are created and transformed
during the development of the activity itself.
Tools carry with them a particular culture, and
therefore the use of tools is an accumulation and
transmission of social knowledge. In ubiquitous
computing, such tools could in a way be viewed
as the ubicomp devices and applications. As far
as activity theory is concerned, development is
not only an object of study, but also a general
research methodology. Gay and Hembrooke have
noted a weakness in the original formulation of
the activity theory model by pointing out that
the model has traditionally been understood as
asynchronic, point-in-time depiction of an activity (Gay & Hembrooke, 2003).
They go on to note that the model does not
depict the transformational and developmental
processes that provide the focus of much recent
activity theory research (Gay & Hembrooke,
2003). In (Boer et al., 2002), Boer et al. do propose
an extension of activity theory across time and the
levels of an organization to explain connections
between different activities as well as the influence that an activity may have on itself. Moreover,
Boer et al. also consider the role that an activity
may play in other activities at different levels of
analysis. Those extensions to the activity theory
can serve at least two purposes; they can help to
explain tensions present in real-world systems
and yield a model with a greater degree of agility
in representing complex, distributed cognition.
Other work (Uden, 2007) describes how activity
theory was used to develop a framework for the

933

Designing for Tasks in Ubiquitous Computing

design of a context-aware mobile learning application. Pinto and Jose (2006) propose ActivitySpot,
a ubicomp framework for localized activities
such as activities that are strongly related to a
specific physical environment and that only can
be achieved there. The framework defines a conceptual model that has been inspired by activity
theory model. In their attempt to develop a context
model for ubiquitous computing, Kaenampornpan
and ONeill in (2004) have relied extensively on
activity theory. They give the following three
reasons for using activity theory:

934

Activity theory provides a simple standard

form for describing human activity. It acknowledges that although, as fellow humans,
we cannot fully understand the full moment-to-moment richness of other humans'
activities, states, goals, and intentions, we do
manage to interact and to interpret others'
actions with an enormously higher degree
of success than any existing context-aware
computer based system. Therefore, in attempting to produce better context-aware
systems, it is neither possible nor necessary
to model all the richness of human activity.
Activity theory takes into account the
concepts of tool mediation and social environment, which are important in the ubiquitous computing world. This is because in
ubicomp, users are allowed to use different
computing devices, both physical and virtual. Moreover, the users can use computing
services anywhere and anytime which means
that they use the services in different social
environments. The social environments
and tools are important elements that have
an influence on users' intentions in doing
activities.
Activity theory models the relationships
amongst the elements. Therefore, it can
be a useful way to model the relationship
between each element in a context model.

Activity theory has also been used in the

arena of peripheral displays.
In (Matthews et al., 2007), activity theory was
used to perform an analysis of peripheral displays.
In the same effort, the authors also used activity
theory to develop an approach for designing and
evaluating peripheral displays.

Situated Action
The situated action model emphasizes the emergent, contingent nature of human activity, that
is, the way activity grows directly out of the
particularities of a given situation. The focus is
situated activity or practice. The situated action
model does not underestimate the importance
of artifacts or social relations or knowledge or
values, but rather its true locus of inquiry is the
everyday activity of persons acting in [a] setting (Lave, 1988). The world of computing has
always faced contextual issues. However, the
current wide adoption and usage of ubiquitous
computing (e.g., cellphones, personal digital assistants, etc.) have made contextual issues arguably more prominent than during any other time
in history of computing. The main reason is that
the ubiquitous devices and applications primarily
are used in real settings and therefore, there is a
need for the ubiquitous devices and applications
to support situated activities. The basic unit of
analysis in situated action models is the activity
of persons-acting in setting. (Lave, 1988). The
unit of analysis is thus neither the individual, nor
the environment, but rather a relation between the
two. The situated action model stresses responsiveness to the environment and the improvisatory
nature of human activity. Users under the influence of the environment, may use or attempt to
use ubiquitous technologies/applications in new
ways that even the designers had not anticipated.
The situated action model, therefore, can be suitable for capturing and accommodating such user
improvisations. On the same note, the situated

Designing for Tasks in Ubiquitous Computing

action model deemphasizes the study of more

durable, stable phenomena that persist across
situations (Nardi, 1996). A central tenet of the
situated action approach is that the structuring of
activity is not something that precedes it, but can
only grow directly out of the immediacy of the
situation (Nardi, 1996; Lave, 1988). The authors
of the effort (Fithian et al., 2003) report that they
mainly used the situated action model during the
design and evaluation of an integrated locationaware event and meeting planner built to work in
a PDA form factor. Their justification for adopting
the situated action model was that they wished
to examine the behavior and performance of users in real-world situations, where environmental
and social factors are a source of both distraction
and motivation (Fithian et al., 2003; Taylor &
Harper, 2002). Fithian et al. actually attribute their
meaningful evaluation results to their choice of
the situated action model.

Distributed Cognition
Flor et al. in (Flor et al., 1991) describe distributed
cognition as a new branch of cognitive science
devoted to the study of: the representation of
knowledge both inside the heads of individuals
and in the world ...; the propagation of knowledge
between different individuals and artifacts ...; and
the transformations which external structures
undergo when operated on by individuals and
artifacts.... By studying cognitive phenomena in
this fashion it is hoped that an understanding of
how intelligence is manifested at the systems level,
as opposed to the individual cognitive level, will
be obtained. It should be observed that ubiquitous
devices and applications are primarily used within
real settings/context (the world). Therefore, it is
important that knowledge pertaining to the real
settings be modeled. As has been the case with
the desktop computing applications, knowledge
about the target user too is important in the arena
of ubiquitous computing. On the same note, it is
worth noting that the users of ubiquitous technolo-

gies tend to operate in real settings and, therefore, often have to simultaneously interact with
other people/individuals and artifacts. Knowledge
pertaining to such artifacts and such other individuals is, therefore, important to the design and
development of the ubiquitous applications and
devices being used. In distributed cognition, the
unit of analysis is a cognitive system composed
of individuals and the artifacts they use (Flor et
al., 1991). Distributed cognition moves the unit
of analysis to the system and finds its center of
gravity in the functioning of the system (Nardi,
1996). In a manner similar to traditional cognitive
science (Newell et al., 1972), distributed cognition is concerned with structure (representations
inside and outside the head) and the transformations these structures undergo. However, the
difference is that cooperating people and artifacts are the focus of interest, not just individual
cognition in the head (Nardi, 1996). Another
aspect that distributed cognition emphasizes is
the understanding of the coordination among
individuals and artifacts. The work reported in
(Spinelli et al., 2002) is an investigation of users
involved in carrying out collaborative activities,
locally distributed and mobile. The investigation
utilizes the distributed cognition framework and
contextual design for representing and analyzing
the work observed. By using distributed cognition
to model cognition across users and artifacts, the
study could look at collaboration from an innovative point of view that highlights how context and
external resources impact collaboration. In (Laru
& Jrvel, 2003), the authors address an effort that
has used distributed cognition and collaborative
learning in order to develop a pedagogical model
of mobile learning. UbiLearn is a ubiquitous and
mobile learning project (Laroussi, 2004). Its work
is based on two mobile learning viewpoints; the
first is the technical oriented perspective which
focuses on a traditional behaviouristic educational paradigm as given and tries to represent
or to support it with mobile technologies. The
second is the pedagogical socio-cognitive and

935

Designing for Tasks in Ubiquitous Computing

distributed cognition paradigms, where we face

traditional designs of teaching and learning to
push community oriented learning (e.g., collaborative learning, problem based learning; informal
and ad-hoc learning, etc.). The work (Fischer
et al., 2004) explores the concept of distributed
cognition in ubiquitous computing from two directions. On the one hand, it explores the unique
possibilities that computational media can have
on distributed cognition (how ubicomp technologies can be used to support the users distributed
cognition). On the other hand, it describes a set
of interrelated socio-technical developments that
support distributed cognition among communities
in ubicomp environments, such as a mobile architecture that links mobile travelers with caregiver
communities and transportation systems. The
architecture embodies a distributed cognition
framework that avoids common cognitive barriers found in current transportation systems (i.e.,
generic maps, schedules, labels, landmarks and
signs) while synthesizing personalized multimodal attention and memory prompts from the
transportation environment to provide travelers
with the right information, at the right time, and
in a form best suited for the individual traveler
(Fischer et al., 2004).

Situated Interaction
It may be resourceful to highlight an interaction
paradigm, namely situated interaction that has
been defined based on and motivated by some
of the above models. Situated interaction refers
to the integration of human-computer interaction
and the users situation in a particular working
context in a mobile environment (Hewagamage
& Hirakawa, 2000). This combination perceives
that the interaction is not only a function of device,
but also strongly dependent on the users activities and context in which the device is used. The
concept of situated interaction can be discerned
in, and may be said to have been inspired by, both
the situation action model and the activity theory

936

model. Situated interaction actually introduces a

new paradigm of computing by extending the conventional applications and also by creating a new
set of applications. It is worth noting that mobile
computing has become popular in enhancing the
shopping experience as discussed in (Newcomb et
al., 2003) where they utilized ideas from situated
computing. They go on to say that understanding
situated interactions, where the customer utilizes
the user interface while shopping, became the
greatest challenge for designing the ubiquitous
user interface.
It is worth noting that the acknowledgement
of such challenges could also be supported by the
adoption of design approaches and methodologies
inspired by participatory design, which is based
on the observation of users activities in authentic
everyday settings where mobile computing takes
place (Rogers et al., 2002; Strmberg et al., 2004);
as an example, micro-learning is an emergent area
of investigation that could find useful resources
in these methods while addressing its objective
of designing and distributing series of very small
units of knowledge to be experienced by learners
(for lifelong learning purposes) as intertwined in
their everyday working practices and ubiquitous
computing activities (Gabrielli et al., 2005).

Evaluation Considerations
Conventional user-centered methods could be appropriately exploited in the development process
of ubiquitous applications. On the same note, some
of the traditional usability evaluation techniques
might become useful when adapted for ubiquitous
computing. For instance, there are several efforts
toward realizing usability principles and heuristics for the design and evaluation of ubiquitous
environments/systems, such as ambient heuristics
(Mankoff et al., 2003) and groupware heuristics
(Baker et al., 2001). On the same note, we actually already have proposed a review of usability
principles for mobile computing (Bertini et al.,

Designing for Tasks in Ubiquitous Computing

2005). We have also developed usability heuristics that are appropriate for evaluation in mobile
computing (Bertini et al., 2006).
Much traditional understanding of work organizations has its roots in Fordist and Taylorist
models of human activity, which assume that
human behavior can be reduced into structured
tasks. HCI has not been spared from this either. In
particular, evaluation methods in HCI have often
relied on measures of task performance and task
efficiency as a means of evaluating the underlying
application. However, it is not clear whether such
measures can be universally applicable when we
consider the current move from rather structured
tasks (such as desktop activities) and relatively
stable settings to the often unpredictable ubiquitous settings. Such primarily task-centric evaluation may, therefore, not be directly applicable
to the ubiquitous computing domain. It would be
interesting to consider investigating methods that
go beyond the traditional task-centric approaches
(Abowd & Mynatt, 2000). It is also worth keeping in mind that tasks on the ubiquitous device
(and elsewhere) tend to be unpredictable and
opportunistic.
In this era of ubiquitous computing, the real
need to take into account the real-world context
has become more crucial than at any other time
in the history of computing. Although the concept of context is not new to the field of usability
(e.g., ISO 9241 guidelines propose a model
consideration of context), evaluation methods
have, however, found it challenging, in practice to
adequately/completely integrate the entire context
during the evaluation process. There are various
ways to address this challenge.
One option is the employment of observational
techniques (originally developed by different
disciplines) to gain a richer understanding of
context (Abowd et al., 2002; Dix et al., 2004).
Main candidates are ethnography, cultural probes,
and contextual design. Another option is to use
the Wizard-of-Oz technique, other simulation
techniques, or even techniques that support the

participants imagination. Prototyping too presents an avenue for evaluating ubiquitous computing applications.

Ethnography
Ethnography is an observational technique that
uses a naturalistic perspective; that is, it seeks to
understand settings as they naturally occur, rather
than in artificial or experimental conditions, from
the point of view of the people who inhabit those
settings, and usually involves quite lengthy periods of time at the study site (Hughes et al., 1995).
Ethnography involves immersing an individual
researcher or research team in the everyday activities of an organization or society, usually for
a prolonged period of time. Ethnography is a well
established technique in sociology and anthropology. The principle virtue of ethnography is its
ability to make visible the real world aspects
of a social setting. It is a naturalistic method
relying upon material drawn from the first-hand
experience of a fieldworker in some setting. Since
ubiquitous devices and applications are mainly
used in real world settings, then ethnography
has some relevance to ubiquitous computing. The
aim of ethnography is to see activities as social
actions embedded within a socially organized
domain and accomplished in and through the
day-to-day activities of participants (Hughes
et al., 1995). Data collected/gathered from an
ethnographic study allows developers to design
systems that take into account the sociality of
interactions that occur in the real world. The
work by Crabtree et al. (Crabtree et al., 2006),
shows how ethnography is relevant to and can
be applied in the design of ubiquitous computing
applications. The ultimate aim of the effort is to
foster a program of research and development
that incorporates ethnography into ubiquitous
computing by design, exploiting the inherent
features of ubiquitous computing applications to
complement existing techniques of observation,
data production, and analysis. While describing

937

Designing for Tasks in Ubiquitous Computing

how mobile computing has been used in the fashion

retail industry, Supawanich et al. highlight challenges such as those pertaining to usability, system
tailoring, and the manager-client user experience
(Supawanich et al., 2005). It is worth noting that
they applied ethnography toward addressing the
foregoing challenges. In the work (Newcomb et
al., 2003), which we have mentioned before, the
authors also have applied ethnography in their
effort to examine how grocery shopping could
be aided by a mobile shopping application for
the consumers. In particular, the authors shopped
with customers and followed them throughout the
task of shopping, observing their shopping habits.
In (Berry & Hamilton, 2006), the authors report
that they used ethnography in order to understand
multimedia students and how they use Tablet PCs
in their everyday design studies.

Cultural Probes
Cultural probes (Gaver et al., 1999a) represent a
design-led approach to understanding users that
stresses empathy and engagement. They were
initially deployed in the Presence Project (Gaver
et al., 1999b), which was dedicated to exploring
the design space for the elderly. Gaver has subsequently argued that in moving out into everyday
life more generally, design needs to move away
from such concepts as production and efficiency
and instead focus and develop support for ludic
pursuits. This concept is intended to draw attention to the playful character of human life,
which might best be understood in a post-modern
sense. Accordingly, the notion of playfulness is
not restricted to whatever passes as entertainment,
but is far more subtle and comprehensive, directing
attention to the highly personal and diverse ways
in which people explore, wonder, love, worship,
and waste time together and in other ways engage
in activities that are meaningful and valuable
to them (Gaver, 2001). This emphasis on the ludic
derives from the conceptual arts, particularly the
influence of Situationist and Surrealist schools of

938

thought (Gaver et al., 1999a). Cultural probes draw

on the conceptual arts to provoke or call forth
the ludic and so illuminate the local culture in
which people are located and play out their lives.
During their course of use, ubiquitous devices and
applications typically get embedded in the users
lives and cultures. For instance, people often get
personally attached to their cellphones. Cultural
probes offer fragmentary glimpses into the rich
texture of peoples lives (Gaver, 2002). Cultural
probes are not analytic devices but reflect the
local culture of participants and are drawn upon
to inspire design. In the Presence Project, cultural
probes inspire design by providing a rich and varied set of materials that help to ground designs in
the detailed textures of the local cultures (Gaver
et al., 1999a). These materials are products of the
probe packs, each consisting of a variety of artifacts relevant to the study. Such artifacts provide
a range of materials reflecting important aspects
of the participants local cultures and, on being
returned to the investigators, these reflections inspire design. For instance, in the Presence Project,
the artifacts include: postcards with questions
concerning participants attitudes to their lives,
cultural environment, and technology; maps asking participants to highlight important areas in
their cultural environment; cameras with instructions asking participants to photograph things of
interest to them and things that bored them; photo
albums asking participants to assemble a small
montage telling a story about participants lives;
and media diaries asking participants to record the
various media they use, when, where, and in whose
company. The original idea of culture probes has
been extended to include technology and thus the
concept, technology probes (Hutchinson et al.,
2003; Paulos &Goodman, 2004; Paulos & Jenkins,
2005). According to (Hutchinson et al., 2003),
technology probes can assist in achieving three
interdisciplinary goals: the social science goal of
understanding the needs and desires of users in
a real-world setting; the engineering goal of field
testing the technology; and the design goal of in-

Designing for Tasks in Ubiquitous Computing

spiring users and researchers to think about new

technologies. It is also possible to consider a probe
that is entirely simulated, such as with paratypes
(Abowd et al., 2005). In a research effort aimed
at exploring issues of dependability in ubiquitous
computing in domestic settings (Crabtree et al.,
2002), cultural probes are one of the qualitative
methods that was used. In this case, some participants agreed to keep personal diaries of their
daily activities. However, all participants were
supplied with polaroid cameras, voice activated
dictaphones, disposable cameras, photo albums,
visitors books, scrapbooks, post-it notes, pens,
pencils and crayons, postcards, and maps. In an
attempt to elicit the methods and guidelines for
designing and developing applications for domestic ubiquitous computing, Schmidt and Terrenghi
in (Schmidt & Terrenghi, 2007) adopted various
methods including cultural probes. In a study of
the possible applications of mobile technology for
industrial designers and architects for their daily
work, Muoz Bravo et al. in (Muoz Bravo et al.,
2007) conducted user studies in which one of the
studies consisted of using cultural probes.

Contextual Inquiry
Contextual inquiry (Holtzblatt et al., 1993) is
a method that aims at grounding design in the
context of the work being performed. Contextual
inquiry recommends the observation of work as
it occurs in its authentic setting, and the usage
of a graphical modeling language to describe the
work process and to discover places where technology could overcome an observed difficulty. It
is worth noting that in its application, contextual
inquiry does combine various methods such as
field research and participatory design methods
(Muller et al., 1993) in order to provide designers with grounded and rich/detailed knowledge
of user work. Contextual inquiry is one of the
parts of what is referred to as contextual design.
Contextual design is a design approach that was
developed by Holtzblatt and Beyer (Beyer et al.,

1998). It is an approach for designing customercentered products based on an understanding of

the existing work contexts and practices. It is worth
noting that ubiquitous devices and applications are
often intended to be used and get used in the real
world where real work (or primary tasks) take(s)
place. Therefore, the design of such devices and
applications should be informed by an understanding of the way customers work (or would like to
work) in the real world. Contextual design starts
with the premise that any product embodies a way
of working. The products function and structure
introduce particular strategies, language, and
work flow on its users. A successful design should
therefore offer a way of working that customers would like to adopt. Contextual design has
seven parts: contextual inquiry; work modeling;
consolidation; work redesign; user environment
design; testing with customers; and putting it
into practice. One of the proponents of contextual
design, Holtzblatt, has actually reported on how
contextual design can be appropriated to produce
a mobile application (Holtzblatt, 2005). It is interesting to observe that the work by Newcomb
et al. (Newcomb et al., 2003), did come up with
a contextual design which was meant to serve
two purposes; these are in the shoppers home
to aid him/her in creating a shopping list, and in
the store for the actual shopping. In the effort by
Schmidt and Terrenghi (Schmidt & Terrenghi,
2007), which we came across before, contextual inquiry too was used for understanding and
proposing methods and guidelines for designing
and developing domestic ubiquitous computing
applications. The previously mentioned work by
Spinelli et al. (2002) on locally distributed and
mobile collaborative activities, which we came
across before, did use contextual design. The
authors defend their choice of contextual design
by stating that the representation of work activities, utilising the methods of contextual design,
aid researchers in conceptualising technologies
that truly meet the informational and communicative needs of dynamic and fragmented users

939

Designing for Tasks in Ubiquitous Computing

and their communities. ... This has allowed us

to develop an understanding of, and to design
for, users and their communities-in-context, by
applying techniques such as affinity diagramming (for theme building) and work models to
capture such essential elements as cultural and
social models of technology use; breakdowns
... in working practices and artefact models ...
that allows us to represent users resources and
their relationship with these resources. In the
process, it also promotes an effective coupling of
well-designed technologies with the fast changing physical environments that their users may
inhabit (Spinelli et al., 2002).

Wizard-of-Oz Simulation and

Supporting Immersion
Another possibility is to use the Wizard-of-Oz
technique or even other simulation techniques
such as virtual reality. The Wizard-of-Oz
technique is an evaluation method where the user
of the system is made to believe or perceive that
he or she is interacting with a fully implemented
system though the whole or a part of the interaction of the system is controlled by a human being,
the wizard , or several of them. Such techniques
are especially appropriate where the ubiquitous
application is not fully complete. However, the
simulation should closely reflect the real context
as much as possible (realistic simulation). There
exist various ubiquitous computing applications
that have at some point been evaluated using the
Wizard-of-Oz technique, for example, (Carter
et al., 2007; Mkel et al., 2001; Rudstrm et
al., 2003), and so on. Another alternative is to
adapt more traditional inspection methods to
the analysis of ubicom settings by enriching the
range and quality of discovery resources provided
to usability experts to support their imagination
and immersion about the real world usage settings. We have recently conducted a study in this
direction where video data about user interaction
with an e-learning course delivered on PDAs were

940

used as additional resources supporting a more

effective performance of cognitive walkthrough
evaluation by usability experts involved in the
study (Gabrielli et al., 2005).

Prototypes
In the formative stages of the design process, low
fidelity prototypes can be used. However, as the
design progresses, user tests need to be introduced.
In the context of ubiquitous computing, user tests
will not only require the inclusion of real users,
real settings, and device interaction tasks, but also
real or primary tasks (or realistic simulations of the
real tasks and of the real settings). As mentioned
previously, realistic simulations of the real tasks
and of the real settings could be adopted as an
alternative. Therefore, there would be the need
to provide a prototype that supports the real tasks
and real settings or their simulations. This does
imply some cost in the design process because the
prototype at this level would need to be robust
and reliable enough in order to support primary
tasks in real settings or the simulations. In fact,
the technology required to develop ubiquitous
computing systems is often on the cutting edge.
Finding people with corresponding skills is difficult. As a result, developing a reliable and robust
ubiquitous computing prototype or application
is not easy (Abowd & Mynatt, 2000; Abowd et
al., 2002).

Open Issues and Conclusion

We have attempted to describe the nature of tasks
in ubiquitous computing. We have then proposed
and discussed various models and methods appropriate for supporting the development process
of ubiquitous computing applications based on
the deeper understanding of the nature of tasks.
However, still there are many other pertinent
aspects which too would need to be addressed
and which we consider worthy of our further

Designing for Tasks in Ubiquitous Computing

investigation. These include: the choice of the

methods; the choice of the models; the classification/categorization and characterization of tasks
for mobile and ubiquitous computing; formal
specification of social and collaborative aspects;
and so forth.

Choice of Methods
We have described several methods appropriate
for evaluating in ubiquitous computing. One of
the major issues is deciding which of the methods
to choose. Of such evaluation methods, one may
want to know which one(s) will be most suitable
for a certain ubicomp application. Considering
evaluation methods in general (not just evaluation
methods for ubicomp), Dix et al. indicate that:
there are no hard and fast rules in this each
method has its particular strengths and weakness
and each is useful if applied appropriately. (Dix
et al., 2004). They, however, point out that there
are various factors worth taking into consideration
when choosing evaluation method(s), namely:

the stage in the lifecycle at which the evaluation is carried out;

the style of evaluation (field or laboratory);
the level of subjectivity or objectivity of the
method;
the type of measures provided by the
method;
the level of information provided by the
method;
the immediacy of the response provided by
the method;
the level of interference or intrusiveness of
the method;
the resources required by the method.

The foregoing factors may be appropriately

borrowed from when we consider the evaluation

of ubicomp applications. According to Carter et

al. (Carter et al., 2007), in determining which
methods for ubiquitous computing to use, (among
paper prototypes, interactive prototypes, Wizard-of-Oz, and probes,) the designer must make
trade-offs between realism , unobtrusiveness, data
sparsity, ambiguity, and cost/time. They go on to
say that paper prototypes and Wizard-of-Oz can
be used to explore ambiguity. Probes that can be
employed in real-world situations over a period
of time can support both realism and sparsity.
Moreover, paper and interactive prototypes may
be the least costly methods, but they may also
be the least flexible methods. It therefore comes
as no surprise that some researchers have begun
carrying out corresponding comparative studies
(Liu & Khooshabeh, 2003; Mankoff & Schilit,
1997).
Many of the methods considered in this chapter
are very open compared to more traditional task
analysis techniques. This reflects the often spontaneously planned and re-planned nature of many
tasks in the wild compared to (relatively) more
constrained office tasks. Methods that embody a
fixed or pre-understood idea of human behaviour
are likely to miss some of the nuanced activity that
is the focus of more open observational techniques
such as ethnography. However, without models it
is hard to move from what is observed to potential,
especially as this potential often involves users
appropriating technology for themselves. For this
prompting to see what could happen, as well as
what does happen, more interventionist methods
in particular forms of technology probes or at
least rich prototypes seem more appropriate. That
is, the more open methods seem best suited for
early and late stages in design for understanding
the initial situation and later for assessing the
impact of a deployment. However in mid-stages,
when establishing potential is more important,
more structured models and more interventionist
methods seem more appropriate.

941

Designing for Tasks in Ubiquitous Computing

Choice of Models
Fithian et al. in (Fithian et al., 2003) observe that
mobile and ubiquitous computing applications
lend themselves well to the models: situated action; activity theory; and distributed cognition. As
for which of these models are most suitable for a
certain mobile or ubiquitous application, the foregoing authors say that the choice depends largely
on the kind of application and of which aspects
of design are in the limelight. They recommend
that the choice be based on a critical analysis of
the users and their knowledge, the tasks, and the
application domain.
In (Fithian et al., 2003), Fithian et al. also note
that basing entire evaluation on just time measurements can be very limiting, especially if the
tasks are benchmarked in a situated action setting.
Although time measurements are important, other
performance measures that may be much more
useful for evaluating such ubicomp applications
include interruption resiliency, interaction suspensions, interaction resumptions, and so forth.
Interestingly, these richer metrics require a
far richer model of what is going on than simpler
end-to-end timing. This reinforces the message
on other areas of evaluation that understanding
mechanism is critical for appropriate and reliable
generalization (Ellis & Dix, 2006).

Classification of Tasks
In a study found in (Carter et al., 2007), Carter
et al. report that respondents felt that the current
mobile tools are poorly matched to the user tasks
of meeting and keeping up with friends and
acquaintances. The study observed that location-based technology might assist users in such
tasks. Moreover, the study found that users would
prefer to have cumbersome and repetitive tasks
carried out by their mobile technology artifacts
(e.g., the device, the application, etc.). Carter et al.
also found that planning tasks vary in nature and
detail depending on the formal or informal nature

942

of the event. It might be interesting to consider

how level of formality could be used as one of the
means of classifying tasks in mobile and ubiquitous computing. Carter et al. observe that events
with differing levels of formality require different
tasks and, therefore, different support. They note
that users showed most interest in the systems
that supported informal gathering, rather than
formal gatherings. Another possible criterion for
classifying or categorizing user tasks could be by
borrowing from the activity theorys framework
for describing human behavior (e.g., activities,
operations, actions, etc.) or more specialized
frameworks such as (Bardram, 2005; Bardram
& Christensen, 2004). The work (Matthews et
al., 2007), proposes the following classification
for the types of activities peripheral displays are
likely to support: dormant, primary, secondary,
and pending.

Characterization of Tasks
In a work which primarily describes the challenges
for representing and supporting users activity in
the desktop and ubiquitous interactions, Voida et
al. in (Voida et al., to appear) characterize activities as follows:

activities are dynamic, emphasizing the

continuation and evolution of work artifacts
in contrast to closure and archiving;
activities are collaborative, in the creation,
communication, and dissemination of work
artifacts;
activities exist at different levels of granularity, due to varying durations, complexity
and ownership; and
activities exist across places, including
physical boundaries, virtual boundaries of
information security and access, and fixed
and mobile settings.

In (Abowd & Mynatt, 2000), Abowd and

Mynatt describe everyday computing as an area

Designing for Tasks in Ubiquitous Computing

of interaction research which results from considering the consequences of scaling ubiquitous
computing with respect to time. They indicate
that designing for everyday computing requires
focus on the following features of informal, daily
activities:

they rarely have a clear beginning or end;

interruption is expected;
multiple activities operate concurrently;
time is an important discriminator;
associative models of information are
needed.

Like Fithian et al.s metrics described above (Fithian et al., 2003), these properties all emphasize
the fact that activities in a ubiquitous interaction
are more fragmented and require more divided
attention than architypal office applications,
although arguably these were never as simple as
the more simplisitic models suggested. However,
the first point also suggests that at a high-level
there may be more continuity, and this certainly
echoes Carter et al.s study (Carter et al., 2007)
with the importance of informal gathering and
communication a life-long goal.

Formal Specification of Social and

Collaborative Aspects
With a formal specification, it is possible to
analyze a system long before it gets designed
or implemented. Although this benefit applies to
virtually all types of systems, it is interesting to
the world of ubiquitous computing where, as we
have noted, at the end of the previous section,
developing a reliable and robust prototype or
application is not an easy undertaking. Formal
specifications, therefore, can be useful in supporting the development of ubiquitous applications. On the same note, in ubiquitous computing
users perform their activities in the real world
settings, where there are other people. In other
words, ubiquitous computing involves context,

which includes other people besides the user.

Therefore, collaborative and social aspects have
a lot of weight in ubiquitous computing. It has
been rightly noted in (Abowd & Mynatt, 2000)
that human beings tailor their activities and recall
events from the past based on the presence (or
even the help) of other people. Therefore, it is
important to consider how we can realize formal
specifications that can represent collaborative
and social aspects for ubiquitous applications. It
is worth observing that much of the research in
ubiquitous computing has focused on mobility
(and other contextual aspects) with regard to an
individual user, with little being done regarding
social and collaborative aspects.
One of the possible approaches to the formal
modeling of social aspects is through the use of
agents. It might be worth investigating to what
degree such agent-based models can be applied in
ubiquitous computing. One such model is OperA
(Dignum et al., 2002a; Dignum et al., 2002b; Dignum, 2004). The authors indicate that the concept
of agents is useful for representing organizational
interaction for two main reasons. The first is that
it enables the reference to any autonomous entity
participating in an interaction, including people.
The second is that it provides theoretical models
for entities and interaction. OperA abstracts
from the specific internal representations of the
individual agents, and separates the modeling of
organizational requirements and aims. Contracts
are used to link the different models and create
specific instances that reflect the needs and structure of the current environment and participants
(Dignum, 2004).
It might also be appropriate to borrow a leaf
from Grid Computing where the need for models
for addressing collaborative and social aspects
has been identified (Liu & Harrison, 2002). According to Liu (Liu, 2003), in Grid Computing
the development of such models has been based
on the early work on information systems (Liu,
2000; Stamper, 1973; Stamper, 1996) and computer-supported collaborative work (CSCW),

943

Designing for Tasks in Ubiquitous Computing

(Liu et al., 2001). One particular model that has

been proposed is the SPS model, which entails
the integrated modeling of semantic, pragmatic
and social aspects (Liu, 2003). Regarding formal
specifications for CSCW in general (and not just
under Grid Computing), one interesting effort
is the work by Johnson (Johnson, 1999), which
describes how formal methods can be used creatively to solve a vast range of design problems
within CSCW interfaces. It is worth noting that the
work does show how mathematical specification
techniques can be enhanced to capture physical
properties of working environments, thereby
providing a link between the physiological studies from ergonomics and the HCI user interface
design techniques.
A related and interesting work is found in
(Musolesi et al., 2004), in which there is a proposal
of a two-level mobility model that is based on
artificially generated social relationships among
individuals carrying mobile devices. The generation process respects the mathematical basis of
social networks theory and, thus, is grounded in
empirical experience of actual social relationships. The second level/stage maps the social
organization onto topographical space such that
the actual generated topography is biased by the
strength of social ties.
At a very low level, more traditional formal
models become applicable as we are below
the level of the more complex considerations of
ubiquitous computing. In particular, variations of
Fitts law have been used extensively to understand
and to design interfaces for pointing tasks on tiny
devices (Guiard & Beaudouin-Lafon, 2004).

Summary
As a way of emphasizing the relevance of the
theme of this chapter, it is worth observing that
there is a growing interest within the research
community regarding tasks in ubiquitous computing. Therefore, it comes as no surprise that

944

we are now seeing the emergence of fields such

as activity-centered design (Gay & Hembrooke,
2003), activity-based computing (Bardram, 2005;
Bardram & Christensen, 2004), and activity-based
ubiquitous computing (Li & Landay, 2006).
As we consider the move from the conventional desktop setting to the real world setting,
various design issues and demands arise when
we consider the nature of tasks the ubiquitous
devices/applications would be expected to support and the real world context in which they will
be used. A close study of the nature of tasks in
ubiquitous computing has the potential to bring
to light some of the requirements in the development of ubiquitous applications.
In particular, we have seen how tasks in ubiquitous environments tend to be more dynamic,
less pre-planned, and more situated than those
commonly assumed to be the case for more traditional desktop applications. In addition, users
are likely to be involved in multiple activities, and
the task involving a ubiquitous device may not be
the primary task for the user either because there
is a real world task(s) such as driving that takes
precedence, or because the device interaction is
merely supporting an ongoing activity such as
social coordination. Interruptions and resumptions of activity become the norm (although
there is plenty of evidence that this is also the
case in the office) and so the need, as advocated
in distributed cognition, to offload memory into
the device becomes important.
Because of the dynamic nature of tasks we
have discussed, various methods and theories
that emphasise the richer nature of human activity, and any methods used to study tasks for
ubiquitous interaction have to be open to seeing
unexpected patterns of activity. However, there
are clearly also generic meta-tasks and common
issues found in many ubiquitous interactions
including offloading of memory, interruption
management, location sensitivity, and so forth.
It is essential to understand the former, situation
specific issues, in order to avoid designs that

Designing for Tasks in Ubiquitous Computing

are not fit for that purpose; however, the latter,

generic issues, offer the potential for lessons to
be learnt across systems and for ongoing fruitful
research directions.

References
Abowd, G. D., & Mynatt, E. D. (2000). Charting
Past, Present, and Future Research in Ubiquitous
Computing. ACM Transactions on Computer-Human Interaction, 7(1), 29-58.
Abowd, G. D., Mynatt, E. D., & Rodden, T.
(2002). The Human Experience. IEEE Pervasive
Computing, 1(1), 48-57.
Abowd, G. D., Hayes, G. R., Iachello, G., Kientz,
J. A., Patel, S. N., & Stevens, M. M. (2005).
Prototypes and paratypes: Designing mobile
and ubiquitous computing applications. IEEE
Pervasive Computing, 4(4), 6773.
Baker, K., Greenberg, S., & Gutwin, C. (2001).
Heuristic Evaluation of Groupware Based on
the Mechanics of Collaboration. In Proceedings
of the 8th IFIP International Conference on
Engineering for Human-Computer Interaction
(pp. 123-140).
Bardram, J. E. (2005). Activity-Based Computing: Support for Mobility and Collaboration in
Ubiquitous Computing. Personal and Ubiquitous
Computing, 9(5), 312-322.
Bardram, J. E., & Christensen, H. B. (2004).
Open Issues in Activity-Based and Task-Level
Computing. Paper presented at the Pervasive04
Workshop on Computer Support for Human Tasks
and Activities (pp. 55-61), Vienna, Austria.
Berry, M., & Hamilton, M. (2006). Mobile Computing, Visual Diaries, Learning and Communication: Changes to the Communicative Ecology of
Design Students Through Mobile Computing. In
Proceedings of the eighth Australasian Computing
Education Conference (ACE2006)Conferences

in Research in Practice in Information Technology (pp. 35-44). Hobart, Tasmania, Australia.

Australian Computer Society, Inc.
Bertini, E., & Kimani, S. (2003). Mobile Devices:
Opportunities for Users with Special Needs. In
Proceedings of Mobile HCI Conference 2003
(pp. 486-491).
Bertini, E., Catarci, T., Kimani, S., & Dix, A.
(2005). A Review of Standard Usability Principles
in the Context of Mobile Computing. International
Journal of Studies in Communication Sciences,
5(1), 111-126.
Bertini, E., Gabrielli, S., Kimani, S., Catarci, T., &
Santucci, G. (2006). Appropriating and Assessing
Heuristics for Mobile Computing. Paper presented
at the International Conference in Advanced
Visual Interfaces (AVI) (pp. 119-126).
Beyer, H., & Holtzblatt, K. (1998). Contextual
Design: Defining Customer-Centered Systems.
Academic Press: Kaufmann Publishers.
Boer, N., van Baalen, P.J., & Kumar, K. (2002).
An activity theory approach for studying the
situatedness of knowledge sharing. In Proceedings of the 35th Annual Hawaii International
Conference on System Sciences (HICSS-3502)
(pp. 1483-1492).
Card, S. K., Moran, T. P., & Newell, A. (1983).
The Psychology of Human-Computer Interaction.
Lawrence Erlbaum Associates.
Carter, S., Mankoff, J., Klemmer, S. R., & Matthews, T. (2007). Exiting the cleanroom: on ecological validity and ubiquitous computing. Journal
of Human-Computer Interaction, 22(1/2).
Crabtree, A., Hemmings, T., Rodden, T., Clarke,
K., Dewsbury, G., Hughes, J., Rouncefield, M.,
& Sommerville, I. (2002). Sore Legs and Naked
Bottoms: Using Cultural Probes in Dependability Research. DIRC Conference on Dependable
Computing Systems. London: The Royal Statistical Society.

945

Designing for Tasks in Ubiquitous Computing

Crabtree, A., Benford, S., Greenhalgh, C., Tennent, P., Chalmers, M., & Brown, B. (2006).
Supporting Ethnographic Studies of Ubiquitous
Computing in the Wild. In Proceedings of the
sixth ACM conference on Designing Interactive
Systems (pp.60-69). ACM Press.
Dignum, V., Meyer, J-J., Weigand, H., & Dignum,
F. (2002a). An Organizational-oriented Model for
Agent Societies. In Proceedings of the International Workshop on Regulated Agent-Based Social
Systems: Theories and Applications (RASTA02),
at AAMAS, Bologna, Italy.
Dignum, V., Meyer, J-J., Dignum, F., & Weigand,
H. (2002b). Formal Specification of Interaction in
Agent Societies. Paper presented at the Second
Goddard Workshop on Formal Approaches to
Agent-Based Systems (FAABS), Maryland.
Dignum, V. (2004). A model for organizational
interaction: based on agents, founded in logic.
Unpublished doctoral dissertation, Utrecht University.
Dix, A. (2002). Beyond intention - pushing
boundaries with incidental interaction. In Proceedings of Building Bridges: Interdisciplinary
Context-Sensitive Computing, Glasgow University, Sept 2002. http://www.hcibook.com/alan/
topics/incidental/
Dix, A., Finlay, J., Abowd, G., & Beale, R. (2004).
Human-Computer Interaction. Prentice Hall
(Third Edition).
Dix, A., Ramduny-Ellis, D., & Wilkinson, J.
(2004b). Trigger Analysis - understanding broken tasks. Chapter 19 in The Handbook of Task
Analysis for Human-Computer Interaction. D.
Diaper & N. Stanton (Eds.) (pp.381-400). Lawrence Erlbaum Associates.
Ellis, G., & Dix, A. (2006). An explorative analysis
of user evaluation studies in information visualisation. Proceedings of the 2006 Conference
on Beyond Time and Errors: Novel Evaluation

946

Methods For information Visualization (BELIV

06) (pp. 1-7), Venice, Italy. ACM Press, New
York, NY.
Fischer, G., Arias, E., Carmien, S., Eden, H.,
Gorman, A., Konomi, S., & Sullivan, J. (2004).
Supporting Collaboration and Distributed Cognition in Context-Aware Pervasive Computing
Environments. Meeting of the Human Computer
Interaction Consortium Computing Off The
Desktop.
Fithian, R., Iachello, G., Moghazy, J., Pousman,
Z., & Stasko, J. (2003). The design and evaluation
of a mobile location-aware handheld event planner. In Proceedings of Mobile HCI Conference
2003 (pp. 145-160).
Flor, N., & Hutchins, E. (1991). Analyzing distributed cognition in software teams: A case study
of team programming during perfective software
maintenance. In J. Koenemann-Belliveau et al.
(Eds.), In Proceedings of the Fourth Annual
Workshop on Empirical Studies of Programmers
(pp. 3659). Norwood, N.J.: Ablex Publishing.
Gabrielli, S., Mirabella, V., Kimani, S., & Catarci,
T. (2005). Supporting Cognitive Walkthrough
with Video Data: Results from a Mobile Learning
Evaluation Study. In Proceedings of Mobile HCI
Conference 2005 Conference (pp. 77-82).
Gabrielli, S., Kimani, S., & Catarci, T. (2005).
The Design of MicroLearning Experiences: A
Research Agenda, Paper presented at the Microlearning Conference 2005.
Gaver, W.H., Dunne, A., & Pacenti, E. (1999a).
Cultural probes. ACM Interactions, 6(1), 21-29.
Gaver, W.H., Hooker, B., & Dunne, A. (1999b).
The Presence Project. London: Department of
Interaction Design.
Gaver, W. (2001). Designing for ludic aspects
of everyday life, ERCIM News, 47. www.ercim.
org/publication/Ercim_News/enw47/gaver.html.

Designing for Tasks in Ubiquitous Computing

Gaver, W. (2002). Domestic Probes. www.crd.

rca.ac.uk/equator/PROBE.htm.
Gay, G., & Hembrooke, H. (2003). Activity-Centered Design: An Ecological Approach to Designing Smart Tools and Usable Systems. Cambridge,
MA. MIT Press.
Gorlenko, L., & Merrick, R. (2003). No Wires
Attached: Usability Challenges in the Connected Mobile World. IBM Systems Journal, 42,
639-651.
Guiard, Y., & Beaudouin-Lafon, M. (2004) Target
Acquisition in Multiscale Electronic Worlds. International Journal of Human-Computer Studies,
61(6), 875-905.
Hewagamage, K. P., & Hirakawa, M. (2000).
Situated Computing: A Paradigm to Enhance the
Mobile Users Interaction. Handbook of Software
Engineering and Knowledge Engineering, 0(0).
World Scientific Publishing Company.
Holland, S., & Morse, D. R. (2001). Audio GPS:
Spatial Audio in a Minimal Attention Interface.
In Proceedings of Mobile HCI Conference 2001
(pp. 28-33).
Holtzblatt, K. (2005). Customer-Centered Design
for Mobile Applications. Personal and Ubiquitous
Computing, 9(4), 227-237.
Holtzblatt, K., & Jones, S. (1993). Contextual
Inqury: Principles and Practice. Participatory
Design: Principles and Practices. Lawrence
Earlbaum, New York.
Horvitz, E. (1999). Principles of Mixed-Initiative
User Interfaces. In Proceedings of the ACM SIGCHI Conference on Human Factors in Computing
Systems (pp. 159-166).
Hughes, J., Rodden, T., King, V., & Anderson, K.
(1995). The Role of Ethnography in Interactive
Systems Design. ACM Interactions, (4), 56-65.
Hutchinson, H., Mackay, W., Westerlund, B., Bederson, B. B., Druin, A., Plaisant, C., Beaudouin-

Lafon, M., Conversy, S., Evans, H., Hansen, H.,

Roussel, N., & Eiderback, B. (2003). Technology
probes: Inspiring design for and with families.
In Proceedings of the ACM SIGCHI Conference
on Human Factors in Computing Systems (pp.
1724). New York: ACM Press.
Johnson, C. (1999). Expanding the role of formal methods in CSCW. In M. Beaudouin-Lafon
(Ed.), Computer Supported Co-operative Work,
Volume 7 of Trends in Software, 221-255. John
Wiley & Sons.
Kaenampornpan, M., & ONeill, E. (2004). An
Integrated Context Model: Bringing Activity
to Context. In Proceedings of the Workshop on
Advanced Context Modelling, Reasoning and
Management at UbiComp 2004.
Kaptelinin, V., Kuutti, K., & Bannon, L. (1995).
Activity Theory: Basic Concepts and Applications.
In Blumenthal et al. (Eds.), Human-Computer Interaction. (LNCS, pp. 189-201). Springer.
Kristoffersen, S., & Ljungberg, F. (1999). Making
Place to Make IT Work: Empirical Explorations
of HCI for Mobile CSCW. In Proceedings of the
International ACM SIGGROUP Conference on
Supporting Group Work, (pp. 276-285).
Laroussi, M. (2004). New E-Learning Services
Based on Mobile and Ubiquitous Computing:
Ubi-Learn Project. Paper presented at the Computer Aided Learning in Engineering Education
(CALIE04). Grenoble, France.
Laru, J., & Jrvel, S. (2003). Applying wireless
technology for co-ordinating collaboration in
distributed university teachers team. Poster at
Computer Supported Collaborative Learning
(CSCL2003).
Lave, J. (1988). Cognition in Practice. Cambridge:
Cambridge University Press.
Leontev, A. N. (1978). Activity, Consciousness,
Personality. Englewood Cliffs, NJ, Prentice
Hall.

947

Designing for Tasks in Ubiquitous Computing

Li, Y., & Landay, J. A. (2006). Exploring ActivityBased Ubiquitous Computing Interaction Styles,
Models and Tool Support. In Proceedings of the
Workshop What is the Next Generation of Human-Computer Interaction? at the ACM SIGCHI
Conference on Human Factors in Computing
Systems.
Liu, K. (2000). Semiotics in Information Systems
Engineering. Cambridge University Press.
Liu, K., Clarke, R., Stamper, R., & Anderson, P.
(2001). Information, Organisation and Technology: Studies in Organisational Semiotics 1.
Kluwer, Boston.
Liu, K., & Harrison, R. (2002). Embedding
Softer Aspects into the Grid. Poster at EUROWEB 2002 - The Web and the GRID: from
e-science to e-business (pp. 179-182). The British
Computer Society.
Liu, K. (2003). Incorporating Human Aspects into
Grid Computing for Collaborative Work. Paper
presented at the ACM International Workshop on
Grid Computing and e-Science. San Francisco,
CA.
Liu, L., & Khooshabeh, P. (2003). Paper or interactive? A study of prototyping techniques for ubiquitous computing environments. In Proceedings of
the ACM SIGCHI Conference on Human Factors
in Computing Systems, extended abstracts, (pp.
130131). New York: ACM Press.
Mkel, K., Salonen, E-P., Turunen, M., Hakulinen, J., & Raisamo, R. (2001). Conducting
a Wizard of Oz Experiment on a Ubiquitous
Computing System Doorman. In Proceedings
of the International Workshop on Information
Presentation and Natural Multimodal Dialogue
(pp. 115-119). Verona.
Mankoff, J., & Schilit, B. (1997). Supporting
knowledge workers beyond the desktop with
PALPlates. In Proceedings of the ACM SIGCHI
Conference on Human Factors in Computing

948

Systems, extended abstracts, (pp. 550551). New

York: ACM Press.
Mankoff J., Dey A. K., Hsieh G., Kientz J.,
Lederer S. & Ames M. (2003). Heuristic Evaluation of Ambient Displays. Proceedings of the
ACM SIGCHI Conference on Human Factors in
Computing Systems (pp. 169-176).
Matthews, T., Rattenbury, T., & Carter, S.
(2007). Defining, Designing, and Evaluating
Peripheral Displays: An Analysis Using Activity
Theory. Journal of Human-Computer Interaction, 22(1/2).
Muller, M.J., Wildman, D.M., & White, E.A.
(1993). A Taxonomy of PD Practices: A Brief
Practitioners Guide. Communications of the
ACM, 36(4), 26-28.
Muoz Bravo, J., Lucero, A., & Aliakseyeu, D.
(2007). Probing the Need for Mobile Technologies for Designers. In Proceedings of the Second
IASTED International Conference on HumanComputer Interaction. Chamonix, France. ACTA
Press.
Musolesi, M., Hailes, S., & Mascolo, C. (2004).
An Ad Hoc Mobility Model Founded on Social
Network Theory. In Proceedings of the Seventh
ACM/IEEE International Symposium on Modeling, Analysis and Simulation of Wireless and
Mobile Systems (MSWiM 2004) (pp. 20-24).
Venezia, Italy.
Nardi, B. (1996). Studying context: a comparison
of activity theory, situated action models, and
distributed cognition. In: B. Nardi (Ed.), Context
and Consciousness: Activity Theory and HumanComputer Interaction (pp. 69-102). MIT Press,
Cambridge, MA.
Newcomb, E., Pashley, T., & Stasko, J. (2003). Mobile Computing in the Retail Arena. In Proceedings of the ACM SIGCHI Conference on Human
Factors in Computing Systems (pp. 337-344).

Designing for Tasks in Ubiquitous Computing

Newell, A., & Simon, H. (1972). Human Problem

Solving. Englewood Cliffs, NJ: Prentice-Hall.
Pascoe, J., Ryan, N., & Morse, D. (2000). Using
While Moving: HCI Issues in Fieldwork Environments. ACM Transactions on Human-Computer
Interaction, 7(3), 417-437.
Paulos, E., & Goodman, E. (2004). The familiar
stranger: Anxiety, comfort, and play in public
places. Proceedings of the ACM SIGCHI Conference on Human Factors in Computing Systems
(pp. 223230). New York: ACM Press.
Paulos, E., & Jenkins, T. (2005). Urban probes:
Encountering our emerging urban atmospheres.
In Proceedings of the ACM SIGCHI Conference
on Human Factors in Computing Systems (pp.
341350). New York: ACM Press.
Pinto, H., & Jose, R. (2006). Activity-centered
ubiquitous computing support to localized activities. In Proceedings of the Conference on Mobile
and Ubiquitous Systems (CMUS) (pp. 119-128).
Guimares.
Rogers, Y., Scaife, M., Muller, H., Randell, C.,
Moss, A., Taylor, I., Harris, E., Smith, H., Price,
S., Phelps, T., Corke, G., Gabrielli, S., Stanton,
D., & OMalley, C. (2002). Things Arent What
They Seem to be: Innovation Through Technology Inspiration. In Proceedings of ACM-SIGCHI
DIS02 (pp. 373-378).
Rudstrm, ., Cster, R., Hk, K., & Svensson,
M. (2003). Paper prototyping a social mobile service. In Proceedings of the MUM Workshop on
Designing for Ubicomp in the Wild: Methods for
Exploring the Design of Mobile and Ubiquitous
Services.
Schmidt, A., & Terrenghi, L. (2007). Methods
and Guidelines for the Design and Development
of Domestic Ubiquitous Computing Applications.
In Proceedings of PerCom2007 (pp. 97-107). New
York, USA.

Spinelli, G., Brodie, J., & Perry, M. (2002). Using

Distributed Cognition and Contextual Design
to Analyse and Represent Collaborative Activities. In Proceedings of CSCSW2002 Conference
workshop Analyzing Collaborative Activity. New
Orleans, Louisiana, USA.
Stamper, R. K. (1973). Information in Business
and Administrative Systems. Wiley.
Stamper, R. K. (1996). Signs, Information, Norms
and Systems. In P. Holmqvist, P. B. Andersen,
H. Klein, & R. Posner (Eds.), Signs of Work (pp.
349-397). Walter de Gruyter.
Strmberg, H., Pirttil, V., & Ikonen, V. (2004).
Interactive Scenarios-building Ubiquitous Computing Concepts in the Spirit of Participatory
Design. Personal and Ubiquitous Computing,
8, 200-207.
Supawanich, A., Rangos, J., Harriman, J., &
Schmitt, G. (2005). Mobile Computing in HighEnd Retail. In Proceedings of the conference
on Designing for User eXperience. American
Institute of Graphic Arts. New York, USA.
Taylor, A., & Harper, R. (2002). Age-old Practices
in the New World: A study of gift-giving between
teenage mobile phone users. In Proceedings of the
ACM SIGCHI Conference on Human Factors in
Computing Systems (pp. 439446). ACM Press.
Uden, L. (2007). Activity theory for designing
mobile learning. International Journal of Mobile
Learning and Organisation, 1(1), 81-102.
Voida, S., Mynatt, E. D., & MacIntyre, B. (2007).
Supporting Activity in Desktop and Ubiquitous
Computing. In V. Kaptelinin, & M. Czerwinski
(Eds.), Designing integrated digital work environments: Beyond the desktop metaphor (pp.
195-222). Cambridge, MA: MIT Press.
Weiser, M. (1991). The Computer for the 21st
Century. Scientific American, 265(3), 91-104.

This work was previously published in Advances in Ubiquitous Computing: Future Paradigms and Directions, edited by S.
Mostefaoui, Z. Maamar, and G. Giaglis, pp. 171-200, copyright 2008 by IGI Publishing, formerly known as Idea Group Publishing (an imprint of IGI Global).
949

950

Chapter 2.32

Task Ontology-Based
Human-Computer Interaction
Kazuhisa Seta
Osaka Prefecture University, Japan

INTRODUCTION
In ontological engineering research field, the
concept of task ontology is well-known as a
useful technology to systemize and accumulate
the knowledge to perform problem-solving tasks
(e.g., diagnosis, design, scheduling, and so on).
A task ontology refers to a system of a vocabulary/concepts used as building blocks to perform
a problem-solving task in a machine readable
manner, so that the system and humans can collaboratively solve a problem based on it.
The concept of task ontology was proposed
by Mizoguchi (Mizoguchi, Tijerino, & Ikeda,
1992, 1995) and its validity is substantiated by
development of many practical knowledge-based
systems (Hori & Yoshida, 1998; Ikeda, Seta, &
Mizoguchi, 1997; Izumi &Yamaguchi, 2002;
Schreiber et al., 2000; Seta, Ikeda, Kakusho, &
Mizoguchi, 1997). He stated:

task ontology characterizes the computational

architecture of a knowledge-based system which
performs a task. The idea of task ontology which
serves as a system of the vocabulary/concepts
used as building blocks for knowledge-based systems might provide an effective methodology and
vocabulary for both analyzing and synthesizing
knowledge-based systems. It is useful for describing inherent problem-solving structure of the existing tasks domain-independently. It is obtained by
analyzing task structures of real world problem.
... The ultimate goal of task ontology research is
to provide a theory of all the vocabulary/concepts
necessary for building a model of human problem
solving processes. (Mizoguchi, 2003)
We can also recognize task ontology as a static
user model (Seta et al., 1997), which captures
the meaning of problem-solving processes, that
is, the input/output relation of each activity in a

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Task Ontology-Based Human-Computer Interaction

problem-solving task and its effects on the real

world as well as on the humans mind.

BACKGROUND
Necessity of Building Task
Ontologies as a Basis of HCI
It is extremely difficult to develop an automatic
problem-solving system that can cope with a
variety of problems. The main reason is that the
knowledge for solving a problem varies considerably depending on the nature of the problems. This
engenders a fact that is sometimes ignored: Users
have more knowledge than computers. From this
point of view, the importance of a user-centric
system (DeBells, 1995) is now widely recognized
by many researchers. Such framework follows a
collaborative, problem-solving-based approach
between human and computer by establishing
harmonious interaction between human and
computer.
Many researchers implement such a framework
with a human-friendly interface using multimedia
network technologies. Needless to say, it is important not only to apply the design principles of
the human interface but also principle knowledge
for exchanging meaningful information between
humans and computers.
Systems have been developed to employ research results of the cognitive science field in order
to design usable interfaces that are acceptable to
humans. However, regarding the content-oriented
view, it is required that the system can understand
the meaning of humans cognitive activities in
order to capture a humans mind.
We, therefore, need to define a cognitive model,
that is, to define the cognitive activities humans
perform in a problem-solving/decision-making
process and the information they infer, and then
systemize them as task ontologies in a machine
understandable manner in order to develop an
effective human-computer interaction.

Problem-Solving Oriented Learning

A task with complicated decision making is referred to as Problem-Solving Oriented Learning
(PSOL) task (Seta, Tachibana, Umano, & Ikeda,
2003; Seta & Umano, 2002). Specifically, this
refers to a task that does not only require learning
to build up sufficient understanding for planning
and performing problem-solving processes but
also to gain the ability/skill of making efficient
problem-solving decisions based on sophisticated
strategies.
Consider for example, a learner who is not
very familiar with Java and XML programming
and tries to develop an XML-based document
retrieval system. A novice learner in a problemsolving domain tries to gather information from
Web resources, investigates and builds up his/her
own understanding of the target area, and makes
plans to solve the problem at hand and then perform
problem-solving and learning processes. Needless
to say, a complete plan cannot be made at once,
but is detailed gradually by iterating, spirally,
those processes while applying a trial and error
approach. Thus, it is important for a learner to
control his/her own cognitive activities.

Facilitating Learners Meta Cognition

through HCI
In general, most learners in PSOL tend to work
in an ad hoc manner without explicit awareness
of meaning, goals and roles of their activities.
Therefore, it is important to prompt construction
of a rational spiral towards making and performing efficient problem-solving processes by giving
significant direction using HCI.
Many researchers in the cognitive science field
proposed a concept whereby metacognition plays
an important role to acquire and transfer expertise
(Brown, Bransford, Ferrara, & Campione, 1983;
Flavell, 1976; Okamoto, 1999). Furthermore,
repeated interaction loops between metacognition activities and cognition activities play an

951

Task Ontology-Based Human-Computer Interaction

Figure 1. The interaction helps the effective planning process

important role in forming an efficient plan for

problem-solving and learning processes.
Figure 1 shows the plan being gradually detailed and refined along the time axis. Figure 1(a)
is a planning process when a learner has explicit
awareness of interactions and iterate metacognition activities and cognition activities spirally,
while Figure 1(b) is a planning process with implicit awareness of them. In PSOL, monitor and
control of problem-solving/learning processes
are typical activities of metacognition while their
performances are ones of cognition. It is natural
that the former case allows efficient plans for
problem-solving workflow more rapidly than the
latter. Without explicit awareness of interaction
loops, a learner tends to get confused and lose
his/her way because nested structures of his/her
work and new information of the target world
impose heavy loads.
Therefore, it is important to implement an
HCI framework that enables effective PSOL by
positioning a learner at the center of the system

952

as a subject of problem solving or learning, and

providing appropriate information to prompt the
learners metacognition effectively.

MAIN ISSUES IN TASK

ONTOLOGY-BASED HCI
In this section, we introduce our approach to supporting PSOL to understand task ontology based
HCI framework.
Rasmussens (1986) cognitive model is adopted
as a reference model in the construction of the task
ontology for supporting PSOL. It simulates the
process of human cognition in problem-solving
based on cognitive psychology. Cognitive activity
in PSOL is related to this model based on which
PSOL task ontology is constructed. This provides
a learner with useful information for effective
performance of cognitive activity at each state,
according to the theoretical framework that was
revealed in the cognitive psychology.

Task Ontology-Based Human-Computer Interaction

Rasmussens (1986) Cognitive Model

Figure 2 represents an outline of Rasmussens
cognitive model known as the ladder model.
Activities in PSOL broadly comprise activities in connection with a problem-solving act and
activities in connection with a learning act (see
Figure 3 in the next section).
An Activation activity in Rasmussens cognitive model corresponds to the situation in which
a problem is given in problem-solving activities,
or one in which a learner detects change in the
real world. An Observe activity corresponds to
observing the details of the change or a gap from
the problem-solving goal. An Identify activity
corresponds to identifying its possible cause. An
Interpret activity corresponds to interpreting the
influence of the change on problem solving and
deciding the problem-solving goal. A Define Task
activity corresponds to determine a problemsolving task for implementing it based on the
problem-solving goal. A Formulate Procedure
activity corresponds to setting up a problem-solving plan to solve the problem-solving task.

Figure 2. Rasmussens (1986) cognitive model

Although basically the same correspondence

applies in learning activities as the case of problem-solving activities, the object of learning activities, mainly focuses on the state of ones own
knowledge or understanding, that is, metacognition activities. Namely, the Activation activity
in Rasmussens cognitive model corresponds to
detecting the change of ones own knowledge state.
The Observe activity corresponds to observing
details or a gap from its own understanding state
(goal state) decided as a goal of learning. The
Identify activity corresponds to identifying its possible cause. The Interpret activity corresponds to
interpreting the influence of its own understanding
state, especially the influence on problem solving
in PSOL, and deciding the goal of learning. The
Define Task activity corresponds to setting up a
learning task for implementing it based on the
problem-solving goal. The Formulate Procedure
activity corresponds to setting up a learning plan
to solve the problem-solving task.
Clarifying a correspondence relationship
between the cognitive activity by a learner in
PSOL and the cognitive activity in Rasmussens
cognitive model permits construction of a problem-solving-oriented learning task ontology as
a basis of human-computer interaction comprehending the properties of PSOL appropriately.
Implementing an interaction between a system
and a learner based on this allows the system
to show effective information to encourage the
learners appropriate decision-making.

Cognitive Model in Problem-Solving

Oriented Learning
Figure 3 shows a cognitive model that captures
detailed working processes of a learner. This
model is PSOL task specific while Rasmussens
model is a task independent one. By making the
correspondence between these models, we can
define an HCI framework based on Rasmussens
theory.

953

Task Ontology-Based Human-Computer Interaction

Figure 3. Learners work in problem-solving-oriented learning (PSOL)

Figures 3(i) and 3(iii) represent the planning

process of the problem-solving plan and learning
plan, respectively, and 3(viii) and 3(x) represent
problem-solving and learning processes in Figure
1, respectively. Figures 3(v) and 3(vi) represent
the monitoring process.
We have presented a problem, say, developing
an XML based document retrieval system in the
upper left corner. Two virtual persons, a problemsolving planner and learning process planner in
the learner, play roles of planning, monitoring,
and controlling problem-solving, and learning
processes, respectively.
With PSOL, a learner first defines a problemsolving goal and refines it to sub-goals which
contribute to achieving goal G (Figure 3(i)). They
are refined to feasible problem-solving plans (Fig-

954

ure 3(ii)); thereafter, the learner performs them

to solve the problem (Figure 3(viii)).
If the learner recognizes a lack of knowledge
in the sub goals and performs problem-solving
plans, we can generate an adequate learning goal
(LG) to get knowledge (Figure 3(iii)) and refine
it to learning process plans (Figure 3(iv)). In
learning processes (Figure 3(x)), s/he constructs
knowledge (Figure 3(iv)) to be required to plan
and perform the problem-solving process. Based
on constructed knowledge, she or he specifies and
performs the problem-solving processes (Figure
3(viii)), to change the real world (Figure 3(vii)). The
learner assesses gaps among goal states (GS), current goal states (CGS) of problem-solving process
plans, and current state (c-state) of the real-world
(Figure 3(v)) and ones among learning goal states

Task Ontology-Based Human-Computer Interaction

(LGS), current learning goal states (CLGS) of

learning process plans and understanding state
(Figure 3(vi)). She or he continuously iterates
these processes until the c-state of the real world
satisfies the GS of problem solving.
It is notable that learners in PSOL have to
make and perform not only problem-solving
plans, but also learning plans in the process of
problem solving. Furthermore, it is important
for the learner to monitor real-world changes
by performing problem-solving processes and
to monitor his/her own understanding states by
performing learning processes and checking and
analyzing whether states of the real world and
understanding states satisfy defined goal states
(Figures 3(v) and 3(vi)). The gap between current
states and goal states causes the definition of new
goals to be dissolved.
Consequently, PSOL impels a learner to
perform complicated tasks with heavy cognitive
loads. A learner needs to manage and allocate
the attentional capacity adequately because of
limited human attentional capacity. This explains
why a novice learner tends to get confused and
lose his/her way.

Task Ontology for

Problem-Solving-Oriented Learning
Figure 4 presents an outline of the PSOL Task
Ontology (Problem-Solving-Oriented Learning
Task Ontology). Ovals in the figure express a
cognitive activity performed by a learner in which
a link represents an is-a relationship.
The PSOL task ontology defines eight cognitive processes modeled in Rasmussens cognitive
model as lower concepts (portions in rectangular
box (a) in the figure). They are refined through
an is-a hierarchy to cognitive activities on the
meta-level (meta activity), and cognitive activities on the object level (base activity). Moreover,
they are further refined in detail as their lower
concepts: a cognitive activity in connection with

learning activities and a cognitive activity in connection with problem-solving activities. Thereby,
a conceptual system is constructed that reflects
the task structure of PSOL. For example, typical
metacognition activities that a learner performs
in PSOL, such as Monitor knowledge state
and Monitor learning plan, are systematized
as lower concepts of metacognition activities in
the Observe activity.
Figure 5 shows a conceptual definition of an
act that identifies a possible cause of why a plan
is infeasible. All the concepts in Figure 5 have
a conceptual definition in a machine readable
manner like this, thus, the system can understand
what the learner tries to do and what information
he/she needs.
Cause identification activities defined include:
the actor of the activity is a learner; a learners
awareness of infeasibility becomes an input
(in%symptom in Figure 5); the lower plan of an
target plan that the learner tries to make it feasible now is made into a reference information
(in%reference in Figure 5). Moreover, this cognitive activity stipulates that a learners awareness
of causes of infeasiblity is output (out%cause
in Figure 5). The definition also specifies that
the causes of the infeasibility include (axioms
in Figure 5): that the sufficiency of that target
plan is not confirmed (cause1 in Figure 5); that
the feasibility of a lower plan, small grained
plan that contributes to realize the target plan, is
not confirmed (cause2 in Figure 5); and that the
target plan is not specified (cause3 in Figure 5).
Based on this machine understandable definition,
the system can suggest the candidate causes of
infeasibility of the object plan, and the information the learner should focus on.
Making this PSOL task ontology into the basis
of a system offers useful information in the situation that encourages appropriate decision-making.
This is one of the strong advantages using PSOL
task ontology.

955

Task Ontology-Based Human-Computer Interaction

Figure 4. A hierarchy of problem-solving- oriented learning task ontology

Figure 5. A definition of identify c-state of NT in executable plan

956

Task Ontology-Based Human-Computer Interaction

An Application: Planning Navigation

as an Example
The screen image of Kassist, a system based on
the PSOL Task Ontology, is shown in Figure 6.
Kassist is an interactive open learner-modeling
environment. The system consists of six panels.
A learner describes a problem-solving plan, own
knowledge state about the object domain, and a
learning process in each panels of (a), (b), and
(c), respectively. Furthermore, a learner can describe the correspondence relationship between
the problem-solving process placed at (a) and
the concept of (b), that is, the correspondence
relationship with the knowledge of the object domain required for carrying out the process of (a);
and the correspondence relationship between the
learning process placed at (c), and the concept of

(b), that is, the correspondence relationship with

the learning process of (c) which constructs an
understanding on the concept of (b). Each shaded
node in (b) represents either knowing or not
knowing the concept. A learner can describe the
correspondence of the concepts and processes in
the object world placed on (a), (b), and (c) with
resource (f), used as those reference information,
so that appropriate information can be referred
to when required.
This provides a learner with an environment
in which she or he can externalize and express
her own knowledge; it then encourages his or her
spontaneous metacognition activities such as the
Activation activity and Observation activity in
Rasmussens model. Moreover, we can implement a more positive navigation function that
encourages a learners metacognition activity

Figure 6. Interactive navigation based on problem solving oriented learning task ontology

957

Task Ontology-Based Human-Computer Interaction

in the subsequent cognitive process by making

ontology the basis of a system.
Consider, for example, this task Investigate
how to develop XML-based document retrieval
system. Assume a situation where a learner does
not know how to tackle this task:
i.

In this situation, a learner clicks the Investigate how to develop XML-based

document retrieval system node on (c);
among the lower learning plans connected
by the part-of links, a plan Learn how to
design module structure, whose feasibility
is not secured is highlighted based on the
ontology; and a learner is shown the causes
of infeasibility with a message connected
lower plan is not feasible.
ii. Then, it shows the cause as a learner has
lack of knowledge to specify the plan.
iii. Moreover, plans influenced by the infeasibility of this learning plan are displayed in the
interpretation process.
iv. Here, a problem-solving plan Design
module structure is highlighted. Such
navigation allows a learner to comprehend
knowledge required to carry out problemsolving and to understand at what stage in
a problem-solving process such knowledge
is needed and their influence.
Thus, a learner can conduct appropriate decision-making by acquiring detailed knowledge
based on this modular design method.
A series of cognitive activities are typical
metacognition activities in PSOL. They include:
a learners awareness of feasibility of a learning
process as a start; monitoring ones own knowledge state; comprehending its influence on a
problem-solving plan; and building a learning plan
for mastering knowledge required for problem
solving. Reference to the appropriate information
offered by a system to a learner encourages his or
her appropriate metacognition activities, which
help implement effective PSOL.

958

Future trends
Ontology-Aware System
The systems which support users to perform
intelligent tasks based on the understanding of
ontologies are called ontology aware systems
(Hayashi, Tsumoto, Ikeda, & Mizoguchi, 2003).
Systemizing ontologies contributes to providing
theories and models, which are human-orientated
to enhance systems abilities of explanation and
reasoning. Furthermore, from the viewpoint
of system development, building systems with
explicit ontologies would enhance their maintainability and extendability. Therefore, future work in
this field should continue developing systems that
integrate ontology and HCI more effectively.

CONCLUSION
This article introduced a task ontology based
human computer interaction framework and
discussed various related issues. However, it is
still difficult and time consuming to build high
quality sharable ontologies that are based on the
analysis of users task activities. Thus, it is important to continue building new methodologies
for analyzing users tasks. This issue should be
carefully addressed in the future, and we hope
more progress can be achieved through collaboration between researchers in the fields of ontology
engineering and human computer interaction.

References
Brown, A. L., Bransford, J. D., Ferrara, R. A., &
Campione, J. C. (1983). Learning, remembering,
and understanding. In E. M. Markman, & J. H.
Flavell (Eds.), Handbook of child psychology (4th
ed.), Cognitive development, 30 (pp. 515-629).
New York: John Wiley & Sons.

Task Ontology-Based Human-Computer Interaction

DeBells, M. (1995). User-centric software engineering, IEEE Expert, 34-41.

Flavell, J. H. (1976). Metacognitive aspects of
problem solving. In L. Resnick (Ed.), The nature
of intelligence (pp. 231-235). Hillsdale, NJ: Lawrence Erlbaum Associates.
Gruber, T. R. (1993). A translation approach to
portable ontologies. Knowledge Acquisition, 5(2),
199-220.
Hayashi, Y., Tsumoto, H., Ikeda, M., & Mizoguchi,
R. (2003). Kfarm: An ontology-aware support
environment for learning-oriented knowledge
management. The Journal of Information and
Systems in Education, 1(1), 80-89.
Hori, M., & Yoshida, T. (1998). Domain-oriented
library of scheduling methods: Design principle
and real-life application. International Journal of
Human-Computer Studies, 49(4), 601-626.
Ikeda, M., Seta, K., & Mizoguchi, R. (1997). Task
ontology makes it easier to use authoring tools.
In the Proceedings of the Fifteenth International
Joint Conference on Artificial Intelligence 97
(IJCAI-97) (pp. 342-347).
Izumi, N., & Yamaguchi, T. (2002). Integration
of heterogeneous repositories based on ontologies
for EC applications development. International
Journal of Electronic Commerce Research and
Applications, 1(1), 77-91.
Mizoguchi, R. (2003). Tutorial on ontological
engineering - Part 1: Introduction to ontological
engineering. New Generation Computing, 21(4),
365-384.
Mizoguchi, R., Tijerino, Y., & Ikeda, M. (1992).
Task ontology and its use in a task analysis
interview systemTwo-level mediating representation in MULTIS. In the Proceedings of the
2nd Japanese Knowledge Acquisition Workshop
(JKAW) (pp. 185-198).

Mizoguchi, R., Tijerino, Y., & Ikeda, M. (1995).

Task analysis interview based on task ontology.
Expert Systems with Applications, 9(1), 15-25.
Okamoto, M. (1999). The study of metacognition in arithmetic word problem solving. Tokyo:
Kazama-shobo Publishing (in Japanese).
Rasmussen, J. (1986). A framework for cognitive task analysis. In Information processing
and human-machine interaction: An approach
to cognitive engineering (pp. 5-8). New York:
North-Holland.
Schreiber, G., Akkermans, H., Anjewierden, A.,
De Hong, R., Shadbolt, N., Van de Veide, W., &
Wielinga, B. (2000). Knowledge engineering and
managementThe common KADS methodology.
Cambridge, MA: The MIT Press.
Seta, K., Ikeda, M., Kakusho, O., & Mizoguchi,
R. (1997). Capturing a conceptual model for
end-user programmingTask ontology as a
static user model. In the Proceedings of the Sixth
International Conference on User Modeling, UM
97 (pp. 203-214).
Seta, K., Tachibana, K., Umano, M., & Ikeda, M.
(2003). Basic consideration on reflection support
for problem-solving oriented learning. In the
Proceedings of the International Conference
on Computers in Education (ICCE 2003), Hong
Kong (pp. 160-168).
Seta, K., & Umano, M. (2002). A support system
for planning problem solving workflow. In the
Proceedings of the International Conference on
Computers in Education (ICCE-02), Workshop
on Concepts and Ontologies for Web-based
Educational Systems, Auckland, New Zealand
(pp. 27-33).

KEY TERMS
Attentional Capacity: Cognitive capacity divided and allocated to perform cognitive task.

959

Task Ontology-Based Human-Computer Interaction

Metacognition: Cognition about cognition.

It includes monitoring the progress of learning,
checking the status of self-knowledge, correcting self-errors, analyzing the effectiveness of
the learning strategies, controlling and changing
self-learning strategies, and so on.
Ontology: A specification of a conceptualization (Gruber, 1993).
Problem-Solving Oriented Learning
(PSOL): Learning not only to build up sufficient
understanding for planning and performing prob-

lem-solving processes but also to gain the capacity

of making efficient problem-solving processes
according to a sophisticated strategy.
Rasmussens Ladder Model: A cognitive
model that models humans decision-making
processes. This model is often used for human
error analysis.
Task Ontology: A system of vocabulary/concepts used as building blocks for knowledge-based
systems.

This work was previously published in the Encyclopedia of Human Computer Interaction, edited by C. Ghaoui, pp. 588-596,
copyright 2006 by Information Science Reference, formerly known as Idea Group Reference (an imprint of IGI Global).

960

961

Chapter 2.33

Social Networking Theories and

Tools to Support Connectivist
Learning Activities
M. C. Pettenati
University of Florence (IT), Italy
M. E. Cigognini
University of Florence (IT), Italy

Abstract
This article considers the affordances of social
networking theories and tools to build new and
effective e-learning practices. We argue that
connectivism (social networking applied to
learning and knowledge contexts) can lead to a
reconceptualization of learning in which formal,
nonformal, and informal learning can be integrated as to build potentially lifelong learning
activities to be experienced in personal learning
environments. In order to provide a guide in the
design, development, and improvement both of
personal learning environments and in the related
learning activities, we provide a knowledge flow
model highlighting the stages of learning and the
related enabling conditions. The derived model is
applied in a possible scenario of formal learning

in order to show how the learning process can be

designed according to the presented theory.

Towards an e-lifelong
learning experience
Formal, nonformal, and informal learning have
become subjects of study and experimentation as
for their potentialities to be carried on through the
network. The pervasiveness of telematic technologies in current learning and knowledge processes
justifies the hopes of success and emerging approaches become always more open, destructured,
and nonformalised. According to this vision,
formal, informal, and nonformal learning can be
seen, such as integration of actions and situations,
that can be developed both in the network and in

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Social Networking Theories and Tools to Support Connectivist Learning Activities

physical contexts. New reflections can therefore

be made on the practice known as e-learning,
starting from a revision of these dimensions.
Formal learning has been defined as a type
of learning that occurs within an organized
and structured context (formal education, incompany training) and is intentional from the
learners perspective. Normally it leads to a formal recognition (diploma, certificate) (Cedefop,
2000; European Commission, 2000). As regards
adults e-learning, formal education in the last
decade has encountered and experimented a
sort of paradox that often witnessed low returns
in terms of knowledge acquisition, compared to
cost investment, which is often significantly high.
Nonformal learning has been defined as
learning embedded in planned activities that
are not explicitly designated as learning,
but that contain an important learning element. Nonformal learning is intentional from
the learners point of view (Cedefop, 2000).
Informal learning is learning resulting from daily
life activities related to work, family, or leisure. It
is often referred to as experiential learning, and
can, to a certain degree, be understood as accidental learning. It is not structured in terms of
learning objectives, learning time, and/or learning
support. Typically, it does not lead to certification. Informal learning may be intentional but
in most cases, it is nonintentional (or incidental/
random) (Cedefop, 2000). Informal learning is an
adaptive process determined by the exploration
need, which is realised in specific experiential
contexts (Calvani, 2005). People acquire their
competence in everyday life, talking, observing
others, trying and making mistakes, working
together with colleagues more or less expert. Informal learning can therefore be intended as the
natural corollary of daily life (Bonaiuti, 2006).
Intentionality of learning is a discriminating factor
shifting nonformal learning in nonintentional
or incidental learning; contrary to what happens
in formal learning, informal learning is not neces-

962

sarily intentional and can be nonrecognized sometimes from the subject himself/herself as knowledge and competence acquisition (Cross, 2006).
According to this perspective, aimed at retrieving and valuing the potentialities embedded in
spontaneous contexts, in this case the network,
the emerging domain of study of informal elearning is receiving greater attention because
of the widespread of social networking practices
and technologies. The online transposition of the
social network is nowadays referred to as social
networking phenomena, and it is related to a set
of available technologies and services allowing
individuals to take part in network-based virtual
communities. Social networking is emerging as a
highly natural practice because it is deeply rooted
in our daily behaviour; spontaneous relations,
interactions, and conversations support informal
learning practices, contributing to the creation and
transmission of knowledge. In informal learning
practices, the social behaviour and the support of
technologies converge toward the network; a
network made by people and resources, a social
network, unified by personal needs or common
goals, interaction policies, protocol and rules,
and telematic systems all together favouring
the growth of a sense of belonging to the net
community.
At the same time, the culture of lifelong learning is gaining importance as one of the most
effective answers to face the challenges brought
by the information and knowledge society (Siemens, 2006): the rapid obsolescence of professional knowledge and skills requires updating
and continuous training as well as recurring and
personalised learning. Under these premises, the
domain of e-lifelong learning is being configured
as a sociotechnical system in which knowledge
and learning are both the form and the content
as for their social and relational meaning. The
subject undergoing an e-lifelong-learning experience crosses this territory doing practices and
strategies of continuous interconnection and com-

Social Networking Theories and Tools to Support Connectivist Learning Activities

bination. In this work the informal e-learning area

constitutes a modularization of the educational
sociotechnical system towards the full enactment
of a lifelong-learning experience that could value
the subject as for his personal and professional
characteristics.
In 2004 the paper Connectivism: A learning theory for the digital age (Siemens, 2004)
launched the theory of connectivism based on a
critique of previous main learning theories synthetically labelled as behaviourism, cognitivism,
and constructivism. According to Siemens, even
the latter theory, which appeared to be the possible
theoretical framework for e-learning practices
(more specifically in its variant named social
constructivism) could not provide an adequate
theoretical support to the instances brought by the
new learning approaches. To fill this gap, Siemens
introduced the connectivism theory. According
to this author, The pipe is more important than
the content within the pipe, meaning that it is
the network itself that is the basis of the learning processes. If the knowledge society requires
the individual to continuously update his/her
knowledge, this cannot happen as a process of
progressive knowledge accumulation; instead,
this can occur through the preservation of our
connections. Learning, as it is also intended in
Wengers community of practice vision (Wenger,
McDermott, & Snyder, 2002) and in Lvys collective intelligence work (Lvy, 1996) is seen
as a mainly social activity. In this perspective,
the true competence for a lifelong learner of the
knowledge society would be the capability to stay
connected and belong to digital communities
with which interests are and can be continuously
shared. Owning a given piece of information is
less important than knowing where and how to
retrieve it.
Hereafter we will refer to e-lifelong learning
as a possibility that each knowledge society citizen has to build: an individual and personalised
lifelong-learning experience that will come across

formal, nonformal, and informal learning stages

with various degrees of uses of technology. In
this domain, theoretical reflection and applied
research is still at the beginning.
In this paper we try to provide a reference
knowledge flow model to support the design of
learning experiences in a networked environment
accounting for new practices and technologies of
social networking currently widespread in the Internet. The model aims also at giving suggestions
to designers of personal learning and personal
knowledge management environments in order to
maximize the advantages deriving from the effective networking to enhance and improve learning
and knowledge management functions.
To this extent, in paragraph 2, we analyse the
affordances brought by the connectivist learning
environment. In paragraph 3 we discuss the new
buzz word of the domain e-learning 2.0 as a pretext
to highlight methodologys and technologys role
in learning. In paragraph 4 we introduce the conceptual view of a personal learning environment
as a place where lifelong learning experience can
occur. Then, in paragraph 5 we revise the social
network technological applications that can serve
the scope of the personal learning environment
features. In paragraph 6 we discuss how learning
can take place in such a context, introducing a
knowledge flow model highlighting the learning
stages and related enabling conditions in a connectivist environment. Eventually, in paragraph
7, we illustrate how the model of paragraph 6
can be applied in a formal learning situation of a
possible higher education scenario.

Affordances of learning in a
connectivist Environment
Scenarios, which become always more common,
highlight that through informal channels, new
learning and knowledge management spaces are
more easily enabled, thanks to people and their

963

Social Networking Theories and Tools to Support Connectivist Learning Activities

ability to network and reciprocally learn in a

natural and spontaneous way. To synthetically
analyse the strengths emerging in the context of
learning in social network, we refer to Table 1.
From this perspective, the main characteristics
of social networks are schematized as for their
sharing/cooperation/collaboration features.
The concept of social networks and their
characteristics have been studied by Siemens
(2006) as for their potential in knowledge management. This author first introduced the term of
connectivism in order to make up for the lack of
theoretical framework to learning in the digital
age: Research in traditional learning theories
comes from an era when networking technologies were not yet prominent. How does learning
change when knowledge growth is overwhelming
and technology replaces many basic tasks we have
previously performed? (Siemens, 2006). Knowing and learning are today defined by connections... Connectivism is a theory describing how
learning happens in a digital age. Connectivism is
the assertion that learning is primarily a network
forming process (2006). Siemens statement adds

social characteristics to the territory in which

learning (lifelong learning) can occur.
Connectivism is characterised by nine principles (2006):
1. Learning and knowledge require diversity
of opinions to present the whole and to
permit selection of best approach.
2. Learning is a network formation process of
connecting specialized nodes or information
sources.
3. Knowledge rests in networks.
4. Knowledge may reside in nonhuman appliances, and learning is enabled/facilitated by
technology.
5. Capacity to know more is more critical than
what is currently known.
6. Learning and knowing are constant, on-going processes (not end states or products).
7. Ability to see connections and recognize patterns and make sense between fields, ideas,
and concepts is the core skill for individuals
today.

Table 1. Social network characteristics (Adapted from Pettenati & Ranieri, 2006b)
Social Network Characteristics
Relation based on individual interests, debate, confront on
Goal
specific topics; multiplicity and heterogeneity of joining
interests and motivations
Belonging
Spontaneous and autonomous motivation
Duration
Nondefined
High level of trust (relevance of reputation), sense
of responsibility, high technological skills, disCohesion and entributed
reflexivity
and
evaluation
(nonautoabling factors
nomous, nor heteronymous but socially spread)
Type of relation: share/evaluate

964

Social Networking Theories and Tools to Support Connectivist Learning Activities

8. Currency (accurate, up-to-date knowledge)

is the intent of all connectivist learning
activities.
9. Decision making is learning. Choosing what
to learn and the meaning of incoming information is seen through the lens of a shifting
reality. While there is a right answer now, it
may be wrong tomorrow due to alterations
in the information climate affecting the
decision.
According to these principles, learning, as
it was already conceptualised in Wenger et al.s
(2002) community of practice vision and in Lvys
(1996) collective intelligence theory, is seen as
a prevalently social activity. Connectivism is
the integration of principles explored by chaos,
network, complexity, and self-organization theories. Knowledge and learning are processes that
occur within nebulous environments of shifting
core elementsnot entirely under the control
of the individual. Learning, defined as knowledge patterns on which we can act, can reside
outside of ourselves (within an organization or a
database), is focused on connecting specialized
information sets (Siemens, 2006). Connectivism
is therefore social networking applied to learning
and knowledge.

A New Role for the Learner:

Continuously Change and Be
Creative
If from one side we value connectivism as a
context in which learning can more favourably
occur, thanks to available technological solutions
(Fallows, 2006), on the other side we acknowledge
that connectivism is also enabled and allowed by
an always stronger user participation to the creation, sharing, use, and management of resources
(contents, relations, applications, etc.) through
social software. Users have certainly become more
aware consumers and producers of resources; they
have become prosumers.

This radical change of the users role is under our eyes; between 2000 and 2004, relevant
literature reported about research investigating
the possible use of blog tools in education without
being capable of bringing credible answers. The
use of the tool at that time had not yet entered
in the praxis and communicative modality of
the people and the potentialities related to the
aspects of metareflection and self-evaluation of
the blog had not yet clearly emerged (Barrett,
2004; Barret 2006; Banzato, 2006). More recently,
the exponential growth of blog use, coupled with
the syndication technologies, which added the
relational dimension to these tools, made the
blog one of the most important social networking
assets used at the moment. This is just an example
that serves to highlight that the determinant variable of the shift in the use of the tool was not due
entirely to technology, but also to the spontaneous
change in the practice of use together with the
overcoming of the diffusion critical threshold that
could make the tool an important instrument in
learning (formal, informal, nonformal) processes.
Another relevant mark of this change in the users role is related to recent news; in December
2006, all the blogosphere rumoured about TIME
magazines cover title: TIMEs Person of the Year
for 2006 is you. According to Lev Grossman
(2006), author of this news, the explosive growth
and the enormous influence of the user-generated
content (such as blogs, video sharing sites, etc.)
should be read in a new way: But look at 2006
through a different lens and youll see another
story, one that isnt about conflict or great men.
Its a story about community and collaboration
on a scale never seen before. Its about the cosmic
compendium of knowledge Wikipedia and the
million-channel peoples network YouTube and
the online metropolis MySpace. Its about the
many wresting power from the few and helping
one another for nothing and how that will not
only change the world, but also change the way
the world changes.

965

Social Networking Theories and Tools to Support Connectivist Learning Activities

E-learning 2.0 or just

learning?
Innovation in technology has brought new (or
revisited) processes and practices, expressed
through new (or revisited) criteria and terms:
the possible educational universe of e-learning,
formal, informal, nonformal, lifelong coupled
with connectivism features has been named (or
tagged) e-learning 2.0 or learning 2.0, analogous
to what happened for the Web 2.0 phenomena
(OReilly, 2005). E-learning 2.0 is therefore a
tag identifying an open, destructured, immersive, and relational learning process amplifying
the learning curve towards the social, relational
side of knowledge coconstruction over the net.
Stephen Downes (2005) illustrated the main
concepts of e-learning 2,0:
1.
2.

3.

it is a type of learning that is tied to the

interests of the learners,
it is an immersive learning that is acquired
by doing, realised not only in the class,
but also in appropriate environments (for
instance museums...),
it is a type of learning in which Web 2.0
technologies connect the learners to the rest
of the world.

However stated, the focus stays on the attention

risen by the affordances that the social-technical
connectivist learning environment can engender,
in light of a background that remains primarily
and unavoidably methodological: the design and
development of educational environment is, beyond any revisiting, a learner/learning-centred
process.
When knowledge technologies change so
radically, they change not only what we
know, but also how we come to know (Laurillard, 2003). Knowledge is the result of a fluid
combination of experience values, contextual
information, and specialist competences, all together providing a reference framework for the

966

evaluation and assimilation of new experiences

and knowledge (Pettenati & Ranieri, 2006b).
Norris, Mason, and Lefrere (2003) highlights the
recursive, dynamic, and networked character of
learning in digital contexts: Knowledge can be
understood as interpreted content, available to a
member of a community and always shaped by
a particular context. Digital representations of
content and context become e-knowledge through
the dynamics of human engagement with them.
The digital elements of e-knowledge can be
codified, combined, repurposed, and exchanged.
The relational aspect of learning and knowledge
processes that is proposed in this work is supported by Wilsons (2002) essay The nonsense
of knowledge management, in which the author
states that: Knowledge is defined as what we
know: knowledge involves the mental processes
of comprehension, understanding and learning
that go on in the mind and only in the mind,
however much they involve interaction with the
world outside the mind, and interaction with others. Whenever we wish to express what we know,
we can only do so by uttering messages of one
kind or another oral, written, graphic, gestural or
even through body language. Such messages do
not carry knowledge, they constitute information, which a knowing mind may assimilate,
understand, comprehend and incorporate into
its own knowledge. Analogously, we attribute
to e-learning a social connotation, for us, elearning is a type of learning that is somewhat
supported by technologies, but it is not necessarily conducted at a distance; it allows interaction
between people and contents, and among people
but (most important) is a type of learning valuing
the social dimension of the underlying knowledge
processes (based on Calvani, 2005) definition,
freely translated and adapted by the authors).
The presence of the e- before all learning
and knowledge concepts is therefore now being
abandoned by many researches to underline that
pervasiveness of technologies do not differentiate
any more traditional contexts from virtual

Social Networking Theories and Tools to Support Connectivist Learning Activities

ones. The traditional concepts of learning and

knowledge, newly restated and deprived by the Internet-ic e-, globally revolutionize the education
and knowledge systems, both in the practices and
in the social processes they imply. As the learner
in an early stage of the knowledge society had to
change and shift identities and practice to access
the same processes over the network (Liguorio &
Hermas, 2005), so the learning and knowledge
practices regain possession of their integrity to
come back to their original complexity, now
completed and evolved by technologies but not
distorted by the same technologies.

Personal Learning Environments as

New Learning Landscapes
In this scenario, many of the innovative technologies of the so-called social software (see next
paragraph) are now playing a crucial role as a
support to learning and knowledge processes.
This does not mean that the scaffold provided up
to now by a formalised online educational environment is to be entirely rebuilt, but the current
extended cognitive context (environmental,
situational, cultural characteristics) in which
the learning dynamics occur must reshape the
learning environment itself (Bonaiunti, 2006).
Metaphorically getting out from the often narrow
space of the learning/course/content management
systems is equivalent to the reasoning conducted
up to now: learning does not take place only in
classes, for a limited period of time, and under
specific certification conditions but (in light of
the sociotechnical context now available) can
easily take place in nonformal, informal, lifelong
settings, prospecting new, challenging and interesting development perspectives. Most evidently,
according to this perspective, new coordination
and integration needs emerge, as for the experiences conducted in heterogeneous systems,
from the functional and from the pedagogical
and methodological viewpoints. A new technological solution is never a didactic solution, Quite

the opposite, it urges to be considered after the

methodological design has been carried out, in
order to properly balance the use of the tools with
respect to learning objectives, learners, available
resources, and infrastructures (Ranieri, 2005).
Stating that the social software (or Web 2.0)
technological apparatus has great educational
potentials would be reductive and misleading;
which type of learning can occur and under
which conditions and ways this learning can occur should always be methodological priorities
to be accounted for during instructional design.
The shift required at this stage of the reasoning
can be conducted with the avail of a specific
environment, personalised on the needs and the
competences of the learner, which could always
be a door to enter learning experiences and a
glue to support the learner in mastering the
heterogeneity of his learning experiences, easily
jumping across formal, informal, unexpected,
and intentional learning activities (Conner, 2004).
This door-glue-environment is represented in
the hopes of the practitioners of this domain, and
in concrete available solutions, by personal learning environments (PLE) that are a concretization
of operational spaces in the network where the
subject is at the centre of his network of learning
resources (Bonaiuti, 2006).
PLEs represent the temporal horizon through
which we look at the potential lifelong learning
of a subject (Tosh, 2005; Tosh & Werdmuller,
2004); they are an open, transversal, and personal environment intercepting and capitalizing
learners competences and skills by aggregating, publishing, combining, and integrating
diverse and distributed resources and contents.
E-portfolios are another way to name PLEs (ePortfolio, 2006); an e-portfolio can be seen as
a type of learning record that provides actual
evidence of achievement. Learning records are
closely related to the Learning Plan, an emerging tool that is being used to manage learning
by individuals, teams, communities of interest
and organizations. The attention given recently

967

Social Networking Theories and Tools to Support Connectivist Learning Activities

to this topic at the eStrategy conference on ePortfolio (Baker, 2006) is but another evidence
of the rising importance of studying this field.
In the context of a knowledge society, where
being information literate is critical, the portfolio
can provide an opportunity to demonstrate ones
ability to collect, organise, interpret and reflect on
documents and sources of information. It is also
a tool for continuing professional development,
encouraging individuals to take responsibility for
and demonstrate the results of their own learning.
Furthermore, a portfolio can serve as a tool for
knowledge management, and is used as such by
some institutions.

Technology to Support the

Learning Experience
The further step in the analysis leads us to
the problem of evaluating and devising which
tools and technologies exist or can be developed in order to match the requirements and
purposes of the personal learning environment.
Technologies and tools now referred to as Web
2.0 software (Fallows, 2006; Hinchcliffe, 2006;
OReilly, 2005) certainly provide both the origin
of this reasoning as well as the goal at which to
aim. Sharing Paul McFedries (McFedries, 2006)
tentative definition according to which Web 2.0
is a second phase of the evolution of the World
Wide Web in which developers create Web sites
that act like desktop programs and encourage
collaboration and communication between users,
we focus on the characteristics of the Web 2.0
applications, highlighting the social perspective
of relation, collaboration, and user-participated
architecture (McFedries, 2006):
1.
2.
3.

968

content is user-created and maintained (peer

production, user-content ecosystem),
user-created and maintained content require
radical trust,
application usability allows rich user experience,

4.
5.

combining data from different sources leads

to creation of new services,
services get better as the number of users
increase in an architecture of participation.

Web 2.0 tools

Folksonomies, cobrowsering, tagging, and social
networking are 2.0 practices. The fil rouge
they share is that they all are expressions of a
shared, diffused cognitive strategy for information retrieval in a spontaneous way, as support to
social sharing tools (such as social bookmarking
tools, image sharing tools, blog search engines,
etc.). Through social tagging, the member of the
community defines a link among resources (sites,
images, videos, audios, etc.) and the terms used to
describe them (Bonaiuti, 2006). This is a bottomup process, starting from a single user adding a
link to a Web site and tagging it at his complete
discretion, using keywords that are meaningful to
himself/herself. Social sharing tools can display
these tags through using a visual approach (which
increases the font size of most popular tags), thus
realising tag clouds that immediately provide users
with a perception of the popularity of the tags.
This folksonomic classification method, which
relies on the spontaneous users contributions (be
he/she an author or a sporadic resource user), leads
to results capable of reflecting the information according to the conceptual model of the population
that creates it (Bonaiuti, 2006). Representing information in classificatory structure is information
in itself: classification incorporates information
and provides the interpretation context that (in
its globality) appears transparent, objective, and
neutral (Surowiecki, 2004).
Practices and applications described as an
example implement ideas derived from social
network theory, which sees social relations
expressed through nodes and links in which
nodes represent the subjects in the network and

Social Networking Theories and Tools to Support Connectivist Learning Activities

the links represent relations between subjects.

Under this theoretical basis, these applications
explicitly represent relations favouring the in-

terconnection among subjects based on interest,

competence, hobby, research goals, study, work,
and so forth (Fini, 2006). Social networking is

Table 2. Sample Web 2.0 applications, a synthetic view

Web application

Description

Social networking, online

social networks

Category of Internet applications to help connect friends,

business partners, or other individuals together using a variety of tools.
Social network search engines are a class of search engines
that use social networks to organize, prioritize, or filter
search results.
A Web log is a Web site where entries are displayed in chronological order. They often provide commentary or news on
a particular subject, typically combining text, images, and
links to other blogs, Web pages, and other media related to
the specific topic.
Specialized search engines for searching blog and news contents.
Ad hoc classification scheme (tags) that Web users invent as
they surf to categorize the data they find online.
Saving and applying keywords to ones personal collection
of Web site bookmarks on a site that enables other people to
share those bookmarks.
Web syndication is a form of syndication in which a section
of a Web site is made available for other sites to use through
to making Web feeds available from a site in order to provide
other people an updated list of content from it (for example
ones latest forum postings, etc.).
A list of tags used in the site with some kind of visual indication of each tags relative popularity (ex. large font). Web
sites that implement tag clouds functions allow both finding
a tag by alphabet and by popularity. Selecting a single tag
within a tag cloud will generally lead to a collection of items
that are associated with that tag.
Web sites combining social bookmarking, blogging, and syndication with a form of nonhierarchical, democratic editorial
control. News stories and Web sites are submitted by users,
and then promoted to the front page through a user-based
ranking system.
Collaborative Web sites that allow users to add, edit, and delete content.
Simultaneous editing of a text or media file by different participants on a network.

Social network search

engines
Blogs

Blog guides
Social tagging (folksonomy)
Social bookmarking
Web syndication, Web feed
management

Tag clouds

Peer production news

Wikis
Collaborative real-time
editing

continued on following page

969

Social Networking Theories and Tools to Support Connectivist Learning Activities

Table 2. continued
Content aggregation and
management, Mashup (Web
application hybrid)
Instant messaging
Podcasting

Personal and group organization

A Web site or Web application that combines content from

more than one source.
A form of real-time communication between two or more
people based on typed text.
Podcasting is the method of distributing multimedia files,
such as audio programs or music videos, over the Internet
using either the RSS or Atom syndication formats, for playback on mobile devices and personal computers.
Allow collaborative tools, project management, calendar,
and so forth.

the connective tissue in which new applications

emerge daily, allowing us to creatively use them
for learning purposes. The following table reports
a synthetic view of some of the currently available
applications matching this philosophy; however,
the list is neither complete nor exhaustive, and is
intended just to be a memorandum for personal
learning environment designers.

A model for effective

learning experience
To comply with the objectives detailed in the
previous paragraphs, we hereafter propose
a schema to illustrate a (simplified) knowledge f low in a connectivist environment.
This model is the result of a reelaboration of four
models:

970

the online collaboration model presented

in Calvani (2005), which was the starting
point of our analysis; this model accounts
for effectiveness conditions and principles
that are considered to be fundamental for
collaboration, as highlighted in reference

literature (Dillenboug, Baker, Blaye, &

OMalley, 1996). However, while in Calvani
(2005) the model is conceived to provide
useful steps to support an online collaborative group, the model is rooted in a formal
educational context;
the second model studied is the one presented
in Ranieri (2006); this model accounts for
individual and collaborative knowledge
construction processes, but it is not specifically conceived focused on telematic
environments, and does not account at all
for collaboration enabling conditions;
the third and fourth models are those presented in Pettenati and Ranieri (2006a,
2006b) in which the authors tried to focus on
framing the reasoning in a social networking
context in order to account for the benefits of
informal learning and online collaboration,
as described in the previous paragraphs of
this paper. Nonetheless, that model did not
account for the knowledge flow (or learning stages) occurring in an online learning
experience, thus being only partially helpful
in the design of a learning experience.

Social Networking Theories and Tools to Support Connectivist Learning Activities

After having confronted the previous three

models with the knowledge flow presented in Siemens (2006), the idea of the model was repurposed
in light of the connectivist idea of the authors
work, thus leading to a new design of the model
now focused on highlighting the possible stages on
an effective connectivist learning experience. In
this new version, the schema can provide more concrete help in the design of online learning activities,
as will be detailed in section 7, using an example.
In Figure 1, the transformation processes that involve personal comprehension and knowledge as
well as social knowledge creation are illustrated in
order to highlight some important processes that
should occur during an effective learning experience. Rectangles in Fig. 1 represent the knowledge
processes whose results can be concretized in various forms of knowledge (not illustrated in Figure1)
such as enunciations, argumentations, justification, shared meanings, collaborative knowledge,
cultural artefacts, and so forth (Ranieri, 2006).

The schema in Figure 1 has no claims to be

exhaustive neither perfectly accurate, as knowledge processes are complex, fluid, iterative, and
certainly not univocally directed, thus leading to
various relations among the composing elements
(represented by rectangles in Figure 1). However
this schema will be used in section 7 to drive the
choice of methodologies and available technologies and their combination in order to serve the
educational needs in a possible higher education
scenario.

Enabling Conditions
The model in Fig. 1 envisages five subsequent
stages (or knowledge processes) that are at the
heart of the schema. The processes are framed by
an external layer, where the enabling conditions
that are relevant for the knowledge processes
development are highlighted:

Figure 1. Knowledge process in a connectivist environment; stages of the learning experience and
enabling conditions

971

Social Networking Theories and Tools to Support Connectivist Learning Activities

1.

2.

3.

4.

5.

972

Basic skills: The acquisition of basic technological as well as online-comunicational

skills is fundamental (Ranieri, 2006) to
allow the subject to positively enter and
live the informal and potentially lifelong
educational experience; the lack of these
conditions can compromise the success of
the entire learning initiative;
Generation and support to motivation: In
informal e-learning contexts the motivation
is spontaneous; it is often induced by fun
and pleasure that individuals have in their
network activity; it is also rooted in the
positive interaction among people (a subject
can more effectively and efficiently pursue
his objective if the other subjects pursue
theirs);
Meaning perception: The subject must
perceive as really meaningful (useful to
himself/herself) the objectives attainable
in the learning activities and acknowledge
that collaboration can derive real advantage;
during the collaboration phases, supporting
the individual self-perception of usefulness
allows the subject to perceive the significance
of its contribution to group activities in order
to consider himself a useful contributor to
others goals;
Group culture: The awareness of being
useful to other community members increases the self-esteem and self-perception
of usefulness of the individual and fosters
the motivation for a wider visibility (for
instance being linked, have positive reputation, produce and/or propose new contents).
These factors increase the sense of positive
group membership;
Social climate: In an informal environment, the sense of belonging (membership)
to a group is spontaneously supported by
the intensity of sharing interests on a topic;
regardless of the expertise (which can be
widely nonhomogeneous among members)

it is still the awareness of the positive interaction with others that sustains mutual
understanding and social grounding; in this
context the (often tacit) agreement of respect,
use of reputation feedback, and respect of a
common socioquette, contribute to build a
positive social climate, making the online
relational environment a trusted environment.

Stages of the Learning Experience

In the centre of Figure 1, a simplified schema of
the stages of a connectivist learning experience
are schematized:
1.

2.

3.

Awareness and receptivity: Siemens (2006)

points out, in this stage, individuals get used
to handling knowledge abundance, and are
firstlconfronted with resources and tools of
the new learning habitat. If this stage is not
sustained by a proper motivational context
as well as the acquisition of basic skills, as
detailed in paragraph 5.1, this phase can
become a frustrating experience and often
cause the learner to drop out from the learning environment.
Connection forming and selection filtering: In this stage, individuals begin to use
tools and understanding acquired during the
previous stage to create and form a personal
network of resources (people and contents).
At this point the learners start to be active in
the learning space in terms of consuming
or acquiring new resources and tools (Siemens, 2006). Affective and emotive factors
such as fun, pleasure, positive interaction,
and increasing sense of meaning are crucial
to building the roots of an effective personal
learning network (ibidem).
Contribution and involvement: In this
stage the learner begins to actively contribute to the learning network, essentially,

Social Networking Theories and Tools to Support Connectivist Learning Activities

4.

becoming a visible node. The learners

active contribution and involvement allows
other nodes on the network to acknowledge
his/her resources, contributions, and ideas,
creating reciprocal relationships and shared
understandings (or, if social technology is
used, collaboratively-created understanding) (ibidem). In this stage group culture and
social climate are key factors enabling the
individual participation and involvement.
Reflection and Metacognition: Reflection
on the knowledge processes and products,
self-reflexivity, and self-evaluation as well
as metacognition (thinking about thinking)
play a prominent role in this stage. Individuals are actively involved in modifying
and rebuilding their own learning network,
acting as network aware and competent
subjects (ibidem). The experience acquired
at this stage within the learning network has
resulted in an understanding of the nuances
of the space and the knowledge inputs, allowing the subject to act both as a provider
of valuable support and to help other networked learners as well as being capable
of accessing just in time and personalized
knowledge to himself/herself.

The learning process depicted can be used as a

model to support design of learning activities and
environments in different educational sets. An example of this approach will be shown in paragraph
7 to build a connectivist learning experience in
a possible higher education scenario.

Building a Connectivist Learning

Experience: A Possible Scenario
This scenario subject is taken by Bonaiuti (2006),
but is significantly detailed, revised, and newly
implemented according to what is presented in this
paper. More specifically, the focus is on building
and exploiting a connectivist online environment

according to the model in Figure 1, to match the

educational purposes.
Scenario: Higher education, course in Urban
Sociology, class topic: Juvenile behaviour in
urban suburbs (Bonaiuti, 2006). Scenario details: this course envisages 2 hours face-to-face
class time and a total of 10 hours to be devoted to
autonomous and group work. The teacher wants
to structure the face-to-face meeting as a role
playing activity. The scenario for the role play
is the following: participants: a math teacher, a
chancellor, two parents, a young boy. The young
boy has been caught by the math teacher outside
of the school while cutting the teachers car
wheels. The teacher has reported the happening
to the chancellor and the student is now risking
expulsion from the school. Debate with arguments in favour of expulsion, neutral, in defence
of the student, and reflective, will be played in
1 hour during the face-to-face class meeting.
The model of Figure 1 can be used to help the
teacher settle his/her teaching activities. To this
extent, the model is exploited during the three
phases of class activities: preclass activities, in
class activities, postclass activities. Instrumental
background: the university in which this scenario
is set has installed a learning environment at the
disposal of teachers and students who want to
avail of it. This environment is a combination
of a traditional e-learning environment and
a personal learning environment conceived
according to the e-learning 2.0 philosophy, as
detailed in the previous part of this work, and
opened to students and alumni who want to keep
alive their online lifelong learning landscape.
Such a combined environment offers both the
course and content management system offered
by traditional e-learning systems coupled with
social-networking creation, sharing, and management tools.

973

Social Networking Theories and Tools to Support Connectivist Learning Activities

tag it with the group name). Students can

use the social bookmarking feature of the
course environment to organize and share
their resources.

1. Preclass Activities: Focus on

Content Management and Sharing
a.

b.

c.

974

Awareness and receptivity: teacher gives

starting tasks to the student using the blog
of the learning environment in the area
reserved for the Urban Sociology course.
In this post, he/she proposes a selected
reference reading (full text of the reading
could be stored in the Course Material area)
and suggests a vision of three short videos
distributed through a video-sharing social
network. The assignment for this phase is
to prepare the debate that will take place
during the face-to-face class: students are
divided in five groups; each group is charged
to represent one role in the scenario: math
teacher, chancellor, two parents, young
boy. Each group, in order to get prepared to
sustain the arguments and debate, can find
external documentation to get closer to the
discussed reality of juvenile behaviour in
urban suburbs (multimedia resources are
allowed, such as podcast, video, text, etc.).
Connection forming, selection, filtering:
in this phase students prepare themselves to
work in a group; each group decides how to
carry on its task (for instance attribute roles
to the group members, share tasks, provide
complementary resources, etc.). This work is
done in spontaneous modes for each group:
some can use the course forum; teacher
can open related threads, or can work in
synchronous mode using voiceIP tool and
sharing ideas and outline, or to dos using the
class wiki or a conceptual map tool; teacher
can provide a tool area for each group of the
course.
Contribution and involvement: in this
phase students actively work using tools and
combining resources to come to a shared
strategy of discussion for the class meeting.
Students post the outline of their argumentations in the course blog or forum (and can

2. In-Class Activity: Focus on Content

Production
a.

b.

Contribution and involvement: during the

first hour of the class the role-playing game
is set. The teacher listens to the debate and
uses a mental map tool to synthesize the
emerging discourse pattern.
Reflection and metacognition: after the
end of the play, the teacher restates what
happened and comments on group contributions. He/she provides ideas and questions
for further insights and reframes all the play
in the context of the topic of the class, with
reference to the selected reading assigned
before the class. At the end of the class,
both the conceptual maps drawn and the
audiorecord of the play are shared on the
course environment.

Students are then given the evaluation assignment to be carried in a post-class phase within
15 days. The assignment is a journalist-style
individual writing of a possible local newspaper
reporting the case of the role play in light of the
analysis of the juvenile behaviour in the urban
suburbs. Students are asked to post the assignment
in their personal blog, tagging it with the course
name and class topic.

3. Post-Class Activities: Focus on

Knowledge Acquisition
a.

Contribution and involvement: students

will work individually to study and analyze
both the material produced by the whole class
and the material provided by the teacher.
Students can work at the writing using either
an online (to have easy backup and document

Social Networking Theories and Tools to Support Connectivist Learning Activities

b.

portability) or desktop word processor, or a

wiki (to keep trace of the evolving versions).
Eventually they will post the assignment in
their personal blog, tagging it as required.
The teacher will easily follow the assignment
submissions using a syndication system.
Reflection and metacognition: to support
the production phase the teacher (or a tutor
acting on his/her behalf), in the meanwhile,
will use the course blog to more deeply
comment on the topic or provide further
resources, as well as posting comments on
students posts (both on the blog, or forum
or wiki). After the evaluation phase (marks
have been given to students assignment in
a privacy-respectful way), the teacher could
invite students to digg their preferred
writing to highlight them in the school
journal.

Conclusion
In this article we tried to provide our interpretation of the current sociotechnical educational
system shaped by technologies and practices
of the knowledge society to locate the role of
learning and learners in a lifelong perspective.
We believe that both users attitudes and available
technologies are mature enough to let us envisage
that each network user could easily engage in a
lifelong learning personal experience if properly
lead by appropriate methodologies, and sustained
by accordingly designed and developed personal
learning environments.
To this extent we provided a model to schematize the knowledge flow occurring during an
effective learning experience in a connectivist
environment. The purpose of this model is twofold: from one side it can be used by personal
learning-environment designers as a guideline
for checking if all phases and enabling conditions are supported by the integrated tools;
on the other side it can be used by instruc-

tors or designers to set up learning activities.

The provided model has some common points with
other models at the state of the art; for instance
in Salmon (2002), the author provided a staged
model for online collaboration activities oriented
to knowledge construction; however, that model,
and the related implementations and example presented in the essay, was developed in the 1.0 age,
where neither technology nor people practice was
mature enough in the sense of social networking
perspective. We believe that a connectivist revisitation of this theory is now necessary, and this
work intended to be a contribution in this sense.
As a conclusion of the paper, we provided a sample
application of the model using the scenario. The
scenario setting chosen for this purpose is a
formal learning higher education context; this
choice should not be read as contrasting with
the lifelong perspective of learning claimed
throughout this work. Instead, it should be read
as a preparatory activity that could enter in
peoples practice diffused by a formal context but
(once mastered as skill) potentially applied in all
other nonformal and informal learning contexts.
Applying the knowledge processes flow model in
other scenarios such as professional training in
enterprises, as well as refining and detailing the
model to these purposes, could be an object of
future study. Many other issues are still open for
further investigation and experimentation: how
could a personal learning environment be used to
develop pedagogical thinking? How can they be
used in professional contexts such as workers for
placements/reintegration?, How can the formal
educational system be changed as to support selfdirected learners?, What arguments can convince
politicians (educational and labour ministries)
to support individual institutions? These questions, as many others, were not addressed in this
work, but can and will be object of future studies.
Whether we call it e-learning, e-learning 2.0,
lifelong learning, knowledge management or not,
the interesting question is to see if and how these
approaches can lead us to a true actualization

975

Social Networking Theories and Tools to Support Connectivist Learning Activities

of a lifelong learning practice for all knowledge

society members.

Acknowledgment
We want to thank Prof. Dino Giuli for giving us
the possibility to carry on our research in this
domain. We are also very grateful to Prof. Antonio Calvani for keeping on asking us stimulating
scientific research questions, pushing us to work in
order to find possible answers. Moreover, we want
to thank the colleagues of the Educational Technology Laboratory and Telematics Technologies
Laboratory for the fruitful discussions occurring
both in formal and nonformal settings.

References
Baker, A. (2006). E-strategies for empowering
learners. ePortfolio Conference, Oxford, England,
on 11-13 October 2006. Retrieved on January 2007,
from, http://www.eife-l.org/news/ep2006
Banzato, M. (2006). Blog e didattica. Dal web
publishing alle comunit di blog per la classe in
rete. TD Tecnologie Didattiche, 38(2).
Barrett, H. (2004). Myonline portfolio adventure.. Retrieved on January 2007, from, http://
electronicportfolios.org/myportfolio/versions.
html
Barrett, H. (2006). Authentic assessment with
electronic portfolios using common software and
etrieved on January 2007, from,
Web 2.0 tools. R
http://electronicportfolios.org/web20.html

Calvani, A. (2005). Reti, comunit e conoscenza: costruire e gestire dinamiche collaborative.

Trento: Erickson.
learning
Cedefop Glossary. (2000). In Making
Cedefop, Thessaloniki.
visible.
Colardin, D., & Bjornavold, J. (2004). Validation of
formal, non-formal and informal learning: Policy
and practices in EU Member States. European
Journal of Education, 39(1), 69-89.
Conner, M. L. (2004). Informal learning. Ageless
learner. Retrieved on January 2007, from, http://
agelesslearner.com/intros/informal.html
Cross, J. (2006). Informal learning for freerange learners. Internet Time Group LLC.
Retrieved on January 2007, from, http://www.
jaycross.com/informal_book/nutshell.html
Denning, S., Pommier, M., & Shneier, L. (2002).
Are there laws of knowledge management? In
J. Mason (Ed.), Are there laws of knowledge
management. Retrieved on January 2007, from,
http://www.educationau.edu.au/globalsummit/
papers/denning.htm
Dillenbourg, P., Baker, M., Blaye, A., & OMalley,
C. (1996). The evolution of research on collaborative learning. In P. Reinmann, & H. Spada (Eds.).
Learning in humans and machines: Towards an
interdisciplinary learning science (189-205).
Pergamon, Oxford.
Downes, S. (2005). E-learning 2.0. eLearn
Magazine, 10.

Bonaiuti, G. (2006). Learning 2.0. I

Quaderni di
Trento: Erickson.
Form@re.

e-Por tfolio. (2006). eStrategies for empowering learners. ePortfolio Conference.

Oxford, England, on 11-13 October 2006
Retrieved on January 2007, from, http://www.
eife-l.org/news/ep2006

Bullen, M. (1998). Participation and critical

thinking in online university distance education.
Journal of Distance Education, 2, 1-32.

European Commission. (2000). Memorandum on

lifelong learning. Brussels: Commission of the
European Communities.

976

Social Networking Theories and Tools to Support Connectivist Learning Activities

Fallows, J. (2006). Homo conexus. Technology Review. July 2006. Retrieved on January 2007, from,
http://www.technologyreview.com/read_article.
aspx?id=17061&ch=infotech.

Norris, D., Mason, J., & Lefrere, P. (2003). Transforming e-knowledge - A revolution in the sharing
of knowledge. Society for College and University
Planning Ann Arbor. Michigan.

Fini, A. (2006, October). Nuove prospettive tecnologiche per l'e-learning: dai Learning Object
ai Personal Learning Environment.Master thesis
in educational sciences. University of Florence.

OReilly, T. (2005). What Is Web 2.0. Retrieved

on January 2007 http://www.oreillynet.com/
pub/a/oreilly/tim/news/2005/09/30/what-is-web20.html

Granieri, G. (2005). Blog generation. Bari: Laterza.

Pettenati, M. C., & Ranieri, M. (2006a).

Informal learning theories and tools to support
knowledge management. In Distributed CoPs.
TEL-CoPs06: 1st International Workshop on
Building Technology Enhanced Learning solutions for Communities of Practice, held in conjunction with the 1st European Conference on
Technology Enhanced Learning Crete, Greece,
October 2, 2006.

Grossman, L. (2006). Times Person of the Year:

You. Times Journal, Wednesday, Dec. 13, 2006.
Retrieved on January 2007, from, http://www.
time.com/time/
Hall, E. T. (1966). The hidden dimension. New
York: Garden-City, Doubleday.
etrieved
Hinchcliffe, D. (2006). Web 2.0 Blog. R
on January 2007, from, http://web2.wsj2.com
Jonassen, D. H. (1999). Costructivist

learning environments. In C. M. Reigeluth (Ed.), Instructional

design theories and models: A new paradigm of
instructional design, Vol.II (pp 215-239). Mahwah, NJ, Lawrence Erlbaum Associates.
Jonassen, D. (2000). Computers in the classrooms:
Mindtools for critical thinking. Englewood Cliffs,
NJ: Prentice Hall.
Laurillard, D.( 2003). The changing university
[Online]. Paper presented to the Instructional
Technology Forum, University of Georgia. Retrieved from http://itech1.coe.uga.edu/itforum/
paper13/paper13.html
Lvy, P.(1996). Lintelligenza collettiva. Per unantropologia del cyberspazio. Milano: Feltrinelli.
Liguorio, B., & Hermas, H. (2005). Identit dialogiche nellera digitale. Trento: Erickson.
McFedries, P. (2006, June). The Web, take two.
IEEE Spectrum, 68.

Pettenati, M. C., & Ranieri, M. (2006b). Dal s

alle reti: nuove pratiche di social network per la
collaborazione in rete. In G. Bonaiuti, Learning
2.0. Erickson: Trento.
Ranieri, M.(2005). E-learning: Modelli e strategie
didattiche. Trento: Erickson.
Ranieri, M., (2006). Formazione e cyberspazio,
divari ed opportunit nel mondo della rete. ed.
ETS.
Roncaglia, G., Calvo, M., Ciotti, F., Zela, M. A.
(2004). Internet 2004 - Manuale per luso della
Roma: Laterza.
rete.
Salmon, G. (2002). E-tivities: The key to active
online learning. Kogan Page, Limited.
Siemens, G. (2004). Connectivism: A learning
theory for a digital age. Elearningspace.org 12
December, 2004. Retrieved on January 2007,
from, http://www.elearnspace.org/Articles/connectivism.htm
Siemens G. (2006). Knowing knowledge. Retrieved on January 2007, from, http://www.
knowingknowledge.com

977

Social Networking Theories and Tools to Support Connectivist Learning Activities

Sorrentino F., & Paganelli F. (2006). Lintelligenza

distribuita. Ambient Intelligence: il futuro delle
tecnologie invisibili. Erickson:Trento.
Surowiecki, J. (2004). The wisdom of crowds:
Why the many are smarter than the few and how
collective wisdom shapes business, economies,
societies and nations. Little, Brown.
Tosh, D. (2005). A concept diagram for the personal learning landscape. Retrieved on January
2007, from, http://elgg.net/dtosh/weblog/398.
html

Tosh, D., & Werdmuller B. (2004). Creation

of a learning landscape: Weblogging and social networking in the context of e-portfolios.
Retrieved on January 2007, from, http://eradc.
org/papers/Learning_landscape.pdf
Wenger, E., McDermott, R. E, & Snyder, W.
M. (2002). Cultivating communities of practice.
Boston: Harvard Business School Press.
Wilson, T. D. (2002). The nonsense of knowledge
management. Retrieved on January 2007, from,
http://informationr.net/ir/8-1/paper144.html

This work was previously published in the International Journal of Web-based Learning and Teaching Technologies, edited
by L. Esnault, Volume 2, Issue 3, pp. 42-60, copyright 2007 by IGI Publishing, formerly known as Idea Group Publishing (an
imprint of IGI Global).

978

979

Chapter 2.34

User-Centered Design
Principles for Online Learning
Communities:

A Sociotechnical Approach for the Design

of a Distributed Community of Practice
Ben K. Daniel
University of Saskatchewan, Canada
David OBrien
University of Saskatchewan, Canada
Asit Sarkar
University of Saskatchewan, Canada

Abstract
This chapter examines current research on online
learning communities (OLCs), with the aim of
identifying user-centered design (UCD) principles
critical to the emergence and sustainability of distributed communities of practice (DCoPs), a kind
of OLC. This research synthesis is motivated by
the authors involvement in constructing a DCoP
dedicated to improving awareness, research, and
sharing data and knowledge in the field of governance and international development. It argues

that the sociotechnical research program offers

useable insights on questions of constructability.
Its attention in particular to participatory design
and human-computer interaction are germane
to designing user-centered online learning communities. Aside from these insights, research has
yet to probe in any systematic fashion the factors
affecting the performance and sustainability of
DCoP. The chapter concludes with a discussion
of UCD principles for online learning community
to support the construction and deployment of
online learning communities.

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

User-Centered Design Principles for Online Learning Communities

Introduction
Increasingly, distributed communities of practice
(DCoPs) are attracting attention for their potential
to enhance learning, to facilitate information
exchange, and to stimulate knowledge creation
across cultural, geographical, and organizational
boundaries. Research shows the utility of
DCoP
on their members is positive (Daniel, Sarkar, &
OBrien, 2004a; Daniel, Poon, & Sarkar, 2005;
Schwier & Daniel, Chapter II, this volume).
Their allure aside, experience indicates that they
may not emerge or flourish even in the presence
of demand from users. In fact the process

of
constructing DCoP is not well understood, and
factors influencing sustainability merit further
research attention.
This chapter introduces the authors involvement in the development of a DCoP. The DCoP in
question is the Governance Knowledge Network
(GKN). This project began in 2001 with the aim
of assessing the interest of academics and practitioners in Canada to develop an online learning
community (OLC) for systematizing the exchange
of information at the intersection of governance
and international development
(Daniel et al.,
2004a).
The surveys of key Canadian stakeholders
in the project indicated considerable data existed,
and recommended the proposed GKN to: actively
engage in dissemination and archiving of data
not widely accessible in the public sphere, profile
community members, promote social network
building and collaboration, and inform members
of current events and opportunities.
Following the identification of the demand and
interest, the second stage of our research involved
the development of a GKN prototype. In this
unchartered course, we were guided by enabling
technology and other DCoP models (World Bank,
UNDP).1 We also turned to research to inform our
efforts on how to effectively sustain the project.
Our synthesis of research in the area identified
promising insights from studies we refer to as
the sociotechnical approach. As applied to DCoP,

980

the sociotechnical approach aims at understanding peoples interaction with technology and the
ensuing communication, feedback, and control
mechanisms necessary for people to take ownership of the design and implementation process.
This chapter focuses on this interaction, as it is
germane to the development and sustainability of
the GKN, in particular, and DCoP more generally.
The chapter is divided into the following sections.
The next section outlines relevant research on
DCoPs and the sociotechnical approach. We next
provide an overview of the GKN OLC project
and present key results from the research that
informed the design of the GKN. A discussion
of various human and technology elements we
consider critical to the initiation, development,
growth, and sustainability of the GKN follows,
and in the next section, we revisit the key human
and technology design issues. Finally, we conclude
the chapter and present UCD principles for OLCs
drawn from the sociotechnical approach.

Related Work
Daniel, Schwier, and McCalla (2003b) observe that
online learning communities have attracted diverse disciplinary interest, but that it is possible to
identify two dominant perspectivestechnological determinism and social constructivism. The
basic tenet of the technology determinism research
is that technology shapes cultural values, social
structure, and knowledge. In technology-related
fields, such as computer science and information
systems, significant attention has been given to
understanding technological developments and
how these changes influence social structures.
The social constructivism perspective, on the
other hand, posits that knowledge and world views
are created through social interaction. Social
constructivism theories have inspired research
on knowledge construction within communities
of practice. Lave and Wenger (1991) assert that
a societys practical knowledge is situated in

User-Centered Design Principles for Online Learning Communities

relations among practitioners, their practice, and

the social organization and political economy of
communities of practice. For this reason, learning should involve such knowledge and practice
(Lave & Wenger, 1991). Between these heuristic
poles there are cross-disciplinary perspectives, of
which it is possible to further discern them into
four subcategories:
1. Applied Technology Perspective: Much of
the work on OLC by computer scientists and
information systems researchers is driven
by a desire to understand and improve computational approaches. Studies in computer
science, information systems, and educational technologies are mainly aimed at understanding technology to develop tools and
systems that support learning environments
(Daniel, Zapata-Rivera, & McCalla, 2003a;
Preece, 2002; Schwier, 2001). Findings have
been utilized for building technologies
that support OLC. For instance, a growing
number of developers and researchers in
industry and universities are investigating
ways to create software packages that add
new functionality to systems supporting
interaction, collaboration, and leaning in
online learning communities (Kim, 2000;
McCalla, 2000; Preece, 2000; Resnick, 2002;
Schraefel, Ho, Milton, & Chignell, 2000).
2. Ethno-Narrative Perspective: Ethnonarrative research is devoted to revealing
personal experiences of being a member
of an OLC. Most studies adopt a narrative
approach, similar to participant observation
inquiry used in anthropology. Researchers in
this tradition have undertaken comparative
analysis of both online learning and temporal
communities (Schwier, 2001). Critics have
disparaged ethno-narrative studies on the
grounds that findings tend to be anecdotal
and lack external validity; their conclusions
are tentative and limited to the groups under
study (cf. Downes, 2001; Rhiengold, 1993,

1999, 2002). Stolterman, Croon, and Argren

(2000) argue that although the generalization and validity of such studies is limited,
understanding personal perceptions of
learning in OLC is essential. It is difficult to
imagine how one can improve the learning
environment of OLC without the subjective
feedback of the learners.
3. Cultural Studies Perspective: Cultural
studies have contributed enormously to
understanding online learning communities.
For instance, research by Brook and Boal
(1995), Dery (1994), and Hershman and
Leason (1996) investigate the relationship
between the virtual and the physical, and
they fall within the context of cultural interpretation research. Approaches employed in
this category include experimental studies,
with an emphasis on cultural events in online
environments. The background disciplines
of this group are diverse, including social
psychology, philosophy, psychology, and
fine arts.
4. Sociotechnical Perspective: The sociotechnical research tradition argues for a balanced approach to integrating cognitive and
technical dimensions of OLC. This approach
emerged from the extension of sociology,
anthropology, and psychology to the study
of HCI. Subsequently this research informed
disciplines, including computer science and
information systems (Heylighten, 1999).
Research in sociotechnical areas addresses
issues such as:
User-Centered Design: Moving the
focus of interest to learners and away
from technology in the design of online
learning (Norman, 1996).
Contextual Enquiry: Understanding
the users context and its potential
influence on the use of technology
(Preece, 2000).
Sociability: Appreciating the importance of community policies for

981

User-Centered Design Principles for Online Learning Communities

interactions, governance, and social

protocols in OLC (Preece, 2000).
Participatory Design: Involving user
participation in the design of OLC
and the effects on learning outcomes
(Mumford, 1987; Nguyen-Ngoc,
Rekik, & Gillet, Chapter XIII, this
volume).
Direct-Manipulation: Creating tools
for users to create their online learning environment and exploring the
effects of functional options such as
menu-driven and graphical interfaces
(Shneiderman, 1998).

Common to this growing body of research

issues is the need for the interplay of human and
technology factors to guide the design, development, deployment, and evaluation of online
learning communities.

Formal and Informal Online Learning

Communities
There are numerous computational tools that support social learning across time and place (Laghos
& Zaphiris, Chapter XI, this volume). New tools
and patterns of communication have enabled
social engagement, information, and knowledge
sharing within social systems now referred to as
OLC. Unlike a temporal community that resides
in a fixed locale and whose members often know
each other well enough to carry effective interactions, OLCs exist in cyberspace and may or may
not be aware of each other (Daniel, Schwier,
& McCalla, 2003). The character of an OLC is
influenced by structural features, which may
include: community size, duration of interaction
and anticipated lifespan, location or distribution
of the community, the homogeneity/heterogeneity of members, and breadth or narrowness of
subject area. Variation of these features gives
rise to diverse OLCs.

982

In Table 1, we simplify this diversity by distinguishing between formal and informal online
learning communities.
Formal online learning
communities have explicit learning goals and
evaluation criteria. Examples would include
courses/programs offered by education institutions or companies (McCalla, 2000; Schwier,
2001). In contrast, informal OLCs achieve learning outcomes through social learning. Examples
would include distributed communities of practice
(Daniel, OBrien, & Sarkar 2004b). A unique
feature of DCoPs is the absence of a teacher or
instructor; rather, in a DCoP, the learners are also
teachers, as members collectively determine the
content and support each other throughout the
learning process. Further differences are contrasted in Table 1.
A growing body of research identifies the
contribution of DCoPs to facilitating information
exchange and knowledge creation, thereby enriching the work of the collective (Brown & Duguid,
1991; Hildreth, Kimble, & Wright, 1998; Lesser
& Prusak, 2000). These positive outcomes have
caught the interest of scholars and knowledge
managers. And yet, there is little comparative
research on the correlates of DCoP performance
or sustainability. We find this surprising, given the
fact that OLCs emerged and proliferated with the
advent of the Internet and then World Wide Web
over a decade ago. The case-study foundations
for comparative research are certainly present,
however (Kalaitzakis, Dafoulas, & Macaulay,
2003; Hartnell-Young, McGuinness, & Cuttance,
Chapter XII, this volume).
Germane to the topic of DCoP emergence
and sustainability is the question of constructability. Can the DCoP features listed in Table
1 be built, or have DCoPs simply migrated from
the temporal to the online world? If we return to
the literature review briefly touched on earlier,
perhaps not surprisingly we would find a different answer to this question depending on the
literature consulted. For example, the sociology

User-Centered Design Principles for Online Learning Communities

Table 1. Features of online learning communities and distributed communities of practice (adapted from
Daniel et al., 2003b)
Formal: Online Learning Communities
(OLCs)
Membership is explicit and identities are
generally known
Participation is often required
High degree of individual awareness (who
is who, who is where)
Explicit set of social protocols for
interaction
Formal learning goals
Possibly diverse backgrounds
Low shared understanding of domain
Loose sense of identity
Strict distribution of responsibilities
Easily disbanded once established
Low level of trust
Lifespan determined by extent in which
goals are achieved
Pre-planned enterprise and fixed goals

and cultural studies literature tends to be skeptical

of the view that DCoPs can be constructed (Kollock & Smith, 1996). By contrast, the computer
science and information systems research, on the
whole, seem more optimistic that robust DCoPs
can be constructed (Preece, 2000; Daniel et al.,
2003b; McCalla, 2000).
Further, informed by user-centered design
principles, Preece formulated the communitycentered development (CCD) framework to guide
practitioners in the field (Preece, 2000). CCD
provides a blueprint for building a DCoP. The
framework encourages designers to: (1) assess
members interests, (2) identify community norms
and appropriate technology, (3) involve stakeholders in prototype design and testing, (4) correct for
poor usability, and (5) foster community network

Informal: Distributed Communities of Practice

(DCoPs)
Membership may or may not be made explicit
Participation is mainly voluntary
Low degree of individual awareness
Implicit and implied set of social protocols for
interactions
Informal learning goals
Common subject matter
High shared understanding of domain
Strong sense of identity
No formal distribution of responsibilities
Less easily disbanded once established
Reasonable level of trust
Lifespan determined by the instrumental/expressive
value the community provides to its members
A joint enterprise as understood and continually
renegotiated by its members

building and identity. Literature informed by

this approach draws attention to the interaction
between human and technology dimensions in
setting the context for the development and sustainability of DCoPs.
CCD integrates a sociotechnical perspective and pays attention to HCI. On the human
dimension side, attention has been drawn to understanding participants goals, motivations, and
perceptions of the learning environment (Daniel
et al., 2003b); trust (Preece, 2002); and culture
and learning needs (Daniel et al., 2004a). On
the technology side, issues include privacy and
security, usability, scalability, and authenticity
(Daniel et al., 2003a; Preece, 2000).
The attention paid by a sociotechnical approach
to HCI makes this framework particularly well

983

User-Centered Design Principles for Online Learning Communities

suited to understanding the development and sustainability of DCoPs. In particular, the relevance
of a sociotechnical approach to the evolution of
the GKN project results from the attention to,
and monitoring of, feedback loops to inform
design and subsequent operation. For example, a
sociotechnical approach cautions against a build
it and wait till they come approach, and favors
a co-design process that enables potential users
to define their goals and areas of concerns. Joint
construction can be regarded as fostering a shared
identity and building networks necessary for the
development of trust and effective ICT-mediated
interaction.

Our Current Research

The GKN project was launched to address a
perceived need to span geography and cross-organizational boundaries to enhance the scholarship on, and the practice of, governance and its
role in advancing international development. The
underlying challenge of praxis is not unique to this
particular subject area. A consultation document
issued by the Social Science and Humanities Research Council of Canada, for example, re-stated
the networking challenge for advancing collaboration and innovation in the humanities and the
social sciences in the following terms:
Canada is a will against geography. It has a
relatively small population, mostly scattered
across more than 5,000 kilometres. It has no centres equivalent to Paris or London that naturally
draw the best minds and greatest talentsto meet
and interact on a regular basis. It does not have
the numerous institutionsthe Americans have to
move people and ideas around. The net resultis
that it is hard for people to know each other well,
to trust each other and to work together over time
and distance. (SSHRC, 2004)

984

With the emergence of ICTs, these obstacles

to the exchange of information and collaboration
were no longer permanent fixtures, though they
have tended to endure.

Research Approach to the Design of

User-Centered Online Learning
Communities
We began our effort to overcome these obstacles
through a participatory design approach (PDA).
Key to PDA is an iterative process that seeks to
address users needs and promotes their involvement in project development (Schuler & Namioka,
1993). A PDA, also known as a cooperative design approach, shares numerous similarities with
Preeces (2000) community-centered approach.
The first step identified potential technologies
capable of spanning geography and nurturing
collaboration in a DCoP. Working on the human
dimension, the project team created a profile of key
stakeholders of 200 individuals from academia,
government, and the non- and for-profit sectors.
This list represented our target population for the
survey of potential users views on knowledge
sharing in the field and interest in participating
in the development of a DCoP.
The users assessment was divided into three
sections:

an assessment of existing communication/

networking mechanisms among potential
community users,
an assessment of the level of awareness of
work undertaken by users and their affiliated
organizations, and
users perceived value of a DCoP and what
services would contribute to its potential
value.

The goal of the users assessment was to

identify a target groups interests, perceived

User-Centered Design Principles for Online Learning Communities

knowledge gaps, thematic content, and potential

design models for the proposed GKN portal.
Following the analysis of the assessment, we
identified design features that matched identified
services together with appropriate technological
requirements. We further contacted those who
had completed the survey by telephone for a follow-up interview. The goal of the interview was
to elicit further information regarding individuals
preferences for content and portal design. These
steps also served the equally important objective
of engaging potential community participants. In
addition, we were able to gauge the reaction to
the objectives of the GKN project and method of
development and implementation. In addition, the
telephone follow-up was an opportunity to initiate
informal connections among various individuals
working in the same area of research.

Results and Discussion

The target population for the survey was close to
200 organizations identified as working in the field
of international development and governance. The
response rate to the survey was 25%. Of those
responding, 38% were university based, 23%
were from provincial and federal government institutions, 30% were from non-governmental and
research organizations, and 9% were from private
consulting firms. The respondents were distributed across Canada: 45% from western Canada,
53% from central Canada, and only 2% from the
eastern part of the country. These figures reflect the
geographical and sectoral diversity of our sample.
Four out of five respondents were interested in
applied research and technical assistance in this
area, and a similar proportion were interested in
influencing, contributing, or participating in the
policy-making process. In addition, over 80% of
respondents revealed that it is important for them
to keep current on new developments in research
and practice. Depending on their organizational

affiliation, 50% to 80% of the respondents were

interested in building collaborative partnerships
for research and technical assistance
We also asked respondents what kind of
research (applied vs. basic research) they were
interested in, and if they were willing to share
a range of potential outputs with potential GKN
members. The majority (90%) responded that they
were interested in applied research. They were
also willing to contribute to, and participate in,
policymaking processes. Participants identified
the potential for the GKN to support their interest
in keeping abreast of current research and practice
in their fields. In terms of collaboration, a large
number of the respondents viewed the GKN as a
potential mechanism to facilitate information exchange and knowledge sharing among members.
These findings were encouraging for, as Lave
and Wenger (1991) suggest, CoP development
when individuals realize the potential to benefit
by sharing knowledge, insights, and experiences
with each other, and enhance their practices and
performances.
Survey data and follow-up interviews revealed
low levels of awareness of contemporary research
and practice in the field. At the same time informants commented on the specialized nature of
their work and the limited number of organizations
active in the field, they also reported that they were
largely unaware of contemporary contributions
to knowledge and action that their counterparts
have made. Though establishing a benchmark of
awareness is problematic, our results indicated a
considerable lack of awareness among researchers
and practitioners working on governance and international development in Canada. The majority
of the participants described current knowledge
on governance and development as fragmented,
and said that there was a serious lack of awareness
among people working on similar issues across
provinces and between organizations. Similarly,
it was observed that a considerable amount of
publicly funded research, reports, and policy

985

User-Centered Design Principles for Online Learning Communities

documents are not exchanged in a systematic

manner. Respondents identified the potential of
a GKN initiative to facilitate relations among
public, private, non-governmental organizations
and academia.
Though overall results revealed that information sharing and knowledge awareness were
fragmented, there was a pattern to the responses.
First, organizations within a sector were more
knowledgeable of current work undertaken by
their counterparts within the same sector than
organizations in different sectors. Second, there
were marked differences in the level of awareness
among counterparts within provinces compared
to those operating outside their provinces. Although there was a high utilization of information and communication technologies as means
to exchange information and data, they were not
used systematically to break down the information barriers across organizations and across
geographic jurisdictions.
Consistent with previous findings (Wenger,
McDermott, & Snyder, 2000), geographic distance is considered an obstacle to knowledge
sharing and utilization, even by those who are
active users of ICTs. Moving from geographic
to language barriers, several respondents underscored the importance of Canadas two official
languages as a potential barrier. Language is
critical to any community, since it is deemed as
a part of a community identity: identity fosters
collaboration and shared understanding within a
community (McCalla, 2000). Turning to services,
the following list identifies the top four stakeholder
recommendations:

986

Design a DCoP to facilitate information

exchange and knowledge sharing.
Provide a platform for sharing lessons,
experiences, and best practices.
Identify and nurture collaboration among
government, research community, academia, NGOs, and development practitioners.

Build linkages and partnerships with other

international research communities to advance policy and practice.

Following the analysis of the data and feedback

to respondents, we identified and profiled different technologies capable of supporting a DCoP
that would perform to stakeholder expectations.
Once the technological elements were identified,
feedback was sought again from participants on
the relevance of these models. This feedback was
integrated in the prototype development of the
GKN portal, which is currently in its formative
stages. As the GKN project moved from a needs
assessment to co-development with interested
partners, human and technology interaction issues
are gaining more importance.
At present, the GKN team has implemented a
beta version of the system, while at the same time
pursuing research into social and technical means
to nurture and support an evolving community.
Currently, we are experimenting with the use
of blended strategies of face-to-face workshops
and videoconferencing as additional avenues to
encourage integration of human and technology
factors. We are also developing an evaluation plan
to assess the importance of the factors identified
earlier to developing and sustaining the GKN
project. In the following section, we describe
the dimensions of HCI that have the potential to
affect the viability and robustness of the GKN
project.

Emergent Human and

Technology Issues
There are multiple factors affecting the emergence
and sustainability of a DCoP. Drawing from the
GKN experience and insights from the sociotechnical approach outlined previously, we maintain
that the following set of factors are important to
HCI. Their influence and relative importance to

User-Centered Design Principles for Online Learning Communities

the emergence and sustainability of a DCoP is

introduced briefly in the following:

Didactics: Learning is a shared experience,

and by extension DCoPs are learning communities. Some OLCs have explicit learning
goals (e.g., formal OLCs created around
WebCT courses), while others have more
implicit goals of sharing ideas, practices, and
knowledge (e.g., DCoPs among corporateoriented/professional communities). The
technology must therefore enable learning,
and perceptions of learning feedback would
likely affect participation.
Trust: Stakeholder surveys revealed that a
key attraction of the proposed GKN online
community would be the ability to share and
retrieve archived data that was not widely
available. The creation of this shared resource would depend on the willingness of
these stakeholders to contribute their data.
Their decision to share data would likely
be influenced by their trust in others in the
community as well as the environment in
which they interact, for instance, questions
such as: How would community members
treat my data? Would my research be reproduced without my permission or quoted out
of context? Creating generalized trust within
a DCoP is difficult to engineer, but likely
a pre-requisite condition for the sharing and
accumulation of data.
Privacy and Security: Privacy and security
tools address individual perceptions of safety
in the community. In an environment where
a person feels their privacy threatened, declining participation is anticipated. In this
regard, computational tools that protect the
privacy and security of individuals must be
provided.
Scalability and Authenticity: Scalability
expresses the ability of a system to accommodate multiple users, and authenticity

refers to the ability of a system to protect

individuals in a community from outsiders.
A DCoP must encourage entrants and their
participation. This dimension is critical to
the growth of the DCoP, whereas authenticity
appears more important to sustainability. For
example, an open system that does not protect users (e.g., from hackers) is susceptible
to negative feedback and eventual decline
of member participation.
Sociability: Sociability relates to the protocols in use for communication and interaction in the community (Preece, 2000).
These protocols may be imposed in the first
instance, but will likely shift in response to
community dynamics. Sociability is of particular importance to constructed online
communities that do not inherit protocols
in use, as would temporal communities that
have migrated to an ICT-mediated environment. This dimension is likely critical to the
sustainability of a DCoP, as protocols in use
will need to reflect members preferences and
practices. As new protocols emerge, technology must accommodate such changes.
Usability: Our research indicated that
interest in the GKN initiative centered on
the promise of instrumental outcomes (e.g.,
access to information, new insights, and
expanded contacts). Here, technology and
human interaction are clearly linked, as
relevant content is dependent on member
input and its ease of retrieval is dependent on
technology. User-centered interface design
and continuous involvement of users are
critical to both the emergence and sustainability of the GKN project.
Culture: An explicit objective of the GKN
project was to bridge organizational and
linguistic boundaries. As organizational
theory suggests that organizations inculcate
and perpetuate cultures that may promote
or discourage inter-organizational infor-

987

User-Centered Design Principles for Online Learning Communities

mation sharing and/or collaboration. Once

organizational or individual participation is
present (a human, not a technical issue), we
are uncertain of how technology may shape
or accommodate different culture(s). Though
others suggest that the viability of DCoPs
depends on the development of a shared
culture, our project is not sufficiently far
advanced to comment on this hypothesis.
Awareness: The ability of ICT tools to
provide awareness among its members is
predicted to have a powerful impact on
members interactions in the community.
More specifically, awareness (e.g., awareness about who is who, and who does and
knows what) can have a significant positive
feedback that would in turn promote participation and contribute to sustainability.

These elements highlighted exert different

forces on technology and human interaction. For
reasons stated, we anticipate that each will have
a bearing on the emergence and sustainability of
the GKN initiative and DCoP more generally.

factors affecting the emergence and sustainability

of a DCoP.
By way of conclusion, we offer the following general UCD principles for designing and
sustaining online learning communities based
on the sociotechnical approach.

Design Principles

Discussion
The sociotechnical approach to the development
of a DCoP suggests that human and technical
factors are interlinked and they co-determine the
emergence, evolution, growth, and sustainability
of DCoPs. For practitioners involved in designing
or developing a DCoP, the variables outlined previously will likely provide a useful starting point
for guiding implementation and identifying key
relationships. For researchers, our preliminary
exploration of these relationships creates a number
of hypotheses for future investigation. As these
relationships have a bearing on both practice and
research, we intend to track these relationships
through user evaluations and internal monitoring. We anticipate that these findings will work
toward a framework for comparative research on

988

Assessing needs of actual or potential users/learners.

Identifying the gap between what is and
what needs to be.
Understanding users and usage contexts.
Profiling learning styles.
Benchmarking existing community models.
Identifying existing technological tools.
Maintaining an iterative design and development processes that keep users/learners
informed.
Providing appropriate tools to support and
mediate learning, social interaction and
facilitate a sense of togetherness.
Exploring navigation tools to enable members to gather information about others
and have access to community interactions
traces of activities.

Didactic Principles

Nurturing open and informal discourse

as members interact to satisfy their own
personal and community learning needs.
Encouraging learners to become active users
and contributors of content.
Supporting different learning styles.
Encouraging participation and discourse
around central themes, ideas, or purposes.
Guiding participants throughout the interaction process, and providing them with clear
directions to attainment of learning goals.
Understanding unique individual learning
needs differences, and encouraging partici-

User-Centered Design Principles for Online Learning Communities

pants to construct their own meaning based

on unique individual experiences.

Sociability Principles

Establishing a clear set of social protocols

for interactions.
Encouraging informal interaction and an
environment conducive to learner/user interaction so that members have opportunities
to test the trustworthiness of others.
Supporting shared objectiveswhich creates a rationale for belonging to the community.
Maintaining relevant content and context
for interaction throughout the lifespan of
the community.
Encouraging ongoing active dialogue among
members.
Maintaining different forms of awareness
(who is who, who knows what, who knows
who knows what, etc.) in the community to
lubricate effective interaction.

Acknowledgments
The research reported in this chapter has been
supported financially by the Policy Branch of
the Canadian International Development Agency
(CIDA), the Social Sciences and Humanities
Research Council of Canada (SSHRC), and the
International Center for Governance and Development at the University of Saskatchewan.

References
Brook, J., & Boal, I. A. (Eds.). (1995). Resisting
the virtual life: The culture and politics of information. San Francisco: City Lights Books.
Brown, J. S., & Duguid, P. (1991). Organizational
learning and communities of practice: Towards a

unified view of working, learning and innovating.

Organization Science, 2, 40-57.
Cothrel, J. (1999). Virtual community today. The
Journal of AGSI, 7, 52-55.
Daniel, B. K., OBrien, D., & Sarkar, A. (2004b,
July 21-25). The cybernetics of distributed communities of practice. Proceedings of the Joint
International Conference on Education and
Information Systems, Technologies and Applications (EISTA 2004) and the International
Conference on Cybernetics and Information
Technologies, Systems and Applications (CITSA
2004), Orlando, FL.
Daniel, B. K., Poon, N., & Sarkar, A. (2005, June
27-July 2). Analysis of patterns of interactions
in video-chat supported virtual communities to
model social capital. Proceedings of the World
Conference on Educational Multimedia, Hypermedia and Telecommunications (ED-MEDIA
2005), Montreal, Canada.
Daniel, B. K., Sarkar, A., & OBrien, D. (2004a,
June 21-26). A participatory design approach for a
distributed community of practice on governance
and international development. Proceedings of the
World Conference on Educational Multimedia,
Hypermedia and Educational Telecommunications (pp. 4606-4613), Lugano, Switzerland.
Daniel, B. K., Schwier, R. A., & McCalla, G. I.
(2003b). Social capital in virtual learning communities and distributed communities of practice.
Canadian Journal of Learning and Technology,
29(3), 113-139.
Daniel, B. K., Zapata-Rivera, D. J., & McCalla,
G. I. (2003a). A Bayesian computational model
of social capital. In M. Huysman, E. Wenger, &
V. Wulf (Eds.), Communities and technologies.
London: Kluwer.
Downes, S. (2001). Learning communities. Retrieved May 1, 2003, from http//www.atl.ualberta.
ca/downes/future/communities.htm

989

User-Centered Design Principles for Online Learning Communities

Gutwin, C., & Greenberg, S. (1998). Design for

individuals, design for groups: Tradeoffs between
power and workspace awareness. Proceedings of
the ACM Conference on Computer-Supported
Cooperative Work (pp. 207-216).

Lesser, E. L., & Prusak, L. (2000). Communities of practice, social capital and organizational
knowledge. In E. L. Lesser, M. A. Fontaine, & J.
A. Slusher (Eds.), Knowledge and communities.
Boston: Butterworth Heinemann.

Heylighten, F. (1999). What are socio-technical approaches and systems science? Principia
socio-technical approach Web. Retrieved from
http://pespmc1.vub.ac.be/CYBSWHAT.html

Lesser, E., & Storck, J. (2001). Communities of

practice and organizational performance. IBM
Systems Journal, 40(4).

Hildreth, P., Kimble, C., & Wright, P. (1998).

Computer-mediated communications and international communities of practice. Proceedings
of the 4th International Conference on Ethical
Issues of Information Technology (Ethicomp98),
Rotterdam, Holland.
Kalaitzakis, E., Dafoulas, G., & Macaulay, L.
(2003, June 22-27). Designing online communities: Community-centered development for
intensively focused user groups. Proceedings of
the 10th HCI International Conference, Greece.

McCalla, G. (2000). The fragmentation of culture,

learning, teaching and technology: Implications
for artificial intelligence in education research.
International Journal of Artificial Intelligence,
11(2), 177-196.
Mumford, E. (1987). Socio-technical systems
design: Evolving theory and practice. In P. Ehn
& M. Kyng (Eds.), Computers and democracy:
A Scandinavian challenge (pp. 59-76). London:
Avebury.

Kim, A. J. (2000). Community building on the

Web: Secret strategies for successful online communities. Berkeley, CA: Peachpit Press.

Norman, D.A. (1996). Cognitive engineering. In D.

A. Norman & S. W. Draper (Eds.), User-centered
systems designnew perspectives on humancomputer interaction (pp. 31-61). Hillsdale, NJ:
Lawrence Erlbaum.

Kim, E.E. (2003). An introduction to open source

communities. Retrieved May 1, 2003, from http://
www.blueoxen.org/research/00007/

Preece, J. (2000). Online communities: Designing

usability and supporting sociability. New York:
John Wiley & Sons.

Kollock, P., & Smith, M. (1996). Managing the

virtual commons: Cooperation and conflict in
computer communities. In S. Herring (Ed.),
Computer-mediated communication: Linguistic,
social, and cross-cultural perspectives. Amsterdam: John Benjamins.

Preece, J. (2002). Supporting community and

building social capital. Communications of the
ACM, 45(4), 37-39.

Krippendorf, K. (1984). Epistemological foundation for communication. Journal of Communication, 84.

Lave, J., & Wenger, E. (1991). Situated learning:
Legitimate peripheral participation. New York:
Cambridge University Press.

990

Resnick, P. (2002). Beyond bowling together:

Socio-technical capital. In J.M. Carroll (Ed.), HCI
in the new millennium (pp. 247-272). New York:
Addison-Wesley.
Rheingold, H. (1993). The virtual community:
Homesteading on the virtual frontier. New York:
Addison-Wesley.
Riedl, M. O. (2001, January 14-17). A computational model and classification framework for

User-Centered Design Principles for Online Learning Communities

social navigation. Proceedings of the International

Conference on Intelligent User Interfaces 2001
(pp. 137-144), Santa Fe, NM.

SSHRC. (2004). From a granting to a knowledge

council. Ottawa: Social Sciences and Humanities
Research Council of Canada.

Schraefel, M. C., Ho, J., Chignell, M., & Milton,

M. (2000). Building virtual communities for research. Proceedings of the International Working
Conference and Industrial Expo on New Advances
and Emerging Trends in Next Generation Enterprises, Buffalo, NY.

Stolterman, E., Croon, A., & Argren, P.-O. (2000).

Virtual communitiesWhy and how they are
studied. Working paper, Department of Informatics, Ume University, Sweden.

Schuler, D., & Namioka, A. (Eds.). (1993). Participatory design: Principles and practices. Hillsdale,
NJ: Lawrence Erlbaum.
Schwier, R. A. (2001). Catalysts, emphases, and
elements of virtual learning communities. Implication for research. The Quarterly Review of
Distance Education, 2(1), 5-18.
Sclove, R. E. (1995). Democracy and technology.
New York: The Guildford Press.
Shneiderman, B. (1998). Designing the user interface. Strategies for effective human-computer
interaction. Boston: Addison-Wesley.
Smith, M. (1992). Voices from the WELL: The
logic of the virtual commons. Masters thesis,
Department of Sociology, UCLA, USA.

Wenger, E., McDermott, R., & Snyder, W. M.

(2002). Cultivating communities of practice: A
guide to managing knowledge. Boston: Harvard
Business School Press.
Wickre, K. (1995). Virtual communities: Are
they real? Are they real enough? Retrieved May
1, 2003, from http://www.thenet-usa.com/mag/
back/0995/community.html

Endnote
1

Presenters at the 1st GKN Workshop on Building Distributed Communities of Practice

for Enhanced Research-Policy Interface,
May 28-31, 2004. Presentations available
at http://www.icgd.usask.ca/workshopPlan.
html

This work was previously published in User-Centered Design of Online Learning Communities, edited by N. Lambropoulos
and P. Zaphiris, pp. 54-70, copyright 2007 by Information Science Publishing (an imprint of IGI Global).

991

992

Chapter 2.35

Content Personalization for

Mobile Interfaces
Spiridoula Koukia
University of Patras, Greece
Maria Rigou
University of Patras, Greece, & Research Academic Computer Technology Institute, Greece
Spiros Sirmakessis
Technological Institution of Messolongi &
Research Academic Computer Technology Institute, Greece

INTRODUCTION
The contribution of context information to content
management is of great importance. The increase
of storage capacity in mobile devices gives users
the possibility to maintain large amounts of content
to their phones. As a result, this amount of content
is increasing at a high rate. Users are able to store
a huge variety of content such as contacts, text
messages, ring tones, logos, calendar events, and
textual notes. Furthermore, the development of
novel applications has created new types of content,
which include images, videos, MMS (multi-media
messaging), e-mail, music, play lists, audio clips,
bookmarks, news and weather, chat, niche information services, travel and entertainment information,

driving instructions, banking, and shopping (Schilit

& Theimer, 1994; Schilit, Adams, & Want, 1994;
Brown, 1996; Brown, Bovey, & Chen, 1997).
The fact that users should be able to store the
content on their mobile phone and find the content
they need without much effort results in the requirement of managing the content by organizing and
annotating it. The purpose of information management is to aid users by offering a safe and easy way
of retrieving the relevant content automatically, to
minimize their effort and maximize their benefit
(Sorvari et al., 2004).
The increasing amount of stored content in mobile devices and the limitations of physical mobile
phone user interfaces introduce a usability challenge in content management. The physical mobile

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Content Personalization for Mobile Interfaces

phone user interface will not change considerably.

The physical display sizes will not increase since
in the mobile devices the display already covers
a large part of the surface area. Text input speed
will not change much, as keyboard-based text
input methods have been the most efficient way
to reduce slowness. While information is necessary for many applications, the human brain is
limited in terms of how much information it can
process at one time. The problem of information
management is more complex in mobile environments (Campbell & Tarasewich, 2004).
One way to reduce information overload and
enhance content management is through the
use of context metadata. Context metadata is
information that describes the context in which
a content item was created or received and can
be used to aid users in searching, retrieving, and
organizing the relevant content automatically.
Context is any information that can be used to
characterize the situation of an entity. An entity
is a person, place, or object that is considered
relevant to the interaction between a user and an
application, including the user and the applications
themselves (Dey, 2001). Some types of context
are the physical context, such as time, location,
and date; the social context, such as social group,
friends, work, and home; and the mental context,
which includes users activities and feelings
(Ryan, Pascoe, & Morse, 1997; Dey, Abowd, &
Wood, 1998; Lucas, 2001).
By organizing and annotating the content, we
develop a new way of managing it, while content
management features are created to face efficiently
the usability challenge. Context metadata helps
the user find the content he needs by enabling
single and multi-criteria searches (e.g., find photos
taken in Paris last year), example-based searches
(e.g., find all the video clips recorded in the same
location as the selected video clip), and automatic
content organization for efficient browsing (e.g.,
location-based content view, where the content
is arranged hierarchically based on the content

capture location and information about the hierarchical relationships of different locations).

DATE, TIME, LOCATION,

AND PROXIMITY
While context can be characterized by a large
number of different types of attributes, the contribution of context attributes to content management is of great importance. We focus on a small
number of attributes, which are considered the
most important in supporting content management
and also have the most practical implementations
in real products, such as date, time, location, and
proximity (nearby Bluetooth devices). Bluetooth
is a short-range wireless technology used to create personal area networks among user mobile
devices and with other nearby devices.
The first two attributes, date and time, are the
most common in use in a wide range of applications. They are used to organize both digital and
analog content, and offer an easy way of searching
and retrieving the relevant content automatically.
For example, many cameras automatically add
the date and time to photographs. Furthermore,
the location where content is created is another
useful attribute for searching the content (e.g.,
home, workplace, summer cottage). Mobile devices give users the possibility to create content
in many different locations. Users can associate
the location with the equivalent content in order
to add an attribute to it that will enable them to
find it easier. Finally, proximity also plays an
important role in content management, as nearby
Bluetooth devices can provide information both in
social and physical context. While each Bluetooth
device can be uniquely identified, information
can be provided on nearby people by identifying
their mobile phones. An example for physical
context is the case of a Bluetooth-based handsfree car kit that can be used to identify that the
user is in a car.

993

Content Personalization for Mobile Interfaces

USABILITY ISSUES AND

PROBLEMS
The expansion of the dimension of context information in order to include location, as well as
proximity context, can be of benefit to users while
they are able to store, access, and share with others their own location-based information such as
videos and photos, and feel the sense of community
growing among them (Kasinen, 2003; Cheverist,
Smith, Mitchell, Friday, & Davies, 2001). But when
it comes to proximity to be included in context
information, the problem of privacy emerges. It
appears that users are willing to accept a loss of
privacy when they take into account the benefits
of receiving useful information, but they would
like to control the release of private information (Ljungstrand, 2001; Ackerman, Darrel, &
Weitzner, 2001).
While context metadata is attached to content,
when users share content, they have to decide if
they share all the metadata with the content or
they filter out all or some part of them. The cost
for memory and transmission of metadata, as it
is textual information, is not an important factor
to influence this decision. When the user receives
location and proximity information attached to
content, he or she may also find out where and
with whom the creator of the content was when
the content was created. As a result, both the
location of the content creator and the location of
nearby people are shared along with the content
information. If this information is private, the
sharing of it could be considered as a privacy
violation. This violation may be multiplied if
the first recipient forwards the content and the
metadata to other users.
However, users seem to be willing to share
context metadata attached to content, as it would be
convenient if context metadata were automatically
available with the content (so that users do not
have to add this information manually). Furthermore, it would be very helpful for the recipient if

994

the received content was annotated with context

metadata so that the recipient does not have to
annotate it manually and be able to manage the
content more easily. For example, in the case of
image and video content, the filtering of context
metadata such as location and people could be
useless, since these same items appearing in the
image or video can be identified visually from
the image content itself.
But what is meaningful information to the
end user? It seems that users want meaningful
information, but they are not willing to put too
much effort in creating it, unless this information is expected to be very useful. In the case of
location, it would be difficult for users to type
the name of the place and other attributes manually, since it would require their time and effort.
Thus it would be important if meaningful context
metadata, which include the required information,
are automatically generated.
Proximity information also needs to be meaningful. In this way, meaningfulness is important
when attaching information on nearby devices
in the form of metadata. If the globally unique
Bluetooth device address and the real name of
the owner of the device could be connected, this
functionality would give meaningful information
to the user.
It is hard to determine which information is
useful, while what is useful information in one
situation might be totally useless in another. For
example, when looking at photo albums, what
is thought to be useful information varies a lot.
When one is looking at family pictures taken
recently, it is needless to write down the names
of the people, since they were well known and
discernable. But it is different looking at family
pictures taken many years ago: the same people
may not be that easily recognizable.
It appears that useful information depends on
a users location, what the information is used
for, and in which time span. In order to create
meaningful information, users need to put much

Content Personalization for Mobile Interfaces

effort into getting the data, organizing it, and

annotating it with context metadata. Ways to
minimize their effort and maximize their benefit
should be developed.

CONCLUSION
The increasing amount of stored content in
mobile devices and the limitations of physical
mobile phone user interfaces introduce a usability
challenge in content management. The efficient
management of large amounts of data requires
developing new ways of managing content. Stored
data are used by applications which should express
information in a sensible way, and offer users a
simple and intuitive way of organizing, searching, and grouping this information. Inadequate
design of user interface results in poor usability
and makes an otherwise good application useless. Therefore, it is necessary to design and built
context-aware applications.
Issues of usefulness and meaningfulness in
utilizing context metadata need to be further
investigated. Usefulness depends on the type of
metadata. As far as location and proximity are
concerned, it appears that the more time has passed
since the recording of the data, the more accurate
the information needs to be. Furthermore, in the
case of location information, the closer to ones
home or familiar places the data refers to, the
more detailed the information needs to be. A main
usability challenge is the creation of meaningful
context metadata automatically, without users
having to add this information manually. There
exist many ways for automatic recording of information about a users context, but the generated
information is not always meaningful.
Another field that requires further research is
privacy. It seems that users are willing to accept
a loss of privacy, provided that the information
they receive is useful and they have control over
the release of private information. Content management provides users with a safe, easy-to-use,

and automated way of organizing and managing

their mobile content, as well as retrieving useful
information efficiently.

REFERENCES
Ackerman, M., Darrel, T., & Weitzner, D. J. (2001).
Privacy in context. Human Computer Interaction,
16, 167-176.
Brown, P. J. (1996). The stick-e document: A
framework for creating context-aware applications.
IFIP Proceedings of Electronic Publishing 96,
Laxenburg, Austria, (pp. 259-272).
Brown, P. J., Bovey, J. D., & Chen, X. (1997).
Context-aware applications: From the laboratory
to the marketplace. IEEE Personal Communications, 4(5), 58-64.
Campbell, C., & Tarasewich, P. (2004). What can
you say with only three pixels? Proceedings of
the 6th International Symposium on Mobile Human-Computer Interaction, Glasgow, Scotland,
(pp. 1-12).
Cheverist, K., Smith, G., Mitchell, K., Friday, A.,
& Davies, N. (2001). The role of shared context
in supporting cooperation between city visitors.
Computers &Graphics, 25, 555-562.
Dey, A. K., Abowd, G. D., & Wood, A. (1998).
CyberDesk: A framework for providing selfintegrating context-aware services. Knowledge Based
Systems, 11(1), 3-13.
Dey, A. K. (2001). Understanding and using
context. Personal & Ubiquitous Computing,
5(1), 4-7.
Kaasinen, E. (2003). User needs for location-aware
mobile services. Personal Ubiquitous Computing, 7, 70-79.
Kim, H., Kim, J., Lee, Y., Chae, M., & Choi, Y.
(2002). An empirical study of the use contexts and
usability problems in mobile Internet. Proceed-

995

Content Personalization for Mobile Interfaces

ings of the 35th Annual International Conference

on System Sciences (pp. 1767-1776).

man-Computer Interaction, Tampere, Finland,

(pp. 357-363).

Ljungstrand, P. (2001). Context-awareness and

mobile phones. Personal and Ubiquitous Computing, 5, 58-61.

KEY TERMS

Lucas, P. (2001). Mobile devices and mobile

dataissues of identity and reference. HumanComputer Interaction, 16(2), 323-336.
Ryan, N., Pascoe, J., & Morse, D. (1997). Enhanced
reality fieldwork: The context-aware archaeological assistant. In V. Gaffney, M. v. Leusen,
& S. Exxon (Eds.), Computer applications in
archaeology.
Schilit, B., & Theimer, M. (1994). Disseminating
active map information to mobile hosts. IEEE
Network, 8(5), 22-32.
Schilit, B., Adams, N., & Want, R. (1994).
Context-aware computing applications. IEEE
Proceedings of the 1st International Workshop
on Mobile Computing Systems and Applications,
Santa Cruz, CA, (pp. 85-90).
Sorvari, A., Jalkanen, J., Jokela, R., Black, A.,
Kolil, K., Moberg, M., & Keinonen, T. (2004).
Usability issues in utilizing context metadata
in content management of mobile devices. Proceedings of the 3rd Nordic Conference on Hu-

Bluetooth: A short-range wireless technology

used to create personal area networks among user
devices and with other nearby devices.
Content Management: Ways of organizing
and annotating content in order to retrieve and
search it more efficiently.
Context: Any information that can be used to
characterize the situation of an entity.
Context Metadata: Information that describes
the context in which a content item was created
or received.
Entity: A person, place, or object that is
considered relevant to the interaction between a
user and an application, including the user and
the applications themselves.
Location: The place where content is created
by the user.
Usability: The effectiveness, efficiency, and
satisfaction with which users can achieve tasks
in the environment of mobile devices.

This work was previously published in the Encyclopedia of Mobile Computing and Commerce, edited by D. Taniar, pp. 116-118,
copyright 2007 by Information Science Reference, formerly known as Idea Group Reference (an imprint of IGI Global).

996

997

Chapter 2.36

Speech-Centric Multimodal
User Interface Design in
Mobile Technology
Dong Yu
Microsoft Research, USA
Li Deng
Microsoft Research, USA

Abstract

Introduction

Multimodal user interface (MUI) allows users to

interact with a computer system through multiple
human-computer communication channels or
modalities. Users have the freedom to choose
one or more modalities at the same time. MUI
is especially important in mobile devices due
to the limited display and keyboard size. In this
chapter, we provide a survey of the MUI design
in mobile technology with a speech-centric view
based on our research and experience in this area
(e.g., MapPointS and MiPad). In the context of
several carefully chosen case studies, we discuss
the main issues related to the speech-centric MUI
in mobile devices, current solutions, and future
directions.

In recent years, we have seen steady growth in the

adoption of mobile devices in peoples daily lives
as these devices become smaller, cheaper, more
powerful, and more energy-efficient. However,
mobile devices inevitably have a small display
area, a tiny keyboard, a stylus, a low speed (usually less than 400 million instructions per second)
central processing unit (CPU), and a small amount
(usually less than 64MB) of dynamic random-access memory. Added to these limitations is the
fact that mobile devices are often used in many
different environments, such as dark and/or noisy
surroundings, private offices, and meeting rooms.
On these devices, the traditional graphical user
interface (GUI)-centric design becomes far less
effective than desired. More efficient and easy-

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Speech-Centric Multimodal User Interface Design in Mobile Technology

to-use user interfaces are in urgent need. The

multimodal user interface (MUI), which allows
users to interact with a computer system through
multiple channels such as speech, pen, display,
and keyboard, is a promising user interface in
mobile devices.
Multimodal interaction is widely observed
in human-human communications where senses
such as sight, sound, touch, smell, and taste are
used. The research on multimodal human-computer interaction, however, became active only
after Bolt (1980) proposed his original concept
of Put That There. Since then, a great amount
of research has been carried out in this area
(Bregler, Manke, Hild, & Waibel 1993; Codella,
Jalili, Koved, Lewis, Ling, Lipscomb, et al., 1992;
Cohen, Dalrymple, Moran, Pereira, Sullivan,
Gargan, et al., 1989; Cohen, Johnston, McGee,
Oviatt, Pittman, Smith, et al., 1997; Deng & Yu,
2005; Fukumoto, Suenga, & Mase, 1994; Hsu,
Mahajan, & Acero 2005; Huang, Acero, Chelba,
Deng, Droppo, Duchene, et al., 2001; Neal &
Shapiro, 1991; Pavlovic, Berry, & Huang, 1997;
Pavlovic & Huang, 1998; Vo, Houghton, Yang,
Bub, Meier, Waibel, et al., 1995; Vo & Wood,
1996; Wang, 1995). Importantly, the body of this
research work pointed out that MUIs can support
flexible, efficient, and powerful human-computer
interaction.
With an MUI, users can communicate with a
system through many different input devices such
as keyboard, stylus, and microphone, and output
devices such as graphical display and speakers.
MUI is superior to any single modality where users can communicate with a system through only
one channel. Note that using an MUI does not
mean users need to communicate with the system
always through multiple communication channels
simultaneously. Instead, it means that users have
freedom to choose one or several modalities when
communicating with the system, and they can
switch modalities at any time without interrupting
the interaction. These characteristics make the
MUI easier to learn and use, and is preferred by

998

users in many applications that we will describe

later in this chapter.
MUI is especially effective and important in
mobile devices for several reasons. First, each
modality has its strengths and weaknesses. For
this reason, single modality does not permit the
user to interact with the system effectively across
all tasks and environments. For example, speech
UI provides a hands-free, eyes-free, and efficient
way for users to input descriptive information or
to issue commands. This is very valuable when in
motion or in natural field settings. Nevertheless,
the performance of speech UI decreases dramatically under noisy conditions. In addition, speech
UI is not suitable when privacy and social condition (e.g., in a meeting) is a concern. Pen input,
on the other hand, allows users to interact with
the system silently, and is acceptable in public
settings and under extreme noise (Gong, 1995;
Holzman, 1999). Pen input is also the preferred
way for entering digits, gestures, abbreviations,
symbols, signatures, and graphic content (Oviatt
& Olsen, 1994; Suhm, 1998). However, it is impossible for the user to use pen input if he/she is
handicapped or under temporary disability (e.g.,
when driving). MUI, on the other hand, allows
users to shift between modalities as environmental
conditions change (Holzman, 1999), and hence,
can cover a wider range of changing environments
than single-modal user interfaces.
Second, different modalities can compensate
for each others limitations and thus provide users with more desirable experience (Deng & Yu,
2005; Oviatt, Bernard, & Levow, 1999; Oviatt &
vanGent, 1996; Suhm, 1998). For example, the
accuracy of a resource-constrained, midsized
vocabulary speech recognizer is low given the
current speech technology. However, if the speech
recognizer is used together with a predictive T9
(text on 9 keys) keyboard, users can greatly increase the text input throughput compared with
using the speech modality or T9 keyboard alone
(Hsu et al., 2005). The gain is obtained from the
mutual disambiguation effect, where each error-

Speech-Centric Multimodal User Interface Design in Mobile Technology

prone modality provides partial information to aid

in the interpretation of other modalities. Another
reason for the improved user experience is users
active error avoidance, where users tend to select
the input modality that they judge to be less error
prone for a particular task and environment (Oviatt
& vanGent, 1996), and tend to switch modalities
to recover from system errors (Oviatt et al., 1999).
Mutual compensation is very important for mobile devices because the ability of every single
modality in the devices is extremely limited (e.g.,
a limited display and keyboard size, and limited
speech recognition accuracy).
Despite the importance of MUI in mobile devices, designing effective MUIs is far from trivial.
Many MUIs in mobile devices are speech centric,
where speech is the central and main modality. In
this chapter, we will focus on main issues on the
design of effective speech centric MUIs in mobile
devices based on our research and experience in
developing MapPointS (Deng & Yu, 2005) and
MiPad (Deng, Wang, Acero, Hon, Droppo, Boulis,
et al., 2002; Huang, Acero, Chelba, Deng, Droppo,
Duchene, et al., 2001). In Section 2, we describe
a generic MUI architecture in mobile setting
that consists of various recognizers for different
input modalities, semantic parsers, a discourse
manager, and a response manager. In Section
3, we discuss special considerations related to
speech modality. In particular, we discuss the
approaches to overcoming resource limitations
on mobile devices, noise robust speech frontends, noise robust modality switching interfaces,
and context-aware language model. In section 4,
we introduce the issues related to robust natural
language understanding including construction
of robust grammars. We discuss the problem
of modality fusion, including modality-neutral
semantic representation, unification approach,
and modality integration, in Section 5. We discuss possible future directions and conclude this
chapter in Section 6.

A generic MUI architecture

The ultimate goal of an MUI is to fulfill the needs
and requirements of the users. This principle is
one of many emphasized in user-centered design (Gould & Lewis, 1985, Norman & Draper,
1986). According to the user-centered design
principle, the acceptability of an MUI can be
judged using three main attributes (Dybkjaer &
Bernsen, 2001; Hone & Graham, 2001; Nielsen,
1993): effectiveness, efficiency, and learnability.
The effectiveness assesses whether users can
complete the tasks and achieve the goals with
the predefined degree of perceived accuracy. It is
usually measured on the targeted user population,
over a specified range of tasks and environments.
The efficiency judges how much effort (cognitive
demand, fatigue, stress, frustration, discomfort,
and so on) and resources (time) are needed for
users to perform specific tasks. It is usually
measured with the total time (including time for
error corrections) taken to complete a task. The
learnability measures whether users can easily
discover the systems functionality and quickly
learn to use the system.
Figure 1 depicts a typical speech-centric MUI
architecture that is aimed to achieve a high level
of effectiveness, efficiency, and learnability. As
shown in the figure, users can communicate with
the system through speech, keyboard, and other
modalities such as pen and camera. Modality fusion usually is the center of an MUI system. There
are two typical ways of fusing information from
different input modalities, namely, early fusion
and late fusion. With the early fusion, signals
are integrated at the feature level and hence, the
recognition process in one modality would affect
that in another modality (Bregler et al., 1993,
Pavlovic et al., 1997; Pavlovic & Huang, 1998; Vo
et al., 1995,). Early fusion is suitable for highly
coupled modalities such as speech and lip movements (Rubin, Vatikiotis-Bateson, & Benoit, 1998;
Stork & Hennecke, 1995). However, early fusion

999

Speech-Centric Multimodal User Interface Design in Mobile Technology

can greatly increase the modeling complexity

and computational intensity due to its nature of
intermodality influence in the recognition phase.
With the late fusion, information is integrated at
the semantic level. The benefit of late fusion is
its isolation of input modalities from the rest of
the system. In other words, individual recognizers trained using unimodal data can be directly
plugged into the system without affecting the rest
of the system. This feature makes the late fusion
easier to scale up to more modalities in the future
than the early fusion. The architecture shown in
Figure 1 utilizes the late fusion approach that has
been widely adopted, for example, by a variety
of systems including Put-That-There (Bolt, 1980),
MapPointS (Deng & Yu, 2005), MiPad (Huang et
al., 2001), ShopTalk (Cohen, et al., 1989), QuickSet
(Cohen, Johnston, McGee, Oviatt, Pittman, Smith,
et al., 1997), CUBRICON (Neal & Shapiro, 1991),
Virtual World (Codella, Jalili, Koved, Lewis, Ling,
Lipscomb, et al., 1992), Finger-Pointer (Fukumoto
et al., 1994), VisualMan (Wang, 1995), and Jeanie
(Vo & Wood, 1996).
In the late-fusion approach depicted in Figure
1, the input signals received by the system are
first processed by semantic parsers associated
with the corresponding modality into the surface
semantics representation. Note that although each
modality has its own semantic parser, the resulting
surface semantics are represented in a common
semantic representation and is thus independent of
the modality. The surface semantics from all the
input modalities are then fused by the discourse
manager component into the discourse semantics
representation (more discussions on this issue in
Section 4). In order to generate discourse semantics, the discourse manager uses the semantic
modal and interacts with the context manager
to utilize and update such information as dialog
context, domain knowledge, users information,
and users usage history. The updated context
information can be used to adapt the language
model, which can improve speech recognition

1000

accuracy and enhance the quality of semantic

parsers for the next user-computer interaction.
The discourse semantics, which is the output
of the discourse manager, is then fed into the
response manager to communicate back to the
user. The response manager synthesizes the proper
responses, based on the discourse semantics and
the capabilities of the user interface, and plays
the response back to the user. In this process,
behavior model provides rules to carry out the
required actions. The combination of discourse
manager and response manager is usually referred
to as the dialog manager.
Note that the components shown in Figure 1
may reside on the mobile devices, or distributed
on other servers in real implementations. In addition, many MUI systems use an agent-based
software solution in which a facility or hub is

Figure 1. A typical speech-centric MUI architecture and its components

Recognizer +
speech Input

Semantic Parser

(central Modality)
Pen Input

Recognizer +

Language

Semantic Parser

Model

Semantic
Keyboard Input

Parser
Semantic
Parser

Other Inputs

Surface Semantics

Context
Manager

Discourse

Semantic

Manager

Model

Discourse Semantics
Responses

Response

Behavior

Manager

Model

Speech-Centric Multimodal User Interface Design in Mobile Technology

used to pass information to and from different

components (or agents) (Kumar & Cohen, 2000;
Schwartz, 1993).
Many best practices and design principles have
been developed for the speech-centric MUI design
in the past decades (Becker, 2001; Dybkjaer &
Bernsen, 2001; Ravden & Johnson, 1989; Reeves,
Lai, J., Larson, J.A., Oviatt, S., Balaji, T.S., Buisine, et al. 2004), which we summarize next.
First, the system should explicitly inform the
user about its state through appropriate feedback
within a reasonable amount of time, so as to avoid
state errors, that is, the users perceived state
is different from the systems perceived state.
The feedback can be in different modalities, but
must be clear and accurate. If speech feedback is
used, recorded speech is usually preferred over
the synthesized speech, due to its higher degree
of naturalness. Note that the recorded speech
usually takes a larger amount of resources than
the synthesized speech. Since the memory and
storage available in mobile devices is very limited, designers should strike a balance between
the use of synthesized speech and of recorded
speech. The system should follow real-world
conventions, and use the words, phrases, and
concepts that are familiar to the users. The system
should also ensure that the output modalities be
well synchronized temporally. For example, the
spoken directions should be synchronized with
the map display.
Second, the system should provide sufficient
flexibility so that users can select the modalities
that are best for the task under the specific environments. For example, the user should be able to
switch to a nonspeech modality when inputting
sensitive information such as personal identification numbers and passwords. A good MUI design
should also allow users to exit from an unwanted
state via commands that are global to the system,
instead of having to go through an extended dialog.
The system should provide enough information
(e.g., through prompts) to guide novice users to

use the system, yet at the same time allow bargeins and accelerators for the expert users to reduce
the overall task completion time.
Third, the system should be designed to allow
easy correction of errors. For example, the system
should provide context sensitive, concise, and effective help. Other approaches include integrating
complementary modalities to improve overall
robustness during multimodal fusion; allowing
users to select a less error-prone modality for a
given lexical content, permitting users to switch
to a different modality when error happens; and
incorporating modalities capable of conveying
rich semantic information.
Fourth, the systems behavior should be consistent internally and with users previous experiences. For example, a similar dialog flow should
be followed and the same terms should be used
to fulfill the same task. Users should not have to
wonder whether the same words and actions have
different meaning under different context.
Fifth, the system should not present more information than necessary. For example, dialogues
should not contain irrelevant or rarely needed information, and the prompts should be concise.
While the best practices summarized are common to all speech-centric MUIs, some special
attention needs to be paid to speech modality and
multimodality fusion due to the great variations
of mobile device usage environments. We address
these special considerations next.

Special considerations for

speech modality
There are two main challenges for the use of
speech modality on mobile devices. First, the
resources on mobile devices, in particular, CPU
speed, memory, and communication bandwidth,
are very limited. Second, speech recognition accuracy degrades substantially in realistic noisy
environments, where there are abrupt changes in

1001

Speech-Centric Multimodal User Interface Design in Mobile Technology

noise, or variable phase-in phase-out sources of

noise as the user moves. For example, the recognition accuracy may drop 30-50% inside a vehicle
and cafeteria from that in a quiet environment
(Das, Bakis, Nadas, Nahamoo, & Picheny, 1993;
Lockwood & Boudy, 1992). Since the mobile
devices will be used in these real-field settings
without a close-talk microphone, robustness
to acoustic environment, that is, immunity to
noise and channel distortion, is one of the most
important aspects to consider when designing
speech-centric MUIs on mobile devices. Speech
recognition accuracy and robustness can usually
be improved with a noise-robust speech front-end,
a noise-robust modality-switching interface, and
a context aware language model.

Resource Constrained Speech

Recognition
Speech recognition on mobile devices is typically
carried out with two options: the distributed recognition (Deng et al., 2002) where the recognition
happens at a remote server (Figure 2) and the local
recognition (Deligne, Dharanipragada, Gopinath,

Figure 2. Illustration of distributed speech recognition where the actual recognition happens at
the server (e.g., PC)

1002

Maison, Olsen, & Printz, 2002; Varga, Aalburg,

Andrassy, Astrov, Bauer, Beaugeant, et al., 2002)
where the recognition is carried out completely on
the mobile device. The distributed recognition can
take advantage of the power of the remote server
to achieve a fast and accurate recognition, while
the local recognition can eliminate the requirement
of the device to have a fast data connection.
In the distributed architecture, the main consideration is the latency required to send data
to and from the server. The latency is typically
determined by the communication bandwidth and
the amount of data sent. To reduce the latency, a
typical approach is to use a standard codec on the
device to transmit the speech to the server where
the coded speech is subsequently decompressed
and recognized (as depicted in Figure 3). However,
since speech recognizers only need some features
of the speech signal (e.g., Mel-cepstrum), an alternative approach is to put the speech front end
on the mobile device and transmit only speech
features to the server (Deng et al. 2002), as shown
in Figure 4. Transmitting speech features can
further save bandwidth because the size of the
features is typically much less than that of the
compressed audio signals.
Besides the advantage of using the computing
power at the server to improve speech recognition
accuracy, there are other benefits of using serverside recognition. One such benefit is its better
maintainability compared to the local recognition approach because updating software on the
server is much easier and more cost effective than
updating software on millions of mobile devices.
It, however, does require the recognizer on the
server to be front end or codec agnostic in order
to materialize this benefit. In other words, the
recognizer should make no assumptions on the
structure and processing of the front end (Deng
et al., 2002). Another benefit of using distributed
recognition is the possibility for the server to
personalize the acoustic model, language model,
and understanding model all at the server, saving
the precious CPU and memory on mobile devices.

Speech-Centric Multimodal User Interface Design in Mobile Technology

Figure 3. Distributed speech recognition architecture: speech input is encoded and sent to the server.
Speech feature extraction happens at the server side

Codec

Encoded
Signal

Speech
Front-End

Speech
Features

Speech
Input

Recognition
Text

Speech
Recognition
Engine

Figure 4. Distributed speech recognition architecture alternative: the speech feature extraction happens
on the mobile devices. Only the features are sent to the server

Speech
Input

Speech
Front-End

Speech
Features
Recognition
Text

In the past, distributed recognition is unquestionably the dominant approach due to the low CPU
speed and small amount of memory available
on the mobile devices. Nowadays, although the
CPU speed and memory size are increasing
dramatically, distributed recognition is still the
prevailing approach over local recognition due to
the advantages discussed previously.

Speech
Recognition
Engine

The major issue of the local recognition

architecture is the low recognition speed and
accuracy due to the slow CPU speed and low
memory available on mobile devices. Speech
recognizers running on mobile devices need to
be specially designed (Deligne et al., 2002, Li,
Malkin, & Bilmes, 2006; Varga, Aalburg, Andrassy, Astrov, Bauer, Beaugeant, 2002) to fit the

1003

Speech-Centric Multimodal User Interface Design in Mobile Technology

requirement since speech recognizers designed

for the desktop or telephony systems cannot be
directly deployed to mobile devices. The greatest
benefit of using the local recognition approach is
its independency of the network connection and
the server and hence, can be used everywhere
under any conditions. Given the consistent improvement of the CPU speed and memory on the
mobile device hardware, in the future, the local
recognition approach is expected to become more
and more popular for simple tasks such as name
dialing and media playing.

Noise Robust Speech Front End

Noise robustness is one of the most important
requirements for speech-centric MUI on mobile
devices. It has attracted substantial attention in
the past several years. Many algorithms have
been proposed to deal with nonstationary noises.
A popular one is an advanced feature extraction
algorithm (jointly developed by Motorola Labs,
France Telecom and Alcatel) that was selected
in February of 2002 as a standard in distributed
speech recognition by the European telecommunications standards institute. The algorithm
defines the extraction and compression of the
features from speech that is performed on a local,
terminal device, for example, a mobile phone.
These features are then sent over a data link to
a remote back-end processor that recognizes
the words spoken. The major components of this
algorithm are noise reduction, waveform processing, cepstrum calculation, blind equalization,
and voice-activity detection. The noise reduction component makes use of two-stage Wiener
filtering (Macho, Mauuary, No, Cheng, Ealey,
Jouvet, et al., 2002).
The stereo-based piecewise linear compensation for environments (SPLICE), which has been
used in the MiPad system (Deng et al., 2002),
is another effective algorithm for noise robust
speech feature extraction. SPLICE is a cepstrum
enhancement algorithm dealing with additive

1004

noise, channel distortion, or a combination of the

two. It is a dynamic, frame-based, bias-removal
algorithm with no explicit assumptions made on
the nature of the noise model. In SPLICE, the noise
characteristics are embedded in the piecewise
linear mapping between the stereo clean and
distorted speech cepstral vectors. SPLICE has a
potential to handle a wide range of distortions,
including nonstationary distortion, joint additive and convolutional distortion, and nonlinear
distortion (in time-domain), because SPLICE can
accurately estimate the correction vectors without
the need for an explicit noise model.

Modality Switching
One of the problems in speech recognition under noisy environment is modality switching.
If the speech recognition engine is always on,
noises and by-talks may be misrecognized as a
legitimate user input and hence, can erroneously
trigger commands.
A widely used modality switching approach
is called push to talk, where the user presses
a button to turn on the speech recognizer, and
releases the button to turn off the recognizer.
Another approach is called tap & talk (Deng
et al., 2002; Huang, Acero, A., Chelba, C., Deng,
L., Duchene, D., Goodman, et al., 2000, Huang
et al., 2001), where the user provides inputs by
tapping the tap & talk field and then talking
to it. Alternatively, the user can select the tap
& talk field by using the roller to navigate and
holding it down while speaking. Tap & talk can
be considered as a combination of push-to-talk
control and indication of where the recognized
text should go. Both the push-to-talk and tap &
talk avoid the speech detection problem that is
critical to the noisy environment under which the
mobile devices are typically deployed.
Figure 5 shows an example of the tap & talk
interface used in the MiPad (Deng et al., 2002).
If the user wants to provide the attendee information for a meeting scheduling task, he/she taps

Speech-Centric Multimodal User Interface Design in Mobile Technology

Figure 5. An example of the Tap & Talk interface (Deng et al., 2002, 2002 IEEE)

the attendees field in the calendar card. When

that happens, the MUI will constrain both the
language model and the semantic model based on
the information on the potential attendees. This
can significantly improve the accuracy and the
throughput. Note that tap & talk functions as a
user-initiative dialog-state specification. With tap
& talk, there is no need for the mobile devices to
include any special mechanism to handle spoken
dialog focus and digression.

Context-Aware Language Model

Here, context refers to any information that can
be used to characterize the situation related to
human-computer interaction. It typically includes
the surrounding environment (e.g., location and
noise condition), the user (e.g., age and gender,
preferences, past interaction experiences, and the
interaction history in the current session), and
the devices (e.g., remaining battery life, available memory, screen-size, screen-contrast, and
speaker volume). Although context-awareness can
be beneficial to all components in an MUI, it is

especially important for improving speech recognition accuracy under noisy environments.
Context information can be utilized in many
different ways in speech modality. One particular approach is to construct the language model
based on the context. For example, the tap &
talk approach (Deng et al., 2002) customizes the
language model depending on the field the user
is pointing to, as mentioned in section 3.3.
Language model can also be customized, based
on the user information and the dialog state. For
example, if the system is expecting the recipient
information, the language model can include only
the names in the global address book. If the user
information is also used, the language model can
also include users contact list and people who
have exchanged e-mails with the user in the past.
An even more effective language model would
weight different names differently, depending on
the frequencies the user exchanged e-mail with
the person, and the recentness of the interaction
(Yu, Wang,

Mahajan, Mau, & Acero, 2003).

Another example of constructing the language

model based on the context and user informa-

1005

Speech-Centric Multimodal User Interface Design in Mobile Technology

tion is described in the speech enabled MapPoint

(Deng & Yu, 2005). Without context information,
the speech recognizer needs to load all location
names and business names in the North America.
This is definitely beyond the ability of most stateof-the-art speech recognizers. However, if the
users location information and/or the interaction
history are known, the system can load only the
location names and business names around the
users current location, and weight all the names
based on the popularity of the names as well as
the users interaction history.
A more advanced context-aware language
model construction technique is discussed by
Wang (2004). This detection-based technique
is used in the second generation of the MiPad
(Wang, 2004). The basic idea of this approach is
to detect the context cues from the users partial
utterances sequentially, and adjust the language
model dynamically for the next part of the utterances. This approach has achieved excellent
user experience.

Language understanding
Good speech recognition accuracy does not always
translate to good understanding of users intents,
as indicated by Wang, Acero, and Chelba (2003).
A robust language-understanding model is needed
to obtain good user experience for speech-centric
MUI applications, especially since speech recognition errors will affect the understanding.
The first issue to address in language understanding is constructing the semantic grammar.
Since the importance of each word to the understanding is different, the words need to be treated
differently. A typical approach is to introduce a
specific type of nonterminals called semantic
classes to describe the schema of an application
(Wang, 2001; Yu, Ju, Wang, & Acero, 2006). The
semantic classes define the concepts embedded in
the linguistic structures, which are usually modeled with probabilistic context-free grammars.

1006

The advantage of introducing the semantic classes

is to make the linguistic realization of semantic
concepts independent of the semantic concepts
themselves. Once the semantic classes are defined,
a robust linguistic grammar can be built using
the approaches similar to the one described by
Yu, et al. (2006).
The transformation from the recognized text
to the semantic representation is usually done
using a semantic parser. For example, in MiPad,
this transformation is done using a robust chart
parser (Wang, 2001). In this parser, the robustness
to ungrammaticality and noise can be attributed
to its ability of skipping minimum unparsable
segments in the input. The algorithm uses dotted
rules, which are standard context free grammar
rules in Backus Naur form plus a dot in front of
a right-hand-side symbol. The dot separates the
symbols that already have matched with the input
words from the symbols that are yet to be matched.
(Wang, 2001, pp. 1556) Since the language models
used in MiPad are dynamically generated based
on the current user information and the tap &
talk field, the parser used in MiPad supports
dynamic grammars. Given that some part of the
users utterances is in the free-style form (e.g.,
the topic of a meeting to be scheduled), they are
modeled as dictation grammar rules. Since speech
recognition is not perfect, the MiPad robust parser
takes into account the N-best list, together with
the associated confidence scores returned from
the speech recognition engine, and combines the
speech recognition score with the parsing score to
obtain the best parsing result. More recent progress
includes using maximum entropy models to classify the tasks and to disambiguate the meaning
of the slots in the recognition result.

Modality fusion
One strong advantage of using MUIs is the improved accuracy and throughput through modality
integration. There are typically two fusion ap-

Speech-Centric Multimodal User Interface Design in Mobile Technology

proaches: early fusion and late fusion. Given that

late fusion has many superior properties over the
early one, as discussed in Section 2, it will be the
focus of our discussion in this section. There are
two tasks in the late fusion: Process and convert
the input signals into a common surface semantic
representation using the semantic parsers (one
specific to each modality), and fuse the surface
semantics into discourse semantics using the
discourse manager.

Semantic Representation and

Unification
The semantic fusion operation requires a meaning representation framework that is common
among modalities, and a well-defined operation
for combining partial meanings.
Many semantic representation formats have
been proposed in the past. For example, in Bolts
(1980) pioneering paper, only very limited modality fusion is required and hence, a simple
semantic representation was used. In the past
decade, researchers (Cheyer & Julia, 1995; Pavlovic & Huang, 1998; Shaikh, Juth, Medl, Marsic,
Kulikowski, & Flanagan, 1997; Vo & Wood,
1996) have converged to using a data structure
called typed feature structures (Kay, 1979) to
represent meanings. Typed feature structure can
be considered as an extended, recursive version
of attribute-value-type data structures, where a
value can, in turn, be a feature structure. It extends
frames (Minsky, 1975) that represent objects and
relations as nested sets of attribute/value pairs,
by using shared variables to indicate common
substructures. A typed feature structure indicates
the kind of entity it represents with a type, and the
values with an associated collection of featurevalue or attribute-value pairs. In the typed feature
structure, a value may be nil, a variable, an atom,
or another typed-feature structure.
The primary operation on typed feature
structure is unification. Typed-feature-structure

unification is an operation that determines the

consistency of two representational structures
and, if they are consistent, combines them into a
single result. (Oviatt, Cohen, Wu, Vergo, Duncan,
Suhm, et. al., 2000, online version pp. 21) Unification can combine complementary input from
different modalities and rule out contradictory
input (Johnston, 1998).
Note that users multimodal inputs may involve
sequentially integrated or simultaneously delivered signal fragments. In other words, temporal
relationships between different input channels
are very important. To fuse modalities, we need
to first determine whether two input fragments
are related. In most of the systems reported, this
is achieved by considering all input contents that
lie within a predefined time window. To do this,
all input fragments need to be time stamped as
soon as they are generated to remove the errors
due to transit delays.
For example, the speech input Show me the
restaurants around here. might have a gestureinput accompanying it either before, during,
or after the actual utterance, and all these three
possibilities should provide the same result. Usually the term before represents a timeframe
of up to several minutes, during represents a
timeframe of 4 to 5 seconds, and after represents
a timeframe of 500ms to 750ms. If these values
are too small, many multimodal inputs will be
considered as unimodal inputs and will not be
integrated. If the values are too large the chances
of an old or invalid user input are likely being accepted as part of a valid multimodal input.
To determine whether two input fragments
should be treated as parts of a multimodal construction or separate unimodal commands, knowledge gained from a user study is very helpful. For
example, it has been shown in Oviatt, DeAngeli,
and Kuhn (1997) that users written input precedes speech during a sequentially integrated
multimodal command. They have also clarified
the distribution of typical intermodal lags.

1007

Speech-Centric Multimodal User Interface Design in Mobile Technology

Semantic Fusion with Uncertain

Inputs
The challenge of semantic fusion with uncertain
inputs is to determine the unified meaning based
on multimodal input fragments associated with
probabilities. This is especially important for
speech-centric MUI because the output of a speech
recognizer is never certain. Note that the unification operation on the typed feature structure
assumes that all input modalities are certain, and
so they cannot be directly applied here. To fuse
modalities with uncertainties, a hybrid symbolic/
statistical architecture that combines statistical
processing techniques with a symbolic unification-based approach is in need. This combined
approach involves many factors when fusing the
semantics. These factors include recognition accuracy of the individual modalities, the way of
combining posterior probabilities, and the prior
distribution of multimodal commands.
Note that a multimodal input gives rise to
three different types of information overlay:
nonoverlayed, overlayed and nonconflicting,
and overlayed and conflicting. Nonoverlayed
information indicates that the input (unimodal
or multimodal) does not have any of the same
information represented multiple times. This is the
simplest condition. Overlayed and nonconflicting
information refers to information segments that
may have been represented multiple times without
a conflict. The overlayed and conflicting information refers to the case that the information has been
provided multiple times and conflicts. There are
many approaches to resolving conflicting information in typed feature structure if no uncertainty
is involved. The unification approach simply
returns the value null when a conflict is detected.
The overlay method returns the first argument
when conflicting information is present. However,
given that the semantic information from different
modalities should not be equally trusted, a better
conflicting information resolving approach can be
found to handle input signals that may or may not

1008

be overlapped in their temporal delivery (Oviatt

et al., 1997). Note that overlayed information may
arise when inputs are from different modalities
(e.g., speech and gesture), or when the same-type
modality information occurs multiple times over
an extended time frame. Both these two conditions
need to be handled.
Conventionally, the probability of the merged
feature structures is the cross product of the probabilities of individual feature structures based
on the assumption that inputs are statistically
independent with each other. In this section, we
describe an alternative statistical approach that
has been used in QuickSet (Wu, Oviatt, & Cohen,
1999). This approach uses the associative map
to reduce the unification pairs and membersteams-committee (MTC) model to refine the
multimodal integration process so that different
weights are assigned to different modes and different constituents.
Associative map defines all semantically
meaningful mapping relations that exist between
different sets of constituents for each multimodal
command. In its simplest form, it can be considered
as a simple process of table lookup. For example, if
an MUI consists of only the speech modality and
the pen modality, we can build a two-dimensional
table. If two inputs from different modalities can
be fused, the value at the corresponding cell is
1; otherwise, the value is 0. The purpose of the
associative map is to rule out considerations of
those feature structures that cannot possibly be
unified semantically.
Members-teams-committee weighs the contributions derived from different modality recognizers based on their empirically-derived relative
reliabilities. MTC consists of multiple members,
multiple teams, and a committee. members are
the individual recognizers that provide a diverse
spectrum of recognition results (local posterior
probabilities). Member recognizers can be on more
than one team. Members report their results to
their recognizer team leader, which then applies
various weighting parameters to their reported

Speech-Centric Multimodal User Interface Design in Mobile Technology

scores. Furthermore, each team can apply a different weighting scheme, and can examine different
subsets of data. Finally, the committee weights the
results of the various teams, and reports the final
recognition results. The parameters at each level of
the hierarchy are trained from a labeled corpus.
(Oviatt, et al., 2000, online version, p. 24).

Conclusion and future

directions
In this chapter, we discussed the importance of
using the MUI in mobile devices, and described
the state-of-the-art technologies in designing
speech-centric MUI in mobile devices. Specifically, we discussed the noise robustness technologies, the reliable modality switching methods,
the context-aware language model, and the robust language-understanding technologies that
contribute to the usability of the speech modality. We also described the modality integration
technologies that are important to improving the
accuracy and throughput of the MUI. Although
these technologies have greatly advanced the
speech centric MUI design and development in
the mobile devices, future research is needed in
the following areas.

Microphone Array Processing

Noise robustness is still a challenging research
area for speech-centric MUIs. Although many
single-microphone noise robustness technologies
(e.g., Deng, et al., 2002; Macho, et al. 2002) have
been proposed to improve speech recognition accuracy under noisy environments, the progress so
far is still limited. Given the continuous decrease
in the hardware price, using microphone array on
mobile devices is a trend to combat noisy acoustic
conditions and to further decrease speech recognition errors. Microphone array algorithms, which
take advantage of the received signal differences
between microphones, can achieve noise suppres-

sion of 10-15 db effectively (Tashev & Malvar,

2005). Future research is needed for more efficient
and effective algorithms using low-cost, low-quality microphone arrays that may be equipped in
speech-centric mobile devices.

Error Handling Techniques

Fragile error handling continues to be a top interface problem for speech-centric MUI (Karat,
Halverson, Horn, & Karat, 1999; Rhyne & Wolf,
1993; Roe & Wilpon, 1994). A great amount of
research work needs to be done in developing
graceful error-handling strategies in speechcentric MUI. First, new statistical methods need
to be developed to reduce errors through mutual
disambiguation between modalities. Second, new
dialog strategies (e.g., mixed initiative) need to be
developed to allow easy correction of the errors.
Third, the system needs to be able to adapt to
different environments and challenging contexts
to reduce errors. Fourth, better robust speech
recognition technologies need to be developed to
increase the speech recognition accuracy under
a wide range of environments.

Adaptive Multimodal Architectures

In most current MUI systems, their behaviors
are predesigned by the developers. The system
does not automatically learn to improve the
performance as users use the system. Given that
mobile devices are usually used by a single user,
it is very important to develop adaptive MUI
architectures.
For example, Oviatt (1999) showed that any
given users habitual integration pattern (simultaneous vs. sequential) is apparent at the beginning
of their system interaction. When the user uses the
system, the interaction pattern remains the same.
An adaptive MUI system that can distinguish and
utilize these patterns to improve the modality
fusion could potentially achieve greater recognition accuracy and interactive speed. Another

1009

Speech-Centric Multimodal User Interface Design in Mobile Technology

example is for the system to gradually change

the behavior (e.g., automatically predict the users
next action) when the user changes from a novice
to an experienced user.
Future research in this area would include what
and when to adapt, as well as how (e.g., through
reinforcement learning) to adapt MUI systems so
that their robustness can be enhanced.

Mixed Initiative Multimodal Dialog

Most current speech-centric MUI systems are
user initiative, where the user controls the dialog flow (for example, through push to talk). A
user-initiative system can be modeled as a set of
asynchronous event handlers. In a more advanced
system, the system should also actively interact
with the user to ask for missing information
(which is called mixed initiative). For example,
if the user wants to search for the phone number
of a business using a mobile device and he/she
forgets to mention the city and state information,
the dialog system should automatically ask the
user for that information through the multimodal
output devices.
Future research should address the design and
development of consistent and efficient conversational interaction strategies that can be used
by different multimodal systems. Multimodal
dialogue systems should be developed within a
statistical framework (Horvitz, 1999) that permits
probabilistic reasoning about the task, the context,
and typical user intentions.

References

Bregler, C., Manke, S., Hild, H., & Waibel, A.

(1993). Improving connected letter recognition
by lipreading. Proceedings of the International
Conference on Acoustics, Speech and Signal
Processing, 1, 557-560.
Cheyer, A., & Julia, L. (1995). Multimodal maps:
An agent-based approach. International Conference on Cooperative Multimodal Communication
(pp. 103-113).
Codella, C., Jalili, R., Koved, L., Lewis, J., Ling,
D., Lipscomb, J., Rabenhorst, D., Wang, C., Norton, A., Sweeney, P., & Turk, C. (1992). Interactive simulation in a multi-person virtual world.
Proceedings of the Conference on Human Factors
in Computing Systems (pp. 329-334).
Cohen, P. R., Dalrymple, M., Moran, D. B., Pereira,
F. C. N., Sullivan, J. W., Gargan, R. A., Schlossberg, J. L., & Tyler, S. W. (1989). Synergistic use
of direct manipulation and natural language.
Proceedings of the Conference on Human Factors
in Computing Systems (pp. 227-234).
Cohen, P. R., Johnston, M., McGee, D., Oviatt, S.,
Pittman, J., Smith, I., Chen, L., & Clow, J. (1997).
Quickset: Multimodal interaction for distributed
applications. Proceedings of the Fifth ACM International Multimedia Conference (pp. 31-40).
Das, S., Bakis, R., Nadas, A., Nahamoo, D. &
Picheny, M. (1993). Influence of background noise
and microphone on the performance of the IBM
TANGORA speech recognition system. Proceedings of the IEEE International Conference on
Acoustic Speech Signal Processing (pp. 71-74).

Becker, N. (2001). Multimodal interface for mobile

clients, Retrieved July 16, 2006, from http://citeseer.ifi.unizh.ch/563751.html

Deligne, S., Dharanipragada, S., Gopinath, R.,

Maison, B., Olsen, P., & Printz, H. (2002). A
robust high accuracy speech recognition system
for mobile applications. IEEE Transactions on
Speech and Audio Processing, 10(8), 551-561.

Bolt, R. A. (1980). Put-that-there: Voice and gesture at the graphics interface. Computer Graphics,
14(3), 262-270.

Deng, L., Wang, K., Acero, A., Hon, H., Droppo,

J., Boulis, C., Wang, Y., Jacoby, D., Mahajan, M.,
Chelba, C., & Huang, X.D. (2002). Distributed

1010

Speech-Centric Multimodal User Interface Design in Mobile Technology

speech processing in MiPads multimodal user

interface. IEEE Transactions on Speech and Audio
Processing, 10(8), 605-619.
Deng, L, & Yu, D. (2005). A speech-centric perspective for human-computer interface - A case
study. Journal of VLSI Signal Processing Systems
(Special Issue on Multimedia Signal Processing),
41(3), 255-269.
Dybkjaer, L., & Bernsen, N.O . (2001). Usability
evaluation in spoken language dialogue system.
Proceedings of the Workshop on Evaluation
for Language and Dialogue Systems, Association for Computational Linguistics 39th Annual
Meeting and 10th Conference of the European
Chapter,,(pp. 9-18).
Fukumoto, M., Suenaga, Y., & Mase, K. (1994).
Finger-pointer: Pointing interface by image processing. Computer Graphics, 18(5), 633-642.
Gong, Y. (1995). Speech recognition in noisy
environments: A survey. Speech Communication,
16, 261-291.
Gould, J. & Lewis, C. (1985). Design for usability:
Key principles and what designers think. Communications of the ACM, 28(3): 300-301.
Holzman, T. G. (1999). Computer-human interface
solutions for emergency medical care. Interactions, 6(3), 13-24.
Hone, K. S., & Graham, R. (2001). Subjective
assessment of speech system interface usability.
Proceedings of the Eurospeech Conference (pp.
2083-2086).
Horvitz, E. (1999). Principles of mixed-initiative
user interfaces. Proceedings of the Conference
on Human Factors in Computing Systems (pp.
159-166).
Huang, X., Acero, A., Chelba, C., Deng, L.,
Droppo, J., Duchene, D., Goodman, J., Hon,
H., Jacoby, D., Jiang, L., Loynd, R., Mahajan,
M., Mau, P., Meredith, S., Mughal, S., Neto, S.,

Plumpe, M., Stery, K., Venolia, G., Wang, K., &

Wang. Y. (2001). MIPAD: A multimodal interaction prototype. Proceedings of the International
Conference on Acoustics, Speech, and Signal
Processing, 1, 9-12.
Huang, X., Acero, A., Chelba, C., Deng, L.,
Duchene, D., Goodman, J., Hon, H., Jacoby,
D., Jiang, L., Loynd, R., Mahajan, M., Mau, P.,
Meredith, S., Mughal, S., Neto, S., Plumpe, M.,
Wang, K., & Wang, Y. (2000). MIPAD: A next
generation PDA prototype. Proceedings of the
International Conference on Spoken Language
Processing, 3, 33-36.
Hsu B.-J., Mahajan M., & Acero A. (2005).
Multimodal text entry on mobile devices. The
ninth bi-annual IEEE workshop on Automatic
Speech Recognition and Understanding (Demo).
Retrieved July 20, 2006, from http://research.
microsoft.com/~milindm/2005-milindm-ASRUDemo.pdf
Johnston, M. (1998). Unification-based multimodal parsing. Proceedings of the International Joint
Conference of the Association for Computational
Linguistics and the International Committee on
Computational Linguistics (pp. 624-630).
Karat, C.-M., Halverson, C., Horn, D., & Karat, J.
(1999). Patterns of entry and correction in large vocabulary continuous speech recognition systems.
Proceedings of the International Conference for
Computer-Human Interaction (pp. 568-575).
Kay, M. (1979). Functional grammar. Proceedings of the Fifth Annual Meeting of the Berkeley
Linguistics Society (pp. 142-158).
Kumar, S., & Cohen, P. R. (2000). Towards a
fault-tolerant multi-agent system architecture.
Fourth International Conference on Autonomous
Agents (pp. 459-466).
Li, X., Malkin J., & Bilmes, J. (2006). A

highspeed, low-resource ASR back-end based on

custom arithmetic. IEEE Transactions on Audio,

1011

Speech-Centric Multimodal User Interface Design in Mobile Technology

Speech and Language Processing, 14(5), 16831693.

Lockwood, P., & Boudy, J. (1992). Experiments
with a nonlinear spectral subtractor (NSS), hidden Markov models and the projection for robust
speech recognition in cars. Speech Communication, 11(2-3), 215-28.
Macho, D., Mauuary, L., No, B., Cheng, Y. M.,
Ealey, D., Jouvet, D., Kelleher, H., Pearce, D., &
Saadoun, F. (2002). Evaluation of a noise-robust
DSR front-end on Aurora databases. Proceedings of the International Conference on Spoken
Language Processing (pp. 17-20).
Minsky, M. (1975). A framework for representing
knowledge. In P. Winston (Ed.), The psychology of computer vision (pp 211-277). New York:
McGraw-Hill.
Neal, J. G., & Shapiro, S. C. (1991). Intelligent
multimedia interface technology. In J. Sullivan &
S. Tyler (Eds.), Intelligent user interfaces (pp.1143). New York: ACM Press.
Nielsen, J. (1993). Usability engineering. San
Diego: Academic Press .
Norman, D. A. & Draper, S. W. (Eds.). (1986).
User-centered system design: New perspectives
on human-computer interaction. Hillsdale, NJ:
Lawrence Erlbaum Associates.
Oviatt, S. L. (1999). Ten myths of multimodal
interaction, Communications of the ACM, 42
(11), 74-81.
Oviatt, S. L., Bernard, J., & Levow, G. (1999).
Linguistic adaptation during error resolution with
spoken and multimodal systems. Language and
Speech (special issue on Prosody and Conversation), 41(3-4), 415-438.
Oviatt, S. L., Cohen, P. R., Wu, L., Vergo, J.,
Duncan, L., Suhm, B., Bers, J., Holzman, T.,
Winograd, T., Landay, J., Larson, J., & Ferro, D.
(2000). Designing the user interface for multi-

1012

modal speech and gesture applications: State-ofthe-art systems and research directions. Human
Computer Interaction, 263-322. (online version),
Retrieved July 20, 2006, from http://www.cse.ogi.
edu/CHCC/Publications/designing_user_interface_multimodal_speech_oviatt.pdf
Oviatt, S. L., DeAngeli, A., & Kuhn, K. (1997).
Integration and synchronization of input modes
during multimodal human-computer interaction.
Proceedings of Conference on Human Factors in
Computing Systems (CHI97) (pp. 415-422).
Oviatt, S. L. & Olsen, E. (1994).
Integration themes
in multimodal human-computer interaction. In
Shirai, Furui, & Kakehi (Eds.), Proceedings of the
International Conference on Spoken Language
Processing, 2, 551-554.
Oviatt, S. L., & vanGent, R. (1996). Error resolution during multimodal human-computer interaction. Proceedings of the International Conference
on Spoken Language Processing, 2, 204-207.
Pavlovic, V., Berry, G., & Huang, T. S. (1997).
Integration of audio/visual information for use in
human-computer intelligent interaction. Proceedings of IEEE International Conference on Image
Processing (pp. 121-124).
Pavlovic, V., & Huang, T. S., (1998). Multimodal
prediction and classification on audio-visual
features. AAAI98 Workshop on Representations
for Multi-modal Human-Computer Interaction,
55-59.
Ravden, S. J., & Johnson, G. I. (1989). Evaluating
usability of human-computer interfaces: A practical method. Chichester: Ellis Horwood.
Reeves, L. M., Lai, J., Larson, J. A., Oviatt, S.,
Balaji, T. S., Buisine, S., Collings, P., Cohen, P.,
Kraal, B., Martin, J. C., McTear, M., Raman, T.
V., Stanney, K. M., Su, H., & Wang, Q. Y. (2004).
Guidelines for multimodal user interface design.
Communications of the ACM Special Issue on
Multimodal Interfaces, 47(1), 57-59.

Speech-Centric Multimodal User Interface Design in Mobile Technology

Rhyne, J. R., & Wolf, C. G. (1993). Recognition-based user interfaces. In H. R. Hartson &
D. Hix (Eds.), Advances in Human-Computer
Interaction, 4, 191-250.

DARPA Spoken Language Technology Workshop.

Retrieved July 20, 2006 from http://www.cs.cmu.
edu/afs/cs.cmu.edu/user/tue/www/papers/slt95/
paper.html

Roe, D. B., & Wilpon, J. G. (Eds.). (1994). Voice

communication between humans and machines.
Washington, D.C: National Academy Press.

Vo, M. T., & Wood, C. (1996). Building an application framework for speech and pen input
integration in multimodal learning interfaces.
Proceedings of IEEE International Conference
of Acoustic, Speech and Signal Processing, 6,
3545-3548.

Rubin, P., Vatikiotis-Bateson, E., & Benoit, C.

(1998). Special issue on audio-visual speech processing. Speech Communication, 26 (1-2).
Schwartz, D. G. (1993). Cooperating heterogeneous systems: A blackboard-based meta approach. Unpublished Ph. D. thesis, Case Western
Reserve University.
Shaikh, A., Juth, S., Medl, A., Marsic, I.,
Kulikowski, C., & Flanagan, J. (1997). An architecture for multimodal information fusion.
Proceedings of the Workshop on Perceptual User
Interfaces, 91-93.
Stork, D. G., & Hennecke, M. E. (Eds.) (1995).
Speechreading by humans and machines. New
York: Springer Verlag.
Suhm, B. (1998). Multimodal interactive error
recovery for non-conversational speech user interfaces. Ph.D. thesis, Fredericiana University.
Tashev, I., & Malvar, H. S. (2005). A new beamformer design algorithm for microphone arrays.
Proceedings of International Conference of
Acoustic, Speech and Signal Processing, 3, 101104.
Varga, I., Aalburg, S., Andrassy, B., Astrov, S.,
Bauer, J.G., Beaugeant, C., Geissler, C., & Hoge,
H. (2002). ASR in mobile phones - an industrial
approach, IEEE Transactions on Speech and
Audio Processing, 10(8), 562- 569.
Vo, M. T., Houghton, R., Yang, J., Bub, U., Meier,
U., Waibel, A., & Duchnowski, P. (1995). Multimodal learning interfaces. Proceedings of the

Wang, J. (1995). Integration of eye-gaze, voice and

manual response in multimodal user interfaces.
Proceedings of IEEE International Conference on
Systems, Man and Cybernetics (pp. 3938-3942).
Wang, K. (2004). A detection based approach to
robust speech understanding. Proceedings of the
International Conference on Acoustics, Speech,
and Signal Processing, 1, 413-416.
Wang, Y. (2001). Robust language understanding in MiPAD. Proceedings of the Eurospeech
Conference (pp. 1555-1558).
Wang, Y., Acero, A., & Chelba, C. (2003). Is
word error rate a good indicator for spoken language understanding accuracy? Proceedings of
the Workshop on Automatic Speech Recognition
Workshop and Understanding (pp 577-582).
Wu, L., Oviatt, S., &. Cohen, P. (1999). Multimodal
integration-A statistical view. IEEE Transactions
on Multimedia, 1(4), 334-341.
Yu, D., Ju, Y. C., Wang, Y., & Acero, A. (2006).
N-gram based filler model for robust grammar
authoring. Proceedings of the International
Conference on Acoustics, Speech, and Signal
Processing1, 565-569.
Yu, D., Wang, K., Mahajan, M., Mau, P., & Acero, A. (2003).
Improved name recognition with
user modeling. Proceedings of Eurospeech (pp.
1229-1232).

1013

Speech-Centric Multimodal User Interface Design in Mobile Technology

Key Terms
Modality: A communication channel between
human and computer, such as vision, speech,
keyboard, pen, and touch.
Modality Fusion: A process of combining
information from different input modalities in a
principled way. Typical fusion approaches include
early fusion, in which signals are integrated at the
feature level, and late fusion, in which information
is integrated at the semantic level.
Multimodal User Interface: A user interface
with which users can choose to interact with a
system through one of the supported modalities,
or multiple modalities simultaneously, based on
the usage environment or preference. Multimodal
user interface can increase the usability because
the strength of one modality often compensates
for the weaknesses of another.
Push to Talk: A method of modality switching
where a momentary button is used to activate and
deactivate the speech recognition engine.

Speech-Centric Multimodal User Interface:

A multimodal user interface where speech is the
central and primary interaction modality.
Typed feature Structure: An extended, recursive version of attribute-value type data structures,
where a value can, in turn, be a feature structure.
It indicates the kind of entity it represents with
a type, and the values with an associated collection of feature-value or attribute-value pairs. In
the typed feature structure, a value may be nil,
a variable, an atom, or another typed feature
structure.
User-Centered Design: A design philosophy
and process in which great attention is given to
the needs, expectations, and limitations of the
end user of a human-computer interface at each
stage of the design process. In the user-centered
design process, designers not only analyze and
foresee how users are likely to use an interface,
but also test their assumptions with actual users
under real usage scenario.

This work was previously published in the Handbook of Research on User Interface Design and Evaluation for Mobile Technology, edited by J. Lumsden, pp. 461-477, copyright 2008 by Information Science Reference, formerly known as Idea Group
Reference (an imprint of IGI Global).

1014

1015

Chapter 2.37

Kinetic User Interfaces:

Physical Embodied Interaction with Mobile

Ubiquitous Computing Systems
Vincenzo Pallotta
University of Fribourg, Switzerland
Pascal Bruegger
University of Fribourg, Switzerland
Bat Hirsbrunner
University of Fribourg, Switzerland

Abstract
This chapter presents a conceptual framework for
an emerging type of user interfaces for mobile
ubiquitous computing systems, and focuses in
particular on the interaction through motion of
people and objects in physical space. We introduce
the notion of Kinetic User Interface as a unifying
framework and a middleware for the design of
pervasive interfaces, in which motion is considered as the primary input modality.

Introduction
Internet and mobile computing technology is
changing the way users access information and

interact with computers and media. Personal Computing in its original form is fading and shifting
towards the ubiquitous (or pervasive) computing
paradigm (Want et al., 2002). Ubiquitous Computing systems are made up of several interconnected
heterogeneous computational devices with different degrees of mobility and computing po
wer.
All
of these devices and appliances are embedded in
everyday objects, scattered in space, capable of
sensing the environment and of communicating
with each other, and carried or exchanged by
people. Therefore, we are facing a new ecology
of computing systems that poses new issues in
their integration and usability. Human-computer
interfaces that were designed for desktop personal computers must be re-conceived for this
new scenario. Due to the different capabilities

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Kinetic User Interfaces

of mobile and embedded devices, the pervasive

computing infrastructure, and the nature of their
expected usage, it is apparent that new types of
user interfaces are needed in order to unleash the
usability of new generation distributed computing
applications (see (Rukzio, 2006) for a classification of mobile devices interfaces). Additionally,
the concept of user interface itself seems to
be no longer adequate to cope with ubiquitous
computing systems. Rather, it is the concept of
interaction and user experience that will take over
(Beaudouin-Lafon, 2004).

Ubiquitous Computing
Ubiquitous Computing (henceforth Ubicomp)
is an emerging research sub-area of Distributed
Systems whose main focus is studying how heterogeneous, networked computing devices can be
embedded in objects of daily use in order to enable
new applicative scenarios and user experiences.
Mark Weiser (1991; 1993; 1994) introduced the
term Ubiquitous Computing in the 90s as a new
way to understand computer technology and to lay
the foundations of an expected and necessary computing paradigm revolution. Weisers vision has
been adopted and interpreted by a great number
of researchers, among whom we consider relevant
for our goals the works of (Abowd & Mynatt,
2000; Abowd et al., 2002; Banavar & Bernstein,
2004; Bellotti et al., 2002; Greenfield, 2006; Norman, 1999; Want et al., 2002). We summarize the
Ubicomp vision in four fundamental points that
motivate our effort of providing a new conceptual
framework for Ubicomp user interfaces:
1.

2.

1016

Todays computer (e.g., the personal computer) will disappear, and the computing
power will fade inside the network infrastructure, as it is already the case to some
extent with existing web-services.
Computing will be extremely distributed
and heterogeneous. This will result from
the interconnection of several computing

3.

4.

devices, each specialized in specific tasks

and scattered in the physical environment
(ranging from embedded devices to highperformance servers).
Computer interfaces will no longer capture
the full attention of users. Rather, computer
applications will run in background most
of the time, accomplishing routinized
operations, and they will try to gain users
attention only when strictly required.
Computer interfaces will be unobtrusive and
based on new emerging interaction models
obtained by direct interaction with physical
objects and with the whole environment.

Ubiquitous Computing is often equated to

(or better, confused with) nomadic computing
(Kleinrock, 1997). Nomadic computing is a form
of computing environment that offers its users
access to data or information from any device
and network while they are in state of motion.
In nomadic computing, the use of portable devices (such as laptops and handheld computers)
in conjunction with mobile communications
technologies enables users to access the Internet
and data on their home or work computers from
anywhere in the world. Mobile connectivity certainly does play an important role in Ubicomp,
but it is not the only one. We consider of central
importance the users mobility intended as the
users ability of moving objects and themselves
in the physical space. In fact, in using Ubicomp
systems, users are no longer forced to sit in front
of a desktop computer and to operate it with
mice, keyboards and local input/output devices.
Users will interact through actions performed on
everyday objects that surround them. As pointed
again by Weiser1:
[ubiquitous computing] is different from PDAs,
dynabooks, or information at your fingertips. It
is invisible, everywhere computing that does not
live on a personal device of any sort, but is in the
woodwork everywhere.

Kinetic User Interfaces

User Interfaces for Ubicomp

Systems
Human-computer interaction (HCI) is a very large
domain of research, which includes all related
aspects of interaction with digital appliances.
Although HCI has been mostly focused on graphical user interfaces (GUIs), we consider here those
aspects that pertain to how users will interact
with computers in the 21st century (Winograd,
2000), that is, by direct manipulation of objects in
their physical environment. Paul Dourish (2001)
defined this paradigm as embodied interaction.
According to his definition, meaningful embodied
interaction with a digital system can be obtained
only if an alignment is maintained between the
physical and the digital world. In other terms,
bits and atoms must live together in peace
(Negroponte, 1995). As recently pointed out in
(Sparacino, 2005, p. 2):
[] computation and sensing are moving from
computers and devices into the environment itself.
The space around us is instrumented with sensors
and displays, and this tends to reflect a widespread
need to blend together the information space with
our physical space.
Embodied interaction is thus aimed at exploring new interaction patterns where people
are exploring how to move the interface off the
screen and into the real world (Shafer et al., 2001).
For instance, in Tangible User Interfaces (TUIs)
(Ulmer and Iishi, 2000; Holmquist et al., 2004),
tangible interaction is intended to replace desktop
GUIs interaction and elements with operations
on physical objects. The motion of objects in the
physical space determines the execution of actions, such as item selection (by means of what
in TUI are called phicons i.e., physical icons),
service requests, database updates, and so forth.
Rekimoto (1997) proposed the Pick&Drop pattern,
an extension of the Drag&Drop pattern, to move

items across computers. In his work on graspable

user interfaces, Fitzmaurice (2000) proposed
to extend the interaction with classical GUI by
means of physical objects (such as LEGO bricks)
over an augmented desktop surface. Tangible and
graspable user interfaces are undoubtedly a great
achievement in HCI. However, they are strongly
biased by GUIs interfaces; nearly no new types
of interaction induced by the nature of the physical space and objects have been proposed other
than replications of those available on ordinary
desktop GUIs.
The development of the ideas introduced by
TUIs, combined with the ubiquity of information over the Internet, has led to a new concept
of what counts as an Ubicomp system. Moving
from the assumption that only information appliances could constitute a ubiquitous system,
the whole material world of things and people
now can be made computational and connected.
The movement of the Internet of Things (ITU,
2005; Bleeker, 2006) is aimed at promoting the
idea that any object can have a virtual identity
in the Internet realm, as long as the object can
embed a unique identifier, which corresponds to
an IP address. This is now possible because of the
larger 128 bits IP address space offered by the new
IPv6 protocol (i.e., 3.41038 addresses). According
to (Greenfield, 2006), the type of infrastructure
that enables Ubiquitous Computing is already
technically feasible and it will soon scale up to a
level that will make it possible to safely connect
a huge number of small heterogeneous devices,
possibly embedded in everyday objects.
Augmented Reality (AR) (Mackay, 1998) and
Wearable Interfaces (Barfield & Caudell, 2001)
are emerging technologies that support embodied
interaction for Ubicomp systems. We believe that
Augmented Reality focuses more on how feedback is provided, whereas Wearable Interfaces
focuses on the types of devices that can support
embodied interaction.

1017

Kinetic User Interfaces

Unobtrusive Interfaces
When HCI intersects Ubicomp, many assumptions
that were made when designing interaction for
ordinary computing devices are no longer valid.
In Ubicomp, computers exist in different forms
and only in a minimal portion as ordinary desktop
computers (i.e., where interaction is performed
through screens, keyboards, mice). As pointed
out by Weiser and other promoters of Ubicomp,
interacting with a ubiquitous system should be
realized through an unobtrusive interface, more
precisely, an interface that does not capture the
full attention of the user, who can still use the
system to perform the foreground tasks
(Nardi,
1996)
. In contrast, an obtrusive interface is one
that requires an unjustified cognitive effort to be
operated, thus interfering with the normal usage
of the system. Weiser & Seely Brown (1996) call
this setting Calm Technology in order to stress
the importance of adapting the computers and their
interfaces to human pace, rather that the other way
around. In this vision, computers should follow
users in their daily activity and be ready to provide
information or assistance on demand. Moreover,
they should not require much attention from the
user by asking information that can be autonomously obtained from the actual usage context.
They must be aware of the context and be able
to adapt their behaviour and interfaces to different usage situations. In other words, ubiquitous
computers must be smart and adaptive.

Kinetic-Awareness
Context-awareness is considered as the most important issue in Ubicomp (Baldauf et al., 2006;
Dourish, 2004; Hong et al., 2005). Specifically,
location-awareness is considered a key component
of context in designing user interfaces for mobile
systems. Location-awareness has been always
treated as a sub-case of context-awareness, and
motion as a form of context change. Location
change is taken as a context change for adapting

1018

the applications behaviour, rather than as an

explicit intentional act within the applications
interface.
We believe that location changes occurring
over time can represent more than just contextchange. It can be considered as input modality.
This does not mean that location or motion context
has to be neglected. Motion input and location
context can be used together in the same way as
in handheld GUIs, where interaction with the
mobile devices can be contextualized through
location or motion. For instance, it does make
sense to consider a different interpretation for an
objects motion that occurs in a different location,
or to interpret the GUI input differently when the
mobile device is moving with different kinetic
properties. To clarify this aspect let us consider
two examples.
The first example is about two possible situations where a paraglide is flying i) over a lake,
or ii) over the ground. Motion input is treated accordingly to the actual situation; when flying over
the lake some manoeuvres are considered risky,
while they are considered safe if flying over the
ground (landing is possible everywhere).
The second example is a situation where
implicit interaction with GUI is enabled if the
handheld device is detected moving with a certain speed. For instance, suppose that a dialogue
box asking for a confirmation pops up on the
handhelds GUI. If the user is moving, at a speed
of more than 5 Km/h, then after a timeout, the
interface assumes that the default choice has been
(implicitly) selected. If the user is moving at lower
speed or is still, the interface will wait for input
without the timeout.
According to Dix et al. (2000), space and location define a new design space for interactive
mobile systems. Mobile devices have increasing capacity in providing location and motion
information as part of their usage context. They
also acknowledged that accounting for motion
as an input modality for location-aware mobile
Ubicomp systems opens potential research op-

Kinetic User Interfaces

portunities. In fact, they propose and instantiate a

taxonomy of the degrees of mobility with available
technologies and applications. They categorize
degrees of mobility along three dimensions level
of mobility (i.e., the physical bound of the device
to the environment), level of independence (i.e.,
the relation of the used location-aware device with
other devices or to the environment), level of cooperativeness (i.e., the extent to which the device
is bound to a particular individual or group). Some
existing mobile devices and applications are then
categorized according to this three-dimensional
taxonomy. Some difficult or even impossible cases,
such as a fixed-pervasive-personal, that might be
instantiated by a fully isolated active cell (e.g., a
prison cell with motion sensors). The combination
that is relevant for us is that of pervasive-mobilepersonal/group devices. Such devices might be
location/motion-aware objects that can be used
to interact with Ubicomp systems.
Systems combining location and motion
awareness, henceforth, will be referred to as kinetic-aware systems. Our analysis of how kinetic
information is taken into account in existing
Ubicomp systems revealed that there are two
main views on the role of kinetic information in
user interfaces:
1.

2.

In the first view, location and motion are used

as a component of the usage context that is
exploited by applications running on the
mobile information computing devices for
adapting their behaviour. In this view, there
is no physical interaction with the place (and
its contained artefacts) where the users are
using their own devices. The environment
is not sensing the presence and the physical action of the user. Moreover, neither the
system nor the environment is supposed to
handle spatio-temporally located events.
In the second view, location change and
motion are considered as primary input
modalities reflecting the users goals and
intentions while using Ubicomp applica-

tions. That is, users can intentionally perform

explicit and implicit actions through physical motion. These actions are recognized
and contextualized in the place where they
occur, by possibly affecting the state of
co-located devices and remote systems.
Interaction through motion with physical
space becomes the main focus, rather than
simply contextualizing applications based on
ordinary user interfaces running on mobile
devices.
While the majority of research on locationaware Ubicomp systems focuses on the first
view on kinetic awareness, we focus instead
on the second view, which can be achieved by
recognizing users goals and intentions from
various properties of motion of objects in the
physical space through what we call the Kinetic
User Interface (KUI).
Kinetic awareness can be seen as part of a
more general paradigm, defined in (Dix et al.,
2004) as context-aware computing. This emerging paradigm poses several challenges. One of
the biggest recognized difficulties for this type
of system is interpreting human activity. We do
not propose algorithms for automatic recognition
of human activities (which is a machine learning task). Instead, we propose a framework in
which activities can be decomposed into smaller
and easier-to-detect patterns. These patterns are
instantiated by acquiring input from sensors that
are linked to software components representing
physical objects. We believe that providing a level
of abstraction for activities will make easier the
design and the implementation of Kinetic User
Interfaces for Ubicomp systems.

Related Works and Technologies

Location-aware services are nowadays available to
mobile Internet users. With the advent of Web2.02,
a great deal of applied research and development
recently has been devoted to embedding Internet

1019

Kinetic User Interfaces

technology into everyday life mobile devices,

ranging from pure entertainment to critical applications such as healthcare, national security,
military (Cceres et al., 2006). The rationale
behind these efforts is to provide the mobile
Internet users with great flexibility in authoring,
publishing, and retrieving information, as well as
in accessing services that are relevant in a given
situation and place. A remarkable example of
Web2.0 mobile application for multimedia information retrieval is SocialLight3, which allows the
tagging of geographical location with multimedia
tags (e.g., shadow-tags). The roaming users can
geo-tag a place either by using a mobile phone
when they are physically present there, or by
attaching the tag on the SocialLight Web page
with a GoogleMap mash-up. Information can
further be retrieved from the Web site through
the mashed-up map or by querying the system
by using a mobile, GPS-enabled Internet device.
Geo-tagging is also the focus of the Mobile Multimedia Metadata project of the Garage Cinema
Lab (Davis et al., 2004), whose main purpose is
to cooperatively annotate geo-located pictures.
(Ashbrook et al., 2006) push this concept further
and propose a roadmap for future research with a
new type of scenario based on location-awareness
and motion tracking, in which users can capture
media while moving, and share their experience
by making the captured media stream available
to Internet users in real-time.
Location-awareness in Ubicomp has been
the main focus of several projects since 1990.
Among the projects that heavily rely on location
context are the Aware Home (Kidd et al., 1999)
at GeorgiaTech, the GUIDE project at Lancaster
University (Chervest et al., 2000), the AURA
project at Carnegie Mellon University (Sousa &
Garlan, 2002), the GAIAs Active Space project at
University of Illinois Urbana Champaign (Romn
et al., 2002), the Interactive Maps at ETHZ (Norrie, 2005), and the Global Smart Places project
(Meyer et al., 2006). It might seem that this is a
widely explored research area, but a closer analy-

1020

sis reveals that all of these projects deal with a

very basic form of motion-awareness. Location
change is taken as a users context change, which
is used for dynamically adapting the applications
behaviour. They all consider a users location as
an additional parameter of an explicit service
request. Then, the services output is delivered
on the handheld device carried by the user or on
a nearby display.
The Cyberguide project (Abowd et al., 1996)
is one among the first attempts in taking a users
motion into account (see also (Schilit, 1995)). A
tourist equipped with indoor (IR beacons) and
outdoor (GPS) localization devices can automatically receive contextual relevant information on a
PDA while moving, and feed a trip journal.
In the EasyLiving project at Microsoft (Brumitt
et al., 2000), a geometric model of the physical
space is used in order to enable physical embodied
interaction by representing the physical relationships between entities in the world. Unfortunately,
this model has a limited scope since it is adapted
to room scale, and it only considers current spatial
relationships between objects, while ignoring their
motion within the tracked space.
The Intelligent Workspace project at MIT
(Koile et al., 2003) is a system that records the
users location history and learns the so-called
activity zones. Activity zones are portions of the
physical space where the user is doing specific
activities and which are repeatedly observed by
the system. Once an activity zone is learned, the
settlement of user in it will automatically trigger
a number of pre-defined services that support
the observed activity. In other words, the system
reacts to context change and, in particular, to the
users location change.
However, none of the above-mentioned systems explicitly recognize motion as the primary
input modality and as a mean of performing a
purposeful (explicit or implicit) action within
the applications interaction space. The following systems make a further step towards a more
explicit notion of kinetic-awareness.

Kinetic User Interfaces

In the Sentient Computing project at AT&T

Cambridge Research labs (Addlesee et al., 2001),
motion tracking is an essential feature for interacting with the system. The system is built around
the ActiveBat infrastructure for tracking location
and motion using ultrawide-band sensors (now
commercialized by Ubisense4). A few applications for the ActiveBat have been proposed, such
as FollowMe that allows users to move their
input-output environments over several devices
scattered in the environment (e.g., phone call
forwarding, virtual desktop displacement), or
the Virtual Mouse that allows users carrying
the ActiveBat device to use it as a mouse over a
wall display. A relevant aspect of this project for
us is the adopted context modelling techniques.
The application-level and the user-level (mental)
model of the environment are kept aligned by the
system (Harter et al., 2002). In other words, when
users perform actions, they update their mental
representation of the resulting state of the environment by directly observing it. This representation
might not be consistent with the information that
applications have gathered through sensors. The
system must take care of checking and possibly
restabilising the lost alignment.
The Sonic City project (Gaye et al., 2003)
exploits motion in the urban landscape as a way
for interactively creating a musical experience.
The user motion is tracked, as well as the current position over the city map. Motion and location contexts are combined with other contexts
obtained through wearable sensors in order to
influence the composition of music content in
real-time. Users of the Sonic City interface can
hear the result of musical composition during
their walking activity.
Other applications of the motion context are
pervasive games and races. Games that involve
location tracking (like trails in (Spence et al.,
2005)) are suited for exploiting the motion context.
For instance, in CatchBob (Nova et al., 2006), a
multi-player game developed at EPFL, the players
motion is tracked and their paths are made visible

to other team members on a digital map. The goal

of the game is to cooperatively perform a task by
looking at the motion of the other team members
on a handheld display and by communicating
direction suggestions in real-time.

Overview of the Chapter

In this chapter, we propose a unifying framework
for the design and implementation of Kinetic User
Interfaces for Ubicomp systems by (i) defining a
set of fundamental concepts and (ii) by presenting a middleware that enables the use of motion
in physical spaces as the primary interaction
modality.
The remainder of this chapter is structured
as follows. In Section 2, we provide some backgrounds and intuitions about the distinguishing
features of KUIs compared to other types of user
interfaces. We also present a few motivating applicative KUI-based scenarios. In Section 3, we
detail the main concepts of KUI and we present
the structure of its middleware architecture. Conclusions are finally discussed in Section 4.

Kinetic User Interfaces

The term Kinetic is derived from the Greek
kinetikos, which means moving of, relating to
or resulting from motion (the action or process
of moving). In physics, kinetic theory explains
the physical properties of matter in terms of the
movement of its constituent parts; kinetic energy
refers to energy, which a body possesses by virtue
of being in motion. Kinetic abilities of humans
are of no question. People move and change
their current spatial location all the time and
in a mostly unconscious way. Humans are also
capable of modulating motion in several ways,
by keeping or varying their speed, by following
different trajectories or patterns (e.g., dancing), or
by executing various types of motion in parallel
(e.g., gesturing while walking).

1021

Kinetic User Interfaces

At different scales and contexts, motion (or

absence of motion) can be recognized as a purposeful action. For instance, if a tourist stops long
enough in front of a statue, it might be reasonable
to assume that he or she is observing the monument. What if the statue (or the environment)
would be smart enough to provide the tourist with
relevant information about its author or style, or,
even smarter, to figure out that the tourist has
already stopped at the statue, to avoid repeating
the old information unless the tourist explicitly
requests it?
A familiar everyday situation occurs when
items are passed through a bar code reader in a
grocery store counter. Their motion (the passage
in a given place) is recognized as a purchase
transaction. Another typical KUI situation takes
place when somebody, possibly unconsciously,
performs dangerous actions such as moving
into dangerous areas. In these cases, a monitoring system could alert the user by signalling the
potential danger.

KUI vs. GUI

Motion in the physical space is such a common
and pervasive phenomenon that we hardly recognize its status of an interaction modality with
a computer. While it is apparent that motion
plays an essential role in WIMP5 Graphical User
Interfaces (GUI), the virtual and limited nature of
GUIs space, the desktop, seems not to afford the
full bodily motion interaction (Beaudoin-Lafon,
2000). For instance, it does make little or no sense
to talk about speed or acceleration of the pointer.
Sometimes, however, these properties are taken
into account by specific types of PC applications
like games or flight simulators.
We introduce the concept of Kinetic User
Interface (KUI) as a way of endorsing Weisers
Ubiquitous Computing vision (Weiser, 1993)
and the Dourishs Embodied Interaction vision
(Dourish, 2001) discussed in the previous section. Accordingly, KUIs are intended to enable a

1022

new interaction model for pervasive computing

systems in which the motion of objects and users
in the physical space are recognized as events and
processes to which the system reacts. To make a
parallel with ordinary, pointer-controlled Graphical User Interfaces (GUIs), moving the pointer
on the display and clicking on a graphical item
is recognized by the system as an intentional act,
which usually triggers a systems reaction on the
software representation of the domain object associated to the selected graphical item and, possibly,
the execution of an action on the domain object
specified in the application currently running on
the computer. Similar to hovering the pointer
over a desktop in GUIs, in KUIs users can trigger input events for the computing environment
by moving themselves and by displacing tracked
objects. Users can exploit the physical space by
executing actions/operations on physical objects,
such as moving, grabbing, touching, juxtaposing,
whose effects are reflected in the application objects. For instance, following a path or executing
a pre-defined motion pattern can be viewed as
similar to mouse gestures in ordinary desktop
GUIs and can consequently trigger reactions by
the gesture-enabled application.
KUIs are not limited to single-user interfaces
and do not impose a unique locus of interaction.
Hence, it enables richer interactions than GUIs
and it is better suited to ubiquitous and mobile
computing environments. Motion as an input
modality can be used alone or in combination
with other input modalities available to the user
for interaction with the system, which are directly
afforded by other mobile devices carried by the
users and by fixed input devices located in the
interaction space (e.g., ordinary point and click,
or speech recognition).

Feedback Management
As in GUIs, an important issue in KUIs is feedback management. Due to the different nature
of physical space with respect to GUIs synthetic

Kinetic User Interfaces

space, feedback cannot be provided in the same

way as for GUIs. Since one of the goals of KUI is
to help build unobtrusive interfaces, we give back
to users only the minimal amount of information
required to inform them that their interaction
with the physical space has been successfully
recognized. In turn, the system should avoid
interfering with the users current activity if
the effects of the recognized action have only a
peripheral importance to the current foreground
task (i.e., if the effects impact only objects that
are not in the current focus of attention of the
user). Moreover, since the physical space already
allows for the direct manipulation of real objects,
feedback should only inform users about those
effects produced in the computing space to (virtual) domain objects.
Although users might not always be aware of
what effects are caused by their motion and the motion of tracked objects, they will be unobtrusively
notified when their motion has been detected and
interpreted by the system. Different than GUIs or
even in Augmented Reality systems, there will
be no need to display a synthetic image of the
moving object. The only graphical components
of the interface will be those corresponding to additional modalities. For example, we can imagine
a scenario where a dialog box is prompted on a
mobile device or on an embedded display when
the user is detected6 to walk by or stop at a specific
point of interest.
A feedback mechanism of control is also
necessary for other reasons, such as privacy; to
grant a certain level of protection, users must be
somehow aware when their presence and motion
is being currently tracked. Consequently, they
must always be given the possibility to stop the
tracking of the mobile device and to be allowed
to use an alternative interaction modality.

KUI Interaction Patterns

Although KUI interaction patterns can be radically
different from GUI patterns, some of the most ef-

fective GUI patterns, such as Drag&Drop, can be

transferred and adapted to KUI interaction with
physical space. For instance, in a KUI-enabled
SmartHome, the user can drag the media being
currently played in the living room and drop it
to the bedroom just by moving a representative
localizable object such as the remote controller.
It is worth noting that the Drag&Drop pattern
is provided as an interaction pattern by the KUI
middleware and can be activated (and recognized)
for specific applications such as the SmartHome
control system7.
Another useful pattern we include in KUI is
continuous tracking. Continuous physical motion
is comparable to mouse-gestures in GUIs. KUIenabled applications are supposed to recognize
certain kinetic patterns that might be naturally
performed by users during other activities or
specific situations.
As an example of the continuous tracking
pattern, consider the scenario where the user is
driving a car and some of the cars motion parameters are obtained by embedded sensors such
as a GPS tracking system and an accelerometer.
The sensors reveal that the car is decelerating in
the proximity of a gas station (i.e., a geo-located
point of interest already known by the application).
This kinetic pattern (deceleration) is detected by
the KUI and interpreted by the application as the
users intention of refuelling at the gas station.
This hypothesis might be corroborated by other
contextual information from the current cars
sensors (e.g., the fuel level being almost zero).
As a result of this behaviour, the system will
pro-actively prompt the driver with the current
gas prices at the approaching gas station. The application might also perform further contextual
inferences and inform the user that keeping the
current speed and considering the current fuel
level he/she can reach the next gas station that has
better gas prices. However, if the system detects
that the fuel level is high or the fuel tank is even
full, it will not react because it can infer that the

1023

Kinetic User Interfaces

driver stops for other (unknown) reasons (e.g., to

take a pause).
This is a clear example of how KUI interaction
differs from ordinary location-aware user interfaces. In this scenario, the driver passing by a gas
station does not need to explicitly inquire about the
gas prices. Current location information is only
used to contextualize the query that is triggered
as a result of the speed change occurring in the
proximity of a given point of interest.
With regard to enabling technologies, continuous motion tracking is already available with
current GPS-based car navigation systems, and
easily can be integrated with personal mobile
devices (e.g., SmartPhones, PDAs) connected to
mobile Internet infrastructures (such as UMTS,
GPRS, WiMax). With ordinary GPS navigation
systems, the user can always check the current
location on a graphical map and the proximity to
point of interests. With KUI, we extend the possibilities of these systems with an additional level
of interactivity and integration with networked
services (Pallotta et al., 2006).

The UbiGlide flight navigator detects the motion of

the paraglide. By interpreting her current activity
UbiGlide infers Janes intention to cross the lake.
UbiGlide then senses the environment, namely, the
winds force and direction, the lake altitude, the
distance between the paraglide and the opposite
shore, and finally concludes that the crossing is
not possible. Jane is informed immediately about
the risk of danger. Later, she is so focused on the
flight that she finds herself approaching a no-fly
zone (e.g., an airplane landing strip). UbiGlide
detects this possibility and alerts her about the
danger.

KUI-Enabled Scenarios

SmartHome and SmartCity

KUI-enabled applications have a different purpose

compared to ordinary location-aware ones. Since
motion is used as an input modality, KUI-based
applications are expected to provide a higher level
of fluidity in interaction and user experience. In
those situations in which the interface should not
interfere with the foreground users activity (which
in turn might or might not be a computer-based
one), KUI will allow unobtrusive interaction with
a computer. We consider here three case studies
that have been developed so far and that exemplify
the benefits of KUI interaction patterns.

Julia, Steve, and their daughter Monica live in

Geneva. Julia is a busy businesswoman and Steve
is a researcher at the university. Monica is in her
last year of college. They live in a flat equipped
with latest IT technology. They own a networked
SmartFridge, which detects when food items go
in and out, and automatically generates a list
of missing items. The UbiShop system looks at
the list generated by the SmartFridge and sends
requests to buy the missing items to any family
members who pass by a grocery store. Monica is
on her way back home after school and passes by
a grocery store. A reminder to buy milk is sent
on her mobile phone by the UbiShop system. She
decides to ignore the reminder since she knows
that another grocery store is on her way home.
Meanwhile, Steve is also on his way home and near
a grocery store. He also receives the reminder and

Safety in Air Sports

Jane is flying with her paraglide over the Alps
and is trying to reach the other side of a lake she
is currently over. Can she do it without any risks?

1024

In this scenario the current paraglide motion

is not only tracked but also interpreted. An activity report is obtained by composing a number of
basic flight movements in order to recognize more
complex behaviour patterns. For instance, spiral
ascension is made of several turns and altitude
changes. Moreover, activities are filtered by other
contextual information such as wind speed and
direction, air humidity, and so forth.

Kinetic User Interfaces

decides to buy the milk. This purchase causes the

deletion of the milk from the shopping list, so that
Monica will no longer be bothered. When back
home, Steve does not put the milk in the fridge.
After a while, the SmartFridge wonders why
the milk has not yet been put inside, so a request
about this item is sent to Steve who had simply
forgot the milk in the car.
This illustrates how KUI can contribute to
context-aware collaboration8 in a mixed urban/
home environment. Family members activities
are coordinated according to actual members
mobility. The system decides to adapt a workflow by interpreting team members behaviors
in context. Here, the role of the KUI interface
is twofold. First, it allows triggering the task
assignment when the user is moving into a zone
where the task could be accomplished. Second,
it detects from the users speed whether he/she is
likely to be willing to perform the assigned task.
For instance, if the user is running, the application
could interpret this motion pattern as being in
a hurry and might decide not to bother the user.
Another interesting aspect where KUI plays a
role is when the user is expected to perform an
action and this action does not occur. This is the
case when, in our scenario, Steve forgot to put
the milk in the fridge. The application subscribed
to a motion event that does not occur within a
time interval.

Safe-Critical Work Environments

Bill is a specialized worker in a chemical plant.
He typically operates an industrial chemical
reactor. He wears a head-mounted display and
he is connected to the main control desk through
wireless radio communication gears. While normally operating the reactor, suddenly Bill starts
running (accelerates) toward the emergency exit.
He is not able to alert the control desk about what
is happening. The KUI interface detects this motion pattern as abnormal and figures out that the

operator is trying to escape from a dangerous

situation. The system then opens the doors on the
pathway toward the exit of the reactor building and
then immediately closes them after the operator
is sensed to have passed through them.
In certain industrial settings such as chemical plants, it is important that the operator keeps
his hands free in order to be able to do his usual
manual work, while at the same time he/she accesses the automated commands and looks at
the supervision information needed to complete
the task. The role of KUI is apparent because it
provides an additional implicit input modality that
might serve to detect sudden instinctive reactions
to dangerous situations.

The KUI Model

In this section, we present the main concepts of
Kinetic User Interfaces that are implemented as
software components in the KUI middleware
architecture.

The KUI Conceptual Taxonomy

In KUI, motion is a main (or primary) interaction
modality afforded by the physical space to users
through the motion of tracked entities. Tracked
entities are any objects or autonomous (possibly
living) things for which we can provide location
and motion information. Tracked entities are
represented by KUI components called Kuidgets.
Interaction with Kuidgets happens when users
affect their motion properties or change spatiotemporal relationships among them (e.g., an object
is entering into an area). For instance, when the
user is driving a car, the motion properties of
its corresponding Kuidget will be continuously
updated with its current position, speed, acceleration, and direction.
The term Kuidget has been chosen to make
the parallel with a GUIs widgets, that is, soft-

1025

Kinetic User Interfaces

ware components that provide public interfaces

for a hardware sensor and whose interaction is
implemented in terms of messages and call-backs
(Winograd, 2001). Kuidgets are the software
counterpart of some real world entities that can
be used for interaction in KUIs. A KUI-enabled
system is thus able to recognize the current location of Kuidgets, and makes sense of their motion
parameters such as path, speed, acceleration,
and direction. Location and motion sensors (e.g.,
GPS or other tracking devices, accelerometers,
compasses, altimeters) typically provide threedimensional location and motion information to
Kuidgets. Kuidgets are classified according to four
main dimensions: geometry; kinetic properties;
degree of autonomy; and type of motion.
From the geometric point of view, Kuidgets can
be arbitrary three-dimensional objects. However,
it makes sense to distinguish between those objects
whose size is not relevant and those for which it
matters. Entities of the first type are considered
as points while others are considered as geometric
shapes. This distinction is application-dependent,
because one entity can be considered a point in
one application and a shape in others. For instance,
a vehicle (a car, a train, a ship, a plane) is a point
when considered as a moving object in the space,
and it is a space when considered as a container
of objects and people. In the KUI model, the
same entity plays two roles at the same time and
is linked to two distinct Kuidgets.
Kuidgets can be fixed or mobile. Fixed Kuidgets
are typically places or landmarks in the physical
space, while mobile Kuidgets are physical entities whose location and motion can be observed
by tracking the entity or can be provided by their
embedded location and motion sensors.
In modelling KUIs dynamics we adopt the
status-event semantics, which means that KUIbased applications should be able to effectively
deal both with the status of objects and with the
events they generate. Thus, as an underlying model
for dealing with Status-Event semantics in KUI,
we adopt the Environment and Artefacts theory for

1026

multi-agent systems as proposed by (Ricci et al.,

2006). Accordingly, we further classify Kuidgets
along their degree of autonomy as artefact
Kuidgets and agent Kuidgets. Artefact Kuidgets
correspond to mobile physical objects that cannot
move autonomously (e.g., a mobile phone, a car).
The motion properties of an artefact Kuidget can
be directly determined and communicated by
the artefact itself (e.g., a GPS sensor + a mobile
network connection) or observed by another entity
(e.g., the infrastructure of the containing object, a
nearby artefact, or an agent). The events of moving, grabbing, dragging, and dropping artefact
Kuidgets are triggered by the detection of their
current kinetic properties in the physical space and
by the actions performed through their interfaces
for direct manipulation (e.g., pressing a button on
the object while moving it). Agent Kuidgets correspond to autonomous moving entities (people,
animals, robots). Agent Kuidgets have a higher
degree of autonomy, and they can induce motion
to other artefact Kuidgets.
At the physical level, there is not much difference between artefacts and agent Kuidgets; they
are essentially KUIs components, and as long as
their corresponding physical objects are providing
their location and motion information, they are
treated equally in KUIs. At a conceptual level,
however, they differ because agents can control
and operate artefacts and have a higher degree of
autonomy. This distinction is particularly useful
when KUI-enabled applications need to determine
the location and motion of Kuidgets in the absence
of up-to-dated information. Motion of artefact
Kuidgets typically has to be somehow causally
linked to agent Kuidgets; sometimes artefacts
cause the motion of agents, and sometimes it is
the other way around.
Artefact Kuidget typically keeps its last observed (communicated) location if they are not
linked to any moving entity (i.e., the law of inertia).
Even in cases where no information can be obtained for an artefact Kuidget, its current location
can be inferred by default just by knowing that

Kinetic User Interfaces

it is unlinked from a moving Kuidget. Different

from artefact Kuidgets, agent Kuidgets have more
freedom. When unlinked from any moving artefact Kuidgets, their current location and motion
status cannot be safely inferred; they can move
without being tracked. Of course, there also might
be the case that an artefact is no longer tracked,
but this case is considered as an error condition
rather than a choice selected by the user. In other
words, in KUI agent Kuidgets can decide when
they can be tracked, while artefacts cannot.
Two Kuidgets can be logically linked and they
can provide location and motion information to
each other. For instance, a car equipped with a
GPS sensor (an artefact Kuidget) can provide
location and motion information to its driver (an
agent Kuidget). Conversely, a user carrying a GPS
sensor can provide location and motion information to the car Kuidget by setting a logical link
between the car and the user who is driving the
car. It is important to notice that when the user
leaves the car, even if the link is destroyed, the
car Kuidget keeps its last location.
It is also our goal to make KUI as general as
possible in order to uniformly cope with different
geographical scales (e.g., tabletop, room, building,
cities) and with different types of location-aware
devices (e.g., GPS, RFID, Wireless cell triangulation, ultrasonic, ultra-wideband, infrared). For
this purpose, we also distinguish between different types of physical space. Following (Dix et
al., 2000), we consider topological and symbolic
spaces9. In topological spaces, objects are localized
by their exact position by means of an absolute coordinate system and through a notion of distance.
In symbolic spaces, locations are considered as
symbolic elements (e.g., rooms, buildings, cities),
and object are localized through spatial relations
with other objects (e.g., in the bedroom, near the
red table). For topological spaces, the two basic
types of references are points and zones, while for
symbolic spaces, entities are explicitly connected
through symbolic spatial relations such as containment, accessibility, and so forth. Different from

(Dix et al., 2000), we do not make any distinction

between real and virtual spaces. In our approach,
virtual spaces are managed by applications and
they do not need to be represented at the KUI
level. However, KUI allows the geo-localization
of virtual objects; these are entities of the computing space mapped to the geographical space
(e.g., a geo-tag or a zone).
The detection of particular spatio-temporal relations between Kuidgets can trigger applicationspecific KUI events. There are several spatio-temporal relations that can be modelled. We propose
here that a basic KUI should provide at least two
types of spatio-temporal relations: proximity and
containment. For these relations it is important
to consider their temporal dimension, namely the
start and end time, and the duration of the relation. For instance, if two mobile Kuidgets (e.g.,
two agents) are observed while moving together
along the same path or into the same location, the
application will be notified with a joint motion
event by the KUI manager. Then, the application
might make inferences, and as a result, establish
an application-dependent relation between the
two Kuidgets. For instance, when the two agent
Kuidgets are moving together, the application can
infer (with the help of other contextual information) that they might be friends. Similarly, when
two Kuidgets that are supposed to jointly move
cease to do so, an event could be triggered that
in certain circumstances could denote an unusual
situation. For instance, a car moving somewhere
while its owner moves elsewhere else might denote
that the car has been stolen.
The last dimension for our classification is the
type of motion of mobile Kuidgets. In order to
cope with most possible situations, we consider
both endogenous and exogenous motion. Endogenous motion occurs when objects move without
any change of location. For instance, a rotating
object that remains in the same place counts as
endogenous motion; another example is the sudden
motion sensors embedded in Apple Macintosh10
laptops that can be exploited as an input modal-

1027

Kinetic User Interfaces

ity in games. Exogenous motion represents the

familiar case in which objects are displaced in
the space. However, even if an entity is affected
by exogenous motion, this does not necessarily
mean that the exact change of location is tracked.
For instance, if motion is detected by an accelerometer, the interface can use this information to
trigger an event. This is the case of a Nintendo
WiiTM controller WiiMote that is used to interact
with games without detecting the exact change of
location of players. More precisely, endogenous
motion pertains to the fact that the spatial coordinate system is centred on the moving object, while
exogenous motion is referenced to an external
coordinate system.

The KUI Middleware

The software architecture we propose for KUI is
supposed to be integrated within a larger contextaware middleware for Ubicomp. As in GUI, KUI
should be independent from the underlying OS,
and should enable rich interaction for contextaware Ubicomp systems. For this reason, we do

Figure 1. KUI middleware

1028

not commit ourselves to a specific middleware.

Rather, we focus on a general pluggable software
component, which would allow us to make KUI
available in an arbitrary context-aware middleware (see (Baldauf et al., 2006) for a review).
However, KUI can be used as a standalone system
if the interface between the KUI manager and
motion-aware mobile devices is made through
ad-hoc drivers.
The KUI middleware is made of three layers
of abstraction (as shown in Figure 1). Below, we
provide details about each layer:

The Observation Layer is the lower level.

Its role is to collect kinetic contexts from
location and motion aware devices and from
other hardware sensors.
The KUI-Space Layer is an object-oriented
environment that contains and manages
Kuidgets state and their semantic relationships. Information flow coming from the
observation layer is used to update the state
of Kuidgets. Location information pertaining to Kuidgets is stored in a suitable data

Kinetic User Interfaces

structure, the GeoDB (e.g. a GIS11), that is

also used to store and manage physical space
references of fixed and mobile Kuidgets for
both topological and symbolic spaces.
The Activity Layer manages the context
history, and aggregates KUI events into
higher-level semantic events (i.e., for the
detection of specific interaction patterns)
that are sent to the applications. In this
layer, representations of situations will be
constructed by aggregating kinetic information from Kuidgets and other contextual
information. Moreover, models of activities
will be matched with the spotted situations
in order to determine the occurrence of
anomalous behaviours.

More specifically, in the KUI software components location and motion information are linked

to either fixed Kuidgets or mobile Kuidgets that

are localizable by means of tracking hardware.

The KUI Components

The KUI Middleware can be seen as a user interface server that connects client applications
to motion-aware input devices. In that sense, our
model is similar to the X-Windows toolkit model.
KUI provides an API to applications for subscribing aggregated events from the Activity Layer and
encapsulates motion-aware devices through the
Observation Layer. The main components of the
KUI middleware are detailed in Figure 2.
A number of location and motion-aware devices can be connected to any Kuidget for which
they cooperate in providing its kinetic information. Kuidgets are identified by universally unique
identifiers (UUID), and are linked to entries in

Figure 2. Sketch of the KUI toolkit

1029

Kinetic User Interfaces

the GeoDB that provides both direct localization

and indirect localization through explicit relationships with other Kuidgets not linked to directly
localizable elements. When the Observation Layer
produces fresh kinetic information for a given
Kuidget, a callback is sent to KUI-Space manager
component, which updates the corresponding geographical information in the GeoDB and manages
the relations between active Kuidgets.
The KUI-Space Manager is responsible for
the aggregation and filtering of low-level events
produced by the Observation Layer. Observations
also can be obtained from the hosting contextaware middleware through context widgets (Dey
et al., 2001). Context widgets are abstractions for
different types of context sensors. They hide the
complexity of the sensor communication protocols
and offer a hot-plugging mechanism for dynamically adding and removing sensors to the system.
As for GUIs, the application does not have to be
modified when the pointing system changes, for
instance, from mouse to pen. In our case, Kuidgets
connected to location and motion Widgets do not
know what kind of sensor is actually in use when
they get kinetic information.
The Relation Manager receives location updates from the KUI-space and processes them
according to the (programmable) relation rules.
Relations between Kuidgets are created, deleted or
updated and dispatched to the KUI-space manager,
which then sends them to the upper Activity Layer
through a notification mechanism. When some
pre-defined geographical relationship change
occurs (e.g., an object enters an active zone), the
Relation Manager is responsible of notifying
these events to the Activity Layer or directly to
the subscribing Kuidgets. These events can be
programmed as triggers that check conditions
matched against the GeoDB, while aggregate
events can be subscribed by objects in the upper
Activity Layer. Relations between Kuidgets can
be created either by the Kuidgets internal logic
in response to Kuidget events triggered when

1030

retrieving information from the GeoDB, or explicitly by applications. The latter case is useful
when we logically link agent and artefact Kuidgets
together, allowing one of them to inherit motion
properties from the other one.
The Activity Layer is responsible of aggregating motion information from one or several
Kuidget, as well as dynamic information generated by the Relation Manager. Relations and
Kuidgets status are used as the building blocks
for the recognition of the previously described
kinetic interaction patterns.

Enabling Technology
The KUI middleware can be implemented on
top of a context-aware middleware such as the
Context Toolkit (Dey et al., 2001) and it can be
integrated within any enterprise architecture like
J2EE or .NET. Kinetic-aware devices typically
will be connected through wireless Internet so
that client-server software architecture is apparently justified. Applications exploit localization
infrastructures for indoor and outdoor tracking.
Indoor localization technologies include RFID
antennas, ultrasonic, ultrawide-band, and IR
sensors. For outdoor localization and motion
tracking, GPS offers the most available tracking
solution, which, combined with wireless Internet
communication (e.g., GPRS, EDGE or UMTS) is
nowadays available on commercial mobile phone
and handheld devices. Additionally, we expect to
detect others (more local) motion parameters (such
as acceleration and direction) by using wearable
sensors like accelerometers and digital compasses.
For this point, it is crucial for the Observation
Layer to be capable of dealing with several location and motion tracking technologies at the
same time and of easily associating them with
Kuidgets. The accuracy of different localization
devices and motion sensors is not considered to
be an issue in this discussion that pertains to the
conceptual framework for the development of user

Kinetic User Interfaces

interfaces based on kinetic input. We expect, in

any case, that motion tracking and wearable sensor
technology will rapidly improve, as already has
been the case for other enabling technology like,
for instance, the diffusion of broadband Internet
connection for the Web.

Conclusion
In this chapter, we explored the notion of kineticawareness in Ubicomp user interfaces by means of
the seamless and transparent integration of objects
motion detection in the physical space as a primary
input modality. Kinetic User Interfaces enable the
users of Ubicomp systems to establish an interaction through continuous tracking of kinetic-aware
mobile devices at different spatial scales and by
the acquisition of kinetic input through motionaware embedded sensors. KUI interfaces allow
the seamless integration of contextual (implicit)
and intentional (explicit) interaction through motion. We presented a conceptual framework for
KUI interfaces and a middleware as the basis for
implementing the KUI component in standard
Ubicomp architectures.
Kinetic-awareness in Ubicomp seems to take
over simple location-awareness. Motion-based
interaction is a complementary notion to contextawareness. It is not just a matter of acting while
moving, but acting by moving. Motion is a great
source of information that leverages new dimensions of user experience in Ubicomp systems. As
noted in (Beaudouin-Lafon, 2004), in post-WIMP
user interfaces, it will be necessary to shift towards
a more holistic view of user interaction. Users are
expected to interact through activities rather than
single actions. Moreover, they will try to achieve
higher-level goals through activities, rather than
to accomplish tasks through planned actions. KUI
provides a framework for designing Ubicomp applications with embodied interaction with a special
focus on unobtrusiveness and fluidity.

References
Abowd, G.D., Mynatt, E.D., & Rodden. T. (2002,
January-March). The Human Experience. Pervasive Computing 1(1), 48-57.
Abowd, G.D., & Mynatt, E.D. (2000). Charting
Past, Present, and Future Research in Ubiquitous
Computing. ACM Transactions on Computer-Human Interaction, 7(1), 2958.
Addlesee, M., Curwen, R., Hodges, S., Newman,
J., Steggles, P., Ward, A.., & Hopper, A. (2001).
Implementing a Sentient Computing System.
IEEE Computer, 34(8), 50-56.
Ashbrook, D., Lyons, K., & Clawson, J. (2006).
Capturing Experiences Anytime, Anywhere.
IEEE Pervasive Computing 5(2), 8-11.
Baldauf, M., Dustdar, S., & Rosenberg, F. (2007). A
Survey on Context Aware Systems. International
Journal of Ad Hoc and Ubiquitous Computing, Inderscience Publishers. forthcoming. Pre-print from:
http://www.vitalab.tuwien.ac.at/~florian/papers/
ijahuc2007.pdf
Banavar, G., & Bernstein, A. (2004). Challenges
in Design and Software Infrastructure for Ubiquitous Computing Applications. Communications
of the ACM, 45(12), 92-96.
Barfield, W., & Caudell, T. (Eds.). (2001) Fundamentals of Wearable Computers and Augmented
Reality. LEA Books.
Beaudouin-Lafon, M. (2000). Instrumental Interaction: An Interaction Model for Designing
Post-WIMP User Interfaces. In Proceedings of
ACM Human Factors in Computing Systems,
CHI 2000, La Haye (Pays-Bas), Avril 2000, CHI
Letters 2(1):446-453, ACM Press.
Beaudouin-Lafon, M. (2004). Designing Interaction, not Interfaces. In Costabile, M.F. (Ed.),
In Proceedings of the working conference on

1031

Kinetic User Interfaces

Advanced visual interfaces AVI 2004, (pp. 15-22).

Minneapolis: ACM Press.
Bellotti, V., Back, M., Edwards, K., Grinter, R,
Lopes, C., & Henderson, A. (2002). Making Sense
of Sensing Systems: Five Questions for Researchers and Designers. In Proceedings of CHI 2002
(pp. 415-422), Minneapolis: ACM Press.
Bleecker, J. (2006). Why Things Matter: A Manifesto for Networked Objects Cohabiting with
Pigeons, Arphids and Aibos in the Internet of
Things. Retrieved from http://research.techkwondo.com/files/WhyThingsMatter.pdf
Brumitt, B., Meyers, B. Krumm, J., Kern, A.,.
& Shafer, S. (2000). EasyLiving: Technologies
for Intelligent Environments. In Proceedings of
Second International Symposium on Handheld
and Ubiquitous Computing HUC2K, (pp. 12-29).
Bristol, UK, Springer Verlag.
Chen, G., & Kotz, D. (2000). A Survey of Contex-Aware Mobile Computing Research. (Tech.
Rep. No. TR2000-381), Darthmouth Science
Department.
Cheverst, K., Davies, N., Mitchell, K., Friday, A.,
& Efstratiou, C. (2000). Developing a Contextaware Electronic Tourist Guide: Some Issues and
Experiences. In Proceedings of CHI 2000 (pp.
17-24). The Netherlands.
Davis, M. King, S., Good, N., & Sarvas, R. (2004).
From Context to Content: Leveraging Context
to Infer Media Metadata. In Proceedings of
12th Annual ACM International Conference on
Multimedia MM2004 (pp. 188-195). New York,
ACM Press.
Dey, A.K., & Abowd, G.D. (2000). CyberMinder:
A Context-Aware System for Supporting Reminders. In Proceedings of the 2nd International
Symposium on Handheld and Ubiquitous Computing HUC2K (pp. 25-27), Bristol, UK, Springer
Verlag.

1032

Dey, A.K., Salber, D., & Abowd, G.D. (2001). A

Conceptual Framework and a Toolkit for Supporting the Rapid Prototyping of Context-Aware
Applications (anchor article of the special issue on
Context-Aware Computing). Human-Computer
Interaction Journal, 16(2-4), 97-166.
Dix, A., Rodden, T., Davies, N., Trevor, J., Friday,
A., & Palfreyman, K. (2000). Exploiting space
and location as a design framework for interactive
mobile systems. ACM Transactions on ComputerHuman Interaction, 7(3), 285-321.
Dix, A. (2002a). Managing the Ecology of Interaction. In Pribeanu, C., & Vanderdonckt, J. (Eds.),
Proceedings of Tamodia 2002 - First International
Workshop on Task Models and User Interface
Design (pp. 1-9). Bucharest, Romania: INFOREC
Publishing House, Bucharest.
Dix, A. (2002b). Beyond intention - pushing
boundaries with incidental interaction. In Proceedings of Building Bridges: Interdisciplinary
Context-Sensitive Computing, Glasgow University, Sept. 9th 2002.
Dix, A., Finlay, J., Abowd, G., & Beale, R. (2004).
Human-Computer Interaction (Third Edition).
Prentice Hall. 2004.
Dourish, P. (2001). Where the Action Is: The
Foundations of Embodied Interaction. Cambridge, MIT Press.
Fitzmaurice, G.W. (1996). Graspable User
Interfaces. Unpublished doctoral dissertation,
Department of Computer Science, University of
Toronto.
Gaye, L., Maz, R., & Holmquist, L.E. (2003).
Sonic City: The Urban Environment as a Musical
Interface. In Proceedings of the Conference on
New Interfaces for Musical Expression NIME-03
(pp. 109-115), Montreal, Canada.
Greenfield, A. (2006). Everyware: the dawning
age of ubiquitous computing. New Riders publishers, Berkeley. 2006.

Kinetic User Interfaces

Harter, A., Hopper, A., Steggles, P. Ward, A.,

& Webster, P. (2002). The anatomy of a Context-Aware Application. Wireless Networks, 8,
187-197.
Hightower, J., & Borriello, G. (2001). Location
systems for ubiquitous computing. IEEE Computer 34(8), 57-66.
Holmquist, L., Schmidt, A., & Ullmer, B. (2004).
Tangible interfaces in perspective: Guest editors
introduction. Personal and Ubiquitous Computing 8(5), 291-293.
Hong, D., Chiu, D.K.W., & Shen, V.Y. (2005).
Requirements elicitation for the design of contextaware applications in a ubiquitous environment.
In Proceedings of ICEC05 (pp. 590-596).
International Telecommunication Union (2005).
The Internet of Things: Executive Report. Retrieved from http://www.itu.int/pub/S-POL-IR.IT2005/e. Geneva, 2005
Kleinrock, L. (1997). Nomadic computing. Telecommunication Systems 7(1-3): 5-15.
Kidd, C.K., Orr, R.J., Abowd, G.D. Atkeson, C.G.,
Essa, I.A., MacIntyre, B., Mynatt, E., Starner, T.E.,
& Newstetter, W. (1999, October 1-2)). The Aware
Home: A Living Laboratory for Ubiquitous Computing Research. Paper presented at the Second
International Workshop on Cooperative Buildings
CoBuild99, Pittsburgh, PA (LNCS 1670).
Koile, K., Tollmar, K., Demirdjian, D., Shrobe, H.,
& Darrell, T. (2003). Activity Zones for ContextAware Computing. In Proceedings of Ubicomp
2003, (pp. 90-106).
Mackay, W.E. (1998). Augmented reality: linking real and virtual worlds. A new paradigm for
interacting with computer. In Proceedings of
International Conference on Advanced Visual
Interfaces AVI98, (pp. 13-21).
Meier, R., Harrington, A. Termin, T., & Cahill, V.
(2006). A Spatial Programming Model for Real

Global Smart Space Applications. In Eliassen, F.

& Montresor, Al. (Eds.) Proceedings of the 6th
IFIP International Conference on Distributed
Applications and Interoperable Systems DAIS 06
(pp. 16-31). Bologna, Italy, 2006. Lecture Notes
in Computer Science, Springer-Verlag.
Nardi, B.A. (Ed.) (1996). Context and consciousness: activity theory and human-computer interaction. Cambridge, MA: MIT Press.
Negroponte, N. (1995). Being Digital. Vintage
Editions. 1996.
Norman, D.A. (1999). The Invisible Computer.
Cambridge, MA. MIT Press.
Norrie, M.C., & Signer, B. (2005, January).
Overlaying Paper Maps with Digital Information Services for Tourists, In proceedings of The
12th International Conference on Information
Technology and Travel and Tourism ENTER
2005, Innsbruck, Austria.
Nova, N., Girardin, F., & Dillenbourg, P. (2006,
May 9-12)). The Underwhelming Effects of Automatic Location-Awareness on Collaboration in a
Pervasive Game. In proceedings of The International Conference on the Design of Cooperative
Systems COOP06 (pp. 224-238). Carry-le-Rouet,
Provence, France.
Pallotta, V., Brocco, A., Guinard, D., Bruegger,
P., & De Almeida, P. (2006, June 27). RoamBlog:
Outdoor and Indoor Geoblogging Enhanced with
Contextual Service Provisioning for Mobile Internet Users. In Proceedings of the 1st International
Workshop on Distributed Agent-Based Retrieval
Tools (pp. 103-121). Cagliari, Italy.
Rekimoto, J. (1997). Pick-and-Drop: A Direct
Manipulation Technique for Multiple Computer
Environments. In Proceedings of International
Conference on 10th annual symposium on User
Interface Software and Technology UIST97
(pp. 31-39).

1033

Kinetic User Interfaces

Ricci, A., Viroli, M., & Omicini, A. (2006, April

18-21). Construenda est CArtAgO: Toward an
Infrastructure for Artifacts in MAS. In Proceedings of the European Meeting on Cybernetics and
Systems Research EMCSR06 (vol. 2 pp. 569574).
University of Wien, Vienna, Austria.
Romn, M., Hess, C.K., Cerqueira, R., Ranganathan, A., Campbell, R.H., & Nahrstedt, K.
(2002). GAIA: A Middleware Infrastructure to
Enable Active Spaces. IEEE Pervasive Computing, 1(4), 74-83.
Rukzio, E. (2006). Physical Mobile Interactions:
Mobile Devices as Pervasive Mediators for Interactions with the Real World. Unpublished doctoral
dissertation, University of Munich.
Salkham, A., Cunningham, R., Senart, A., &
Cahill, V. (2006). A Taxonomy of Collaborative
Context-Aware Systems (Tech. Rep. No. TCD-CS2006-30). Dept. of Computer Science, University
of Dublin, Trinity College.
Shafer, S.A.N., Brumitt, B., & Cadiz, J. (2001).
Interaction issues in context-aware intelligent
environments. Human-Computer Interaction,
16(2), 363-378.
Schilit, B.N. (1995). System architecture for
context-aware mobile computing. Unpublished
doctoral dissertation, Columbia University, 1995.
Retrieved from http://citeseer.ist.psu.edu/schilit95system.html
Sousa, J.P., & Garlan, D. (2002). AURA: an
Architectural Framework for User Mobility in
Ubiquitous Computing Environments. In Proceedings of the IEEE Conference on Software
Architecture (pp. 29-43), Montreal.
Sparacino, F. (2005). Intelligent Architecture:
Embedding Spaces with a Mind for Augmented
Interaction. In Costabile, M.F. & Patern, F.
(Eds.): Proceedings of the INTERACT 2005
- Human-Computer Interaction IFIP TC13 In-

1034

ternational Conference (pp. 2-3). LNCS 3585

Springer Verlag.
Spence, M., Driver, C., & Clarke, S. (2005). Sharing Context History in Mobile. Context-Aware
Trails-Based Applications. Paper presented at the
1st international workshop on exploiting context
histories in smart environments ECHISE 2005,
Part of PERVASIVE 2005, Munich, Germany,
2005.
Ullmer, B., & Ishii, H. (2000). Emerging Frameworks for Tangible User Interfaces. IBM Systems
Journal 9(3-4), 915-931.
Want, R., Pering, T., Borriello, G., & Farkas, K.I.
(2002). Disappearing Hardware. IEEE Pervasive
Computing 1(1), 26-35.
Weiser, M. (1991). The Computer for the 21st
Century. Scientific American, 265(3), 94-104.
Weiser, M. (1993). Hot topic: Ubiquitous computing. IEEE Computer, October 1993, 71-72.
Weiser, M. (1994). The world is not a desktop.
ACM Interactions, 1(1), 7-8.
Weiser, M. & Seely Brown, J. (1996). The Coming Age of Calm Technology (Tech. Rep.) Xerox
PARC.
Winograd, T. (2000). Towards a Human-Centered
Interaction Architecture. In Carroll, J. (Ed). Human-Computer Interaction in the New Millennium. Addison-Wesley, Reading, MA.
Winograd, T. (2001). Architectures for context.
Human-Computer Interaction, 16(2,3,4).

Endnotes

2
3

http://www.ubiq.com/hypertext/weiser/
UbiHome.html
http://en.wikipedia.org/wiki/Web_2.0
http://www.socialight.com

Kinetic User Interfaces

4
5

http://www.ubisense.com
WIMP stands for Windows, Icons, Menus,
Popups.
The motion detection can be obtained either
by the mobile device itself (e.g. a GPS-enabled handheld) or by external device or
infrastructure (e.g. a badge tracked by a
sensing space).
This interaction pattern is similar to the
Teleport application (Addlesee et al., 2001),
which allows users wearing ActiveBadges

10
11

to move their desktop environments from

a PC to another.
See (Salkham et al., 2006) for an overview.
We changed the names attributed to these
types of space by (Dix et al., 2000). Our
topological corresponds to their cartesian and our symbolic corresponds to
their topological.
http://www.apple.com/
Geographical Information Systems

This work was previously published in Advances in Ubiquitous Computing: Future Paradigms and Directions, edited by S.
Mostefaoui, Z. Maamar, and G. Giaglis, pp. 201-228, copyright 2008 by IGI Publishing, formerly known as Idea Group Publishing (an imprint of IGI Global).

1035

Section III

Tools and Technologies

This section presents extensive coverage of specific tools and technologies that humans interact with
and react to in their daily lives. These chapters provide an in-depth analysis of devices and tools such
as portable music players, mobile phones, and even blogs. Within these rigorously researched chapters,
readers are presented with countless examples of the technologies that support and encourage societal
development and their resulting impact.

1037

Chapter 3.1

Enabling Technologies for

Pervasive Computing
Joo Henrique Kleinschmidt
State University of Campinas, Brazil
Walter da Cunha Borelli
State University of Campinas, Brazil

INTRODUCTION
Bluetooth (Bluetooth SIG, 2004) and ZigBee
(ZigBee Alliance, 2004) are short-range radio
technologies designed for wireless personal area
networks (WPANs), where the devices must have
low power consumption and require little infrastructure to operate, or none at all. These devices
will enable many applications of mobile and pervasive computing. Bluetooth is the IEEE 802.15.1
(2002) standard and focuses on cable replacement
for consumer devices and voice applications for
medium data rate networks. ZigBee is the IEEE
802.15.4 (2003) standard for low data rate networks
for sensors and control devices. The IEEE defines
only the physical (PHY) and medium access
control (MAC) layers of the standards (Baker,
2005). Both standards have alliances formed by
different companies that develop the specifications for the other layers, such as network, link,
security, and application. Although designed for

different applications, there exists some overlap

among these technologies, which are both competitive and complementary. This article makes a
comparison of the two standards, addressing the
differences, similarities, and coexistence issues.
Some research challenges are described, such
as quality of service, security, energy-saving
methods and protocols for network formation,
routing, and scheduling.

BLUETOOTH
Bluetooth originated in 1994 when Ericsson
started to develop a technology for cable replacement between mobile phones and accessories.
Some years later Ericsson and other companies
joined together to form the Bluetooth Special
Interest Group (SIG), and in 1998 the specification 1.0 was released. The IEEE published the
802.15.1 standard in 2002, adopting the lower

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Enabling Technologies for Pervasive Computing

layers of Bluetooth. The specification Bluetooth

2.0+EDR (Enhanced Data Rate) was released in
2004 (Bluetooth SIG, 2004).
Bluetooth is a low-cost wireless radio technology designed to eliminate wires and cables
between mobile and fixed devices over short distances, allowing the formation of ad hoc networks.
The core protocols of Bluetooth are the radio,
baseband, link manager protocol (LMP), logical
link control and adaptation protocol (L2CAP),
and service discovery protocol (SDP). The radio
specifies details of the air interface, including
frequency, modulation scheme, and transmit
power. The baseband is responsible for connection establishment, addressing, packet format,
timing, and power control. The LMP is used for
link setup between devices and link management,
while the L2CAP adapts upper-layer protocols
to the baseband layer. The SDP is concerned
with device information and services offered by
Bluetooth devices.
Bluetooth operates on the 2.4 GHz ISM (Industrial, Scientific, and Medical) band employing
a frequency-hopping spread spectrum (FHSS)
technique. There are 79 hopping frequencies, each
having a bandwidth of 1MHz. The transmission
rate is up to 1 Mbps in version 1.2 (Bluetooth SIG,
2003) using GFSK (Gaussian frequency shift keying) modulation. In version 2.0+EDR new modes
of 2 Mbps and 3 Mbps were introduced. These
modes use GFSK modulation for the header and

access code of the packets, but employ different

modulation for data. The /4 differential quadrature phase-shift keying (DQPSK) modulation
and 8 differential phase-shift keying (DPSK)
modulation are employed in 2 Mbps and 3 Mbps
mode, respectively.
The communication channel can support both
data (asynchronous) and voice (synchronous)
communications. The synchronous voice channels
are provided using circuit switching with a slot
reservation at fixed intervals. The asynchronous
data channels are provided using packet switching
utilizing a polling access scheme. The channel is
divided in time slots of 625 s. A time-division
duplex (TDD) scheme is used for full-duplex
operation.
Each Bluetooth data packet has three fields: the
access code (72 bits), header (54 bits), and payload.
The access code is used for synchronization and
the header has information such as packet type,
flow control, and acknowledgement. Three error
correction schemes are defined for Bluetooth. A
1/3 rate FEC (forward error correction) is used for
packet header; for data, 2/3 rate FEC and ARQ
(automatic retransmission request). The ARQ
scheme asks for a retransmission of the packet
any time the CRC (cyclic redundancy check)
code detects errors. The 2/3 rate FEC is a (15,10)
Hamming code used in some packets. The ARQ
scheme is not used for synchronous packets such
as voice.

Figure 1. Piconet and scatternet

piconet

slave
master

1038

scatternet

Enabling Technologies for Pervasive Computing

The devices can communicate with each other

forming a network, called piconet, with up to eight
nodes. Within a piconet, one device is assigned as
a master node and the others act as slave nodes.
In the case of multiple slaves, the channel (and
bandwidth) is shared among all the devices in
the piconet. Devices in different piconets can
communicate using a structure called scatternet,
as shown in Figure 1. A slave in one piconet can
participate in another piconet as either a master or
slave. In a scatternet, two or more piconets are not
synchronized in either time or frequency. Each of
the them operates in its own frequency-hopping
channel while the bridge nodes in multiple piconets
participate at the appropriate time via TDD. The
range of Bluetooth devices depends on the class
power, ranging from 10 to 100 meters.

ZIGBEE
ZigBee has its origins in 1998, when Motorola
started to develop a wireless technology for
low-power mesh networking (Baker, 2005). The
IEEE 802.15.4 standard was ratified in May 2003
based on Motorolas proposal. Other companies
joined together and formed the ZigBee Alliance
in 2002. The ZigBee specification was ratified in
December 2004, covering the network, security,
and application layers (Baker, 2005).
ZigBee has been designed for low power
consumption, low cost, and low data rates for
monitoring, control, and sensor applications (Akyildiz, Su, Sankarasubramaniam, & Cayirci, 2002).
The lifetime of the networks are expected to be
of many months to years with non-rechargeable
batteries. The devices operate in unlicensed bands:
2.4 GHz (global), 902-928 MHz (Americas), and
868 MHz (Europe). At 2.4 GHz (16 channels), the
raw data rates can achieve up to 250 Kbps, with
offset-quadrature phase-shift keying (OQPSK)
modulation and direct sequence spread spectrum
(DSSS). The 868 MHz (1 channel) and 915 MHz
(10 channels) bands also use DSSS, but with

binary-phase-shift keying (BPSK) modulation,

achieving data rates up to 20 Kbps and 40 Kbps,
respectively. The expected range is from 10-100m,
depending on environment characteristics.
Each packet, called PHY protocol data unit
(PPDU), contains a preamble sequence, a start of
frame delimiter, the frame length, and a payload
field, the PHY service data unit (PSDU). The 32-bit
preamble is designed for acquisition of symbol and
chip timing. The payload length can vary from 2
to 127 bytes. A frame check sequence improves
the reliability of a packet in difficult conditions.
There are four basic frame types: data, acknowledgement (ACK), MAC command, and beacon.
The ACK frame confirms to the transmitter that
the packet was received without error. The MAC
command frame can be used for remote control
and nodes configuration.
In 802.15.4 two channel-access mechanisms
are implemented, for non-beacon and beacon
network. A non-beacon network uses carriersense medium access with collision avoidance
(CSMA-CA) with positive acknowledgements
for successfully received packets. For a beaconenabled network, a structure called superframe
controls the channel access to guarantee dedicated
bandwidth and low latency. The network coordinator is responsible for set up of the superframe
to transmit beacons at predetermined intervals
and to provide 16 equal-width time slots between
beacons for contention-free channel access in each
time slot (IEEE Std. 802.15.4, 2003; Gutierrez,
Callaway, & Barret, 2003).
A ZigBee network can support up to 65.535
nodes, which can be a network coordinator, a
full function device (FFD), or a reduced function device (RFD). The network coordinator has
general network information and requires the
most memory and computing capabilities of the
three types. An FFD supports all 802.15.4 functions, and an RFD has limited functionalities to
reduce cost and complexity. Two topologies are
supported by the standard: star and peer-to-peer,
as shown in Figure 2. In the star topology, the

1039

Enabling Technologies for Pervasive Computing

Figure 2. ZigBee topologies

star topology

peer-to-peer
topology

PAN
coordinator

reduced function device

full function device

communication is performed between network

devices and a single central controller, called the
PAN coordinator, responsible for managing all the
star functionality. In the peer-to-peer topology,
every network device can communicate with any
other within its range. This topology also contains
a PAN coordinator, which acts as the root of the
network. Peer-to-peer topology allows more complex network formations to be implemented, such
as the cluster-tree. The cluster-tree network is a
special case of a peer-to-peer network in which
most devices are FFDs.

COMPARING ZIGBEE AND BLUETOOTH

Bluetooth and ZigBee have been designed for
different applications, and this section makes
a comparison between some features of both
technologies, such as data rate, power consumption, network latency, complexity, topology, and
scalability (Baker, 2005).
In applications where higher data rates are
required, Bluetooth always has advantages,
especially the 2.0+EDR version (Bluetooth SIG,
2004). While ZigBee mainly supports applications
as periodic or intermittent data, achieving rates
up to 250 Kbps, Bluetooth can support different
traffic types, including not only periodical data,
but also multimedia and voice traffic.

1040

ZigBee devices are able to sleep frequently for

extended periods to conserve power. This feature
works well for energy savings, but increases the
network latency because the node will have to
awake in order to transmit or receive data. In
Bluetooth, the devices do not sleep very often
because the nodes are frequently waiting for new
nodes or to join other networks. Consequently,
data transmission and networks access is fast.
Bluetooth devices in sleep mode have to synchronize with the network for communication, while
in ZigBee this is not necessary.
The Bluetooth protocol stack is relatively
complex when compared to ZigBee. The protocol
stack size for ZigBee is about 28 Kbytes and for
Bluetooth approximately 100 Kbytes (Geer, 2005).
Bluetooth is also more complex if we consider
the number of devices. A piconet can have only
eight nodes, and a scatternet structure has to be
formed to accommodate more nodes (Persson,
Manivannan, & Singhal, 2005; Whitaker, Hodge,
& Chlamtac, 2005). The Bluetooth SIG does not
specify the protocols for scatternet formation.
This task is easier in ZigBee networks, since no
additional protocols have to be used. In terms of
scalability the ZigBee also has some advantages,
because network growth is easier to be implemented with flexible topologies. A Bluetooth
network growth requires a flexible scatternet
formation and routing protocol.
The applications have to consider these
characteristics of both protocols when deciding
which is the most advantageous for that specific implementation. Bluetooth will fit better in
short-range cable replacement, extending LANs
to Bluetooth devices and in industries for communication between fixed equipment and mobile
devices or machine-to-machine communication
(Baker, 2005). ZigBee is most likely to be applied in wireless sensor networks and industries
wireless networks, or any other application where
battery replacement is difficult and the networks
have to live for months to years without human
intervention. Many networks may also implement

Enabling Technologies for Pervasive Computing

Table 1. Comparison between Bluetooth and ZigBee

Characteristic

Bluetooth

ZigBee

Data rate

1 Mbps (version 1.2)

3 Mbps (version 2.0)

20-250 Kbps

Expected battery duration

Days

Years

Operating frequency

2.4 GHz ISM

868 MHz, 902-928 MHz,

2.4 GHz ISM

Security

64 bit, 128 bit

128 bit AES

Network topology

Piconet and scatternet

Star, peer-to-peer, cluster

tree

Protocol stack size

~100 KB

~28 KB

Transmission range

10-100 meters (depending

on power class)

10-100 meters (depending

on the environment)

Network latency (typical)

New device enumeration
12 s
Changing from sleep to active 3 s
Active device channel access 2 ms
Applications

30 ms
15 ms
15 ms

Cable replacement, wireless Remote control, sensors,

USB, handset, headset
battery-operated products

both protocols in complementary roles using the

more suitable characteristic of each for that application. Table 1 shows some features of both
technologies.

RESEARCH CHALLENGES
In the Bluetooth specification there is no information on how a scatternet topology should be
formed, maintained, or operated (Persson et
al., 2005; Whitaker et al., 2005). Two scatternet
topologies that are created from separate approaches can have different characteristics. The
complexity of these tasks significantly increases
when moving from single piconets to multiple
connected piconets.
Some research challenges in Bluetooth scatternets are formation, device status, routing, and
intra and inter-piconet scheduling schemes. Each

device needs to determine its role with respect

to (possibly) multiple piconets, whether master
and/or slave. Whitaker et al. (2005) state:
There is a large degree of freedom in the number
of feasible alternative scatternets, which defines
a significant combinatorial optimization problem.
This is made more difficult by the decentralized
nature of the problem, characterized by a lack of
a centralized entity with global knowledge.
The task of packet routing in a scatternet also
is not so easy because the packet may have to be
transmitted in multiple piconets until it reaches
its destination. In a Bluetooth piconet, the master
controls the channel access. A slave can send a
packet only if it receives a polling packet from
the master. Some slaves may participate in multiple piconets, so they become more important
than others and the scheduling scheme may give

1041

Enabling Technologies for Pervasive Computing

priority for these slaves. The nodes involved in

many piconets can only be active one at a time,
and the scheduling strategy has to consider this
characteristic. As stated in Whitaker et al. (2005),
many factors influence the design of scatternet
protocols, such as distribution of devices, scalability, device differentiation, environmental
dynamism, integration between coordination
issues, and level of centralization. The design of
efficient protocols could make Bluetooth fit for a
wider range of applications.
Although in ZigBee the formation of networks with many nodes is not a great problem,
the management of a network with thousands of
nodes has not been addressed and may be very
difficult (Geer, 2005; Zheng & Lee, 2004). Since
ZigBee specification was released after Bluetooth,
many important issues have not been addressed,
and some distributed protocols will have to be
designed for these networks. Both standards
have security features, including algorithms for
authentication, key exchange, and encryption, but
its efficiency still has to be analyzed in networks
with many nodes.
Other important issue concerning ZigBee and
Bluetooth is the coexistence of both devices, as
they use the same 2.4 GHz band, and channel
allocation conflicts are inevitable between these
WPAN technologies (Chen, Sun, & Gerla, 2006;
Howitt & Gutierrez, 2003). This band is also used
by wireless LANs based on IEEE 802.11 standard
cordless phones and microwave ovens. Interference between near devices may be very common,
so coexistence strategies have to be implemented.
It is important to study the characteristics of each
channel allocation scheme and how each channel allocation scheme interacts with the others.
The discussion on the coexistence issue between
IEEE 802.11 and the IEEE 802.15-based WPAN
technologies has been included in the IEEE
802.15.2 standard.

1042

CONCLUSION
Bluetooth and ZigBee are wireless technologies
that may enable many applications of ubiquitous
and pervasive computing envisioned by Weiser
(1991). Millions of devices are expected to be
equipped with one or both technologies in the
next few years. This work addressed some of
the main features and made some comparisons
between them. Some research challenges were
described. These issues must be properly studied
for the widespread use of ZigBee and Bluetooth
technologies.

REFERENCES
Akyildiz, I. F., Su, W., Sankarasubramaniam,
Y. & Cayirci, E. (2002). A survey on sensor networks. IEEE Communications Magazine, 40(8),
102-114.
Baker, N. (2005). ZigBee and Bluetooth: Strengths
and weakness for industrial applications. IEEE
Computing and Control Engineering Journal,
16(2), 20-25.
Bluetooth SIG. (2004). Specification of the Bluetooth system. Core, version 2.0 + EDR. Retrieved
from http://www.bluetooth.com
Bluetooth SIG. (2003). Specification of the Bluetooth system. Core, version 1.2. Retrieved from
http://www.bluetooth.com
Chen, L., Sun, T., & Gerla, M. (2006). Modeling channel conflict probabilities between IEEE
802.15 based wireless personal area networks.
Proceedings of the IEEE International Conference
on Communications, Istanbul, Turkey.
Geer, D. (2005). Users make a beeline for ZigBee sensor technology. IEEE Computer, 38(12),
16-19.

Enabling Technologies for Pervasive Computing

Gutierrez, J., Callaway, E., & Barret, R. (2003).

IEEE 802.15.4 low-rate wireless personal area
networks: Enabling wireless sensor networks.
Institute of Electrical & Electronics Engineer
(IEEE).
Howitt, I., & Gutierrez, J. (2003). IEEE 802.15
low rate-wireless personal area network coexistence issues. Proceedings of the IEEE Wireless
Communication and Networking Conference (pp.
1481-1486), New Orleans, LA.
IEEE Std. 802.15.1. (2002). IEEE standard for
wireless medium access control (MAC) and
physical layer (PHY) specifications for wireless
personal area networks.
IEEE Std. 802.15.4. (2003). IEEE standard for
wireless medium access control (MAC) and
physical layer (PHY) specifications for low-rate
wireless personal area networks (LR-WPANs).

KEY TERMS
Carrier-Sense Medium Access with Collision Avoidance (CSMA-CA): A network contention protocol that listens to a network in order to
avoid collisions.
Direct Sequence Spread Spectrum (DSSS):
A technique that spreads the data into a large
coded stream that takes the full bandwidth of
the channel.
Frequency Hopping Spread Spectrum
(FHSS): A method of transmitting signals by
rapidly switching a carrier among many frequency
channels using a pseudorandom sequence known
to both transmitter and receiver.
Medium Access Control (MAC): A network
layer that determines who is allowed to access the
physical media at any one time.

Persson, K. E., Manivannan, D., & Singhal, M.

(2005). Bluetooth scatternet formation: Criteria,
models and classification. Elsevier Ad Hoc Networks, 3(6), 777-794.

Modulation: The process in which information signals are impressed on an radio frequency
carrier wave by varying the amplitude, frequency,
or phase.

Weiser, M. (1991). The computer for the 21st century. Scientific American, 265(3), 94-104.

Pervasive Computing: An environment

where devices are always available and communicate with each other over wireless networks
without any interaction required by the user.

Whitaker, R. M., Hodge, L. E., & Chlamtac, I.

(2005). Bluetooth scatternet formation: A survey.
Elsevier Ad Hoc Networks, 3(4), 403-450.
Zheng, J., & Lee, M. J. (2004). Will IEEE 802.15.4
make ubiquitous networking a reality? A discussion on a potential low power, low bit rate standard.
IEEE Communications Magazine, 42, 140-146.
ZigBee Alliance. (2004). ZigBee specification
version 1.0. Retrieved from http://www.zigbee.
com

Scatternet: A group of independent and

non-synchronized piconets that share at least one
common Bluetooth device.
Sensor Network: A network of spatially distributed devices using sensors to monitor conditions at different locations, such as temperature,
sound, pressure, and so forth.
Wireless Personal Area Network (WPAN):
A logical grouping of wireless devices that is
typically limited to a small cell radius.

This work was previously published in Encyclopedia of Mobile Computing and Commerce, edited by D. Taniar, pp. 272-276,
copyright 2007 by Information Science Reference, formerly known as Idea Group Reference (an imprint of IGI Global).

1043

1044

Chapter 3.2

Augmenting Collaboration with

Personalization Services
Christina E. Evangelou
Research Academic Computer Technology Institute, Greece
Manolis Tzagarakis
Research Academic Computer Technology Institute, Greece
Nikos Karousos
Research Academic Computer Technology Institute, Greece
George Gkotsis
Research Academic Computer Technology Institute, Greece
Dora Nousia
Research Academic Computer Technology Institute, Greece

Abstract
Collaboration is considered as an essential element
for effective learning since it enables learners to
better develop their points of view and refine their
knowledge. Our aim being to facilitate communities of practice members as learners, we argue
that collaboration tools should provide personalization features and functionalities in order to fit
the specific individual and community learning
requirements. More specifically, we propose a
framework of services supporting personalization
that being embedded in collaboration tools, can

act as catalysts for individual and community

learning. The proposed set of services has derived
after the careful consideration of a generic learner
profile, developed to formalize human actors in
settings where learning takes place.

Introduction
As organizations start to acknowledge the significance of communities of practice (CoPs) in helping
them meet their business needs and objectives, new
efforts to better facilitate the process of learning
in these communities are constantly emerging

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Augmenting Collaboration with Personalization Services

(Quan-Haase, 2005). CoPs, also referred to as

knowledge networks, is a term commonly used
to identify institutionalized, informal networks
of professionals managing domains of knowledge
(Gongla & Rizzuto, 2001). Such communities are
formed by groups of people who share an interest
in a domain of human endeavour and engage in a
process of collective learning (Wenger, 1998). It is
this very process of collective learning that creates
bonds between them since such communities are
formed by groups of people who are willing to
share and elaborate further on their knowledge,
insights, and experiences (Wenger & Snyder,
2000). Being tied to and performed through
practice, learning is considered of premium
value by practitioners for improving their real
working practices (Steeples & Goodyear, 1999).
Situated learning in particular, that is learning
that normally occurs as the function of an activity, context, and culture, is closely related to the
social interactions in the community context.
Above and beyond learning situated in explicitly defined contexts such as the school classroom,
seminars, or even e-learning approaches, modern
learning theories strongly support the value of
communities and collaborative work as effective
settings for learning (Hoadley & Kilner, 2005).
More specifically, they emphasize on collaborative
learning work that refers to processes, methodologies, and environments, where professionals
engage in a common task and where individuals
depend on and are accountable to each other.
When speaking about collaborative learning,
we espouse the Wengers perspective of learning as a social phenomenon in the context of our
lived experience of participation in the world
(Wenger, 1998). As regards to it, an especially
valued activity involves information exchanges
in which information is constructed through addition, explanation, evaluation, transformation,
or summarising (Gray, 2004; Maudet & Moore,
1999). Discoursing, in particular, is considered
as an essential element for effective learning,
since it enables learners to better develop their

points of view and refine their knowledge. This

is because, in discoursing, participants focus
on the same issues, share their knowledge, and
learn to negotiate conflicting opinions in order to
reach a commonly accepted solution (Veerman,
Andriessen, & Kanselaar, 1998).
Still, it is generally acknowledged that traditional software approaches supporting collaboration are no longer sufficient to support
contemporary communication and collaboration
needs (Moor & Aakhus, 2006). Research findings on the usage of collaboration tools show
that learners are not sufficiently supported in
expressing personal ideas and opinions, nor are
provided with adequate means for the articulation
and sharing of their knowledge. Taking this into
account, our work concerns the design of tools
that enable discoursing to support collaborative
work, emphasis given to aspects, such as the
sharing of knowledge and the building of trust.
We envisage collaboration tools that can promote
learning and encourage creative, parallel, and
lateral thinking during collaboration. Towards
this, we argue that personalized services can be
of great value, as they as they enable the provision
of services tailored according to an individuals
(or communitys when applicable) skills, needs,
and preferences. In this paper, we present a set of
personalization services that has been developed
to address the requirements for efficient and effective collaboration between CoP members who
can act as catalysts for individual and community
learning. Thus, we first performed a comprehensive literature and practice survey of related issues
regarding communities of practice, collaboration, and learning. Then, we developed a generic
Learner Profile model to formalize CoP members
as human actors in settings where learning takes
place. The Learner Profile proposed in this paper
contributes to the proper user modelling required
for the development of virtual environments for
collaboration.
The remainder of this paper is structured
as follows. The following section discusses

1045

Augmenting Collaboration with Personalization Services

user modelling issues and presents the Learner

Profile model of our approach. Next to that, we
provide information about the acquisition of the
data required for the population of the proposed
Learner Profile. Then we present the proposed set
of personalized collaboration services towards
learning and their relation to the proposed Learner
Profile. In the following we discuss implementation issues regarding the embedment of the
proposed set of services to collaboration tools.
We conclude this paper with final remarks and
future work directions.

User models are an essential part of every adaptive

system. In the following, we discuss design issues,
and we present the proposed learner profile model
of our approach. The specification of this model
is oriented to the development of the personalized
services appropriate for learners and/or CoPs.

the user model, this may derive from the user,

the tool, and third-party applications. More
specifically, fields that can be filled up by users
constitute the user-derived information (e.g.,
login name, password, address, etc.). In contrast,
fields that are calculated and filled up by the tool
are machine-derived information (e.g., level of
participation, average response time, etc.). Furthermore, some fields can be filled up both from
the user and machine (preferences, resources, etc.).
In addition, there can be fields that are calculated
by external or third-party tools (or applications).
Although user and machine-derived information
can be easily gathered, third-party tools have to
be aware of the user profile and the communication means with the tool in order to interchange
data of the user profile. For this purpose, the
user profile template is available through an xml
schema definition to third-party requestors via
Web services. In the storage layer, user records
are stored in a relational database and manipulated
through SQL queries.

Design Issues

The Learner Profile

The primary design aims of our approach in modelling users as learners was to achieve extensibility
and adaptability of the user profile as well as the
ability to exchange user information between the
proposed personalized collaboration services and
third-party services. In this context, the proposed
learner profile comprises both computational and
noncomputational information. Computational
information comprises information such as the
name, contact details, education, training, and so
forth, of users, as well as information about the
community they belong to. The noncomputational
information is calculated after the processing
of the users individual behaviour during their
participation in system activities. This type of
information comprises fields that can be defined
during run-time, whenever a new requirement
for a new kind of user information is raised. As
regards the source of the information stored in

In successful learning organizations, individual

learning is continuous, knowledge is shared, and
the organizational culture must support learning
(Gephart, Marsick, Van Buren, & Spiro,1996;
Marsick & Watkins, 1999;). Learning entities
transform themselves by acquiring new knowledge, skills, or behaviours in their everyday
work, through study, instruction, or experience.
That is why software tools facilitating working
practices for individuals, communities, or organizations should also be conceived and designed
as environments where learning takes place. In
our approach, collaboration tools should satisfy
the community members needs to construct
and refine their ideas, opinions, and thoughts in
meaningful ways, in order to successfully assist
individual and community learning. At the same
time, individual standpoints should be articulated
in such a way that can be proven useful for the rest

User modelling

1046

Augmenting Collaboration with Personalization Services

of the communitys members. In addition to that,

support should be offered for the development of
learning skills, such as the interaction with other
actors, as well as growth of the learners autonomy
and self-direction. Moreover, identification of CoP
members individual characteristics, as well as
the culture, norms, and incentive schemes of the
community should be appropriately handled.
Research findings about learners modelling
prove that due to the complexity of human actors
and the diversity regarding the learning context,
the development of a commonly accepted learner
profile is a highly complex task (Dolog & Schfer,
2005). For instance, the learner model proposed
in Chen and Mizoguchi (1999) depicts a learner
as a concept hierarchy, but it does not refer to issues such as the learning object, or the learners
interactions with their environment and other
people. However, it provides interesting information about a learners cognitive characteristics,
and it provides a representation of knowledge
assessment issues. Another related approach, the
PAPI Learner conceptual model, comprises
preferences, performance, portfolio, and other
types of information (PAPI, 2000). Yet, this model
is too generic, as its primary aim is to be portable
in order to fit a wide range of applications, and it
does not provide any information about a learners
profile dynamic aspects. The IMS Learner Infor-

mation Package specification (IMS LIP, 2001) is

a useful collection of information that addresses
the interoperability of Internet-based Learner
Information Systems with other systems that
support the Internet learning environment.
After the careful consideration of the above, we
developed a generic Learner Profile (see Figure 1)
that can be employed for the representation of both
individuals and communities as learners (Vidou,
Dieng-Kuntz, El Ghali, Evangelou, Giboin, Jacquemart, & Tifous, 2006). The proposed model
can be used for developing customized services
for both individual and group learners. More
specifically, the proposed Learner Profile consists
of two types of information, namely static information and dynamic information in compliance
with the computational and noncomputational
data presented. Static information is considered
as domain independent in our approach. The
Learner Profile dynamic information elements
were chosen to reflect ones individual behaviour
during his participation in a specific CoPs collaboration activities. Thus, all four dynamic elements, that is, preferences, relations, competences,
and experience are to be implicitly or explicitly
defined through the learners interaction with a tool
supporting collaboration. Preferences regarding
the use of resources and services provided by the
tool, as well as relations among individuals, CoPs,

Figure 1. The proposed Learner Profile

Learner Profile
Static
Information

Dynamic Information

Individual
Information

Preferences

Relations

Community
Information

Competences

Experience

Domain independent

Domain specific

1047

Augmenting Collaboration with Personalization Services

and learning items (e.g., argument, URL, or document) can reveal the learners different personality
types and learning styles. Competences refer to
cognitive characteristics such as the creativity,
reciprocity, and social skills. Experience reflects
learners familiarity and know-how regarding a
specific domain. It should be noted that all dynamic elements of the proposed Learner Profile
can be of assistance towards learning. Nevertheless, the domain of the issue under consideration
is a decisive factor. Thus, dynamic aspects of a
learners profile are treated as domain specific in
our approach.

Acquiring the Learner

Profile data
In order to enable the operation of personalized
collaboration services, the Learner Profile has to
be populated with the appropriate data. Such data
can be acquired in two ways: explicitly from the
users preferences, and implicitly based on the
users behaviour while using the system. Static
information of the Learner Profile is explicitly
provided by the user as a required initialization
step of the registration procedure. While such
information is usually provided when registering
to the system, users should be able to edit this set
of profile information at any time. Such explicit
data acquisition constitutes a subjective way of
profiling, since it depends on the statements made
by the user (e.g., experience level, competences,

etc.). Their subjective nature may influence personalization services in an unpredictable way
(e.g., suggesting to a novice user a document that
requires advanced domain knowledge because
the user misjudged his experience or competence
level). To cope with such issues, we are currently
in the process of designing methods that assess
explicitly stated profile data, based on the users
behaviour. We refer to these ways as implicit or
behaviour-based data acquisition. In general, the
aim of implicit or behaviour-based data acquisition
is to assess experience, domains, competences of
an individual user based on his behaviour. Implicit
data acquisition utilizes the users actions and
interactions, and attempts to extract information
that can permit assessing or augmenting a user
profile data.
A special part of the systems architecture
should be dedicated to support implicit data acquisition and interpretation. It consists of a number
of modules, each of which is responsible for a
particular task (see Figure 2). More specifically, the
User Action and Tracking module is responsible
for observing user actions and recording them in
a special repository of the infrastructure called
the Action and Event Store. The Action and Event
Store only maintains all actions and events that
are useful for implicit user action analysis and
does not interpret them in any way. Analysis
and interpretation of the gathered data as well
as triggering of the appropriate computations
(i.e., system reactions) is the main responsibility
of the Action Interpretation

Engine. The Action

Figure 2. Data acquisition and interpretation structure

User Action Tracking Module

Actions/ Event Store

SystemData Base

1048

Interpretation Engine

Rule Store

Augmenting Collaboration with Personalization Services

Interpretation Engine analyses the available information in the actions and event store and triggers
computations that either update accordingly the
user profile or execute a particular action. The
interpretation engine can be configured using
rules that are also stored within the infrastructure,
making the interpretation engine rule based. A rule
essentially specifies under which circumstances
(i.e., the events and actions of a particular user in
the store) an action is triggered. The rule-based
nature of the interpretation engine makes the
engine itself extensible so that even more cases
of implicit data acquisition and interpretation are
able to be supported.
Based on the explicit or implicit data, explicit or
implicit adaptation mechanisms can be supported
within the collaboration tool. Explicit adaptation
mechanisms refer to approaches where the tool
adapts its services based on the explicitly stated
characteristics or preferences of the user. Users
are usually aware of explicit adaptations since they
themselves triggered the initiation and presence of
the respective services. On the other hand, implicit
adaptation mechanisms refer to approaches that
adapt the systems services to the user, based on
his/her actions within it. Such mechanisms work
in the background; Users are usually unaware
of the origin of these services since they did not
explicitly initiate their activation and, thus, do not
perceive their operation. Implicit personalization
mechanisms are automatically triggered by the
system utilizing implicit or behaviour-based data
in the proposed Learner Profile.
In order to enable the foreseen functionalities
(such as dynamic update of user information,
adaptation of the tool according to the user needs,
etc.), the most important actions of the entire set of
users actions should be tracked down. As regards
the User Action Tracking Mechanism, the recorded
data about user actions contain information about
who did the action, when, what type of action was
executed, and what objects were affected by the
action. In this way, it will be possible for the system
to give valuable feedback to other mechanisms so

as to be able to both examine and calculate dynamic

user characteristics. Furthermore, a variety of
statistical reports that cover both the overall and
the specific views of usage of the system should
also be produced.
Furthermore, a rule-based approach has been
chosen so as to facilitate incorporation of new
rules once they are observed, or modification of
existing ones if they prove to be too restrictive
or even harmful. More specifically, we propose
the development of a set of rules that deal with
resource access, as access to resources are logged,
and a number of rules operate on the logged data
to provide additional information to resources
and/or user profiles. These can be based on the
frequency of access, as well as the competence
and experience levels of users (e.g., a document
that is frequently accessed by novice users should
augment the documents metadata with elements
that mirror this fact, so that this document can
be recommended to any novice user entering
a discussion). A second set of rules observing
discussion contribution could control how user
behaviour in the context of discussions will affect
the users competence and experience (e.g., users
that actively and frequently participate can be
assigned with a high experience level). Another
useful indicator associated to the proposed learner
profile is the reasoning about how a competence
level of a particular user changes in time. This
may provide useful insights about the learning
capabilities of the particular user and the usefulness of the system.

Personalized services for

collaboration tools
One of the major challenges in developing software is that all users are different, in that they
vary in terms of intelligence, knowledge, training, experience, personality, and cognitive style.
Therefore, collaboration tools should provide a
set of personalized services, with the emphasis

1049

Augmenting Collaboration with Personalization Services

given to the individuals learning

styles. In the
following we present a set of services employed for
enhancing software tools supporting collaboration
towards learning.
The proposed set of services
has resulted out of a thorough investigation of
the related literature, existing case studies that
consider diverse aspects of learning within communities, as well as a transversal analysis of a set
of interviews with real CoP members engaged in
various domains of practice.

Awareness
According to the findings of our research, CoPs'
members consider system awareness services
as the most helpful ones for collaboration tools.
Participation awareness provides information
about CoP members, online members as well as
the discourse moves of individual CoP members.
Users will be able to see which user is online, how
the space changed by a particular member, and so
forth. Social awareness provides information on
how members are related to other members in the
CoP, and includes statistics about how and how
many times members within a CoP communicate
with each other and social networks representing
the community. Based on the data populated in
the Learner Profile, personalized services can
provide the proper set of notification actions for
the provision of helpful personalized information
about system events to CoP members. For instance,
a collaboration tool could alert users about the
entrance of another user to the system, or about
new content insertion into the system.
In order to enable this personalized awareness, terms such as related or interesting
that define a relation between the user and the
content should be determined by the user himself,
or automatically by the system through the manipulation of some characteristics from the user
profile. Furthermore, system awareness can play
an important role to assist the familiarization on
the new learners of the system. By both informing
the CoP moderator about the entrance of a new

1050

member and proposing some starting guidelines

to the incomer, this service can assist the learning of the way of participation within a CoP. On
the other hand, the awareness can provide the
moderators with the activity monitoring service
that helps the moderator to better understand
and manage the whole CoPs procedures. That,
in turn, contributes to the process of learning the
CoPs moderator role.
Awareness services can also be of use towards
the self-evaluation of the participation of a community member, providing him with valuable
feedback about his overall contribution to the
community and assisting him in collaborative
learning as well as in self-reflecting. Using statistic reports populated according to the Learner
Profile, such services can measure the level of
the members contribution to the collaboration
procedure. More specifically, these kinds of services can provide with reports about the actual
usage of the resources posted by a member, the
citations of their resources, or the actual impact
of posts to the overall process. In this way, one
can be aware of the overall impression that other
members have about his participation.

Allocation of Resources
Allocation of resources is another service that
being personalized in collaboration tools can
facilitate learning activities, especially for autonomous learners. As regards to searching for
instance, a Learners Profile can provide useful
information to rank search resources according to a number of factors, such as the learners
preferences, or even his/her competence and
experience level. In this way, the system will be
able to adapt to an individual users needs. Moreover, the information about the users domains of
interest will provide additional information with
which a search can be better contextualized, thus
leading to more relevant results. Furthermore,
reasoning mechanisms could be employed for
providing the necessary filtering features for

Augmenting Collaboration with Personalization Services

capturing and reusing the knowledge shared in

past collaboration activities. In this vein, filtering and recommendation of content services can
further support learning. For instance, some of
the attached documents of posted positions that
contribute to the strengthening of an argument
should be suggested for view to the users according to their Learner Profile. Furthermore,
a document library could recommend some
documents that are related to a specific learner
(e.g., experienced learners recommendations
or popular documents). Thus, members will be
able to extend their knowledge through explicit
learning of associated content.
Services for classifying other learners according to their domain of expertise can also assist
learning in the community. Such services enable
the community members to request for suggestion,
find and communicate with their coworkers in a
knowledgeable way. Furthermore, if coinciding
with a communitys norms and wills, such services
could also be used for the assignment of weights
regarding the weight of a members arguments.
In addition, services that keep tracking of the
members activity contribute to the procedure
of learning by example, in which a member can
learn during watching another ones practice in
collaborative activities.

Visualization
It has been widely argued that visualization of
collaboration conducted by a group of people
working collaboratively towards solving a common problem can facilitate the overall process
in many ways, such as in explicating and sharing individual representations of the problem,
in maintaining focus on the overall process, as
well as in maintaining consistency and in increasing plausibility and accuracy (Evangelou,
Karacapilidis, & Tzagarakis , 2006; Kirschner,
Buckingham-Shum, & Carr, 2003;). Personalized
representation of the associated processes, such as
the process of discoursing or knowledge sharing,

is an essential feature for tools providing effective

environments for learning. Furthermore, personalized visualization of context should provide
learners with a working environment that fits to
their preferred visualization style. System personalization includes alterations in colours, fonts,
and text effects, enabling and disabling pieces of
information in the working panel, predefinition
of system responses in user actions and so forth.
In this direction, taxonomies and classification
schemes should be employed, wherever possible,
as a means for guiding users cognition. In any
case, it should be noted that there is no panacea
for the design of user-friendly interfaces; the related practices should be interpreted, refined, and
exploited according to the needs of the different
types of learners involved in the particular environment. Appropriate navigation and help tools
should also be provided for users with diverse
expertise. Adaptive User Interfaces should adapt
themselves to the learner by reasoning about the
user, based on his Learner Profile.

Building Trust
Privacy policies and access control services are
a critical requirement for the employment of all
these services, as well as for the building of trust
between the CoP members and the software application. These should be provided in order to satisfy
the learner/users need to know what information
about them is recorded, for what purposes, how
long this information will be kept, and if this
information is revealed to other people. Furthermore, the security assurance, while establishing
connections between users and services, or while
accessing stored information, should be taken into
consideration as well. Towards this end, two major
techniques are broadly used to provide denial of
access to data, that is, anonymity and encryption.
Anonymity cuts the relation between the particular
user and the information about him/her, while
information encryption provides protection of
the exchanged personal data. In our approach, we

1051

Augmenting Collaboration with Personalization Services

employed the Platform for Privacy Preferences

Project (P3P) approach, a W3C recommendation
that supports the description of privacy policies
in a standardized XML-based form, which can
be automatically retrieved and interpreted by the
user client (Cranor, Langheinrich, Marchiori,
Presler-Marshall, & Reagle, 2002).

Implementation issues
According to current trends in developing Webbased tools, for reasons such as the reusability of
components and agility of services, our approach
builds on top of a service-oriented environment.
In order to exploit advantages enabled by the
Service Oriented Architecture (SOA) design
paradigm, the proposed set of services should be
based on Web service architecture so as to enable
the reusability of the implemented modules, as
well as the integration or the interoperation with
other services (from external systems). An overall
design for the enhancement of tools supporting
collaboration with personalized functionality
towards learning is depicted in Figure 3. In this
approach, we sketch a generic architecture design
in which a Learner Profile Service is the basis for

the storage and the provision of each learners

characteristics to a set of proposed services that
contribute to the systems personalization. In
order to support extensibility, the learning profile
service can be dynamically augmented with new
learners characteristics during run-time. Furthermore, targeting to the openness of the service, the
service can provide the learner profile schema in
the form of XML Schema Definition (XSD) in the
service requestors. Considering the set of proposed
services as nonexhaustive, our approach is open
for the addition of new personalized services (see
Figure 3, block New Service) and can use the
Simple Object Access Protocol (SOAP) for both
internal and external communication.

Conclusion
In this paper, we presented a set of services enhancing CoPs interactions and collaborative work
based on a generic Learner Profile model. Our
approach concerns an alternative form of online
learning with different forms of interaction, and
a new way of promoting community building.
Its purpose is to aid researchers and developers
in the development of personalized collaboration

Figure 3. The proposed services

Learner Information Acquisition Service
Awareness Service

Learner Profile Service

Personalized Search
Personalized Presentation
Content Filtering and Recommendation
Related Learners Activity
Participation Evaluation
Learner Search by Expertise
Privacy Policies and Access Control
New Services

1052

Augmenting Collaboration with Personalization Services

systems, that is, tools that adapt their structure and

services to the individual users characteristics and
social behaviour. Our main goal being to support
individual and community learning, the proposed
set of services is based on personalized features
and functionalities. We argue that it can further
support learning, as well as the achievement of
learning objectives, as it can assist CoP members
in the development of learning skills such as the
interaction with other actors, growth of their
autonomy, and self-direction. Nevertheless, in
order to be creatively adapted in CoPs everyday
practices, the proposed services must fit into the
specific culture, norms, and incentive schemes
of the community. Our future work directions
concern the appropriate handling of these issues
as well as the full development of the proposed
set of personalization services and its evaluation
in diverse CoPs.

Acknowledgment
Research carried out in the context of this paper
has been partially funded by the EU PALETTE
(Pedagogically Sustained Adaptive Learning
through the Exploitation of Tacit and Explicit
Knowledge) Integrated Project (IST FP6-2004,
Contract Number 028038). Also, the authors
thank the anonymous referees for their helpful
comments and suggestions on the previous versions of this paper.

References
Chen, W., & Mizoguchi, R. (1999). Communication content ontology for learner model agent in
multi-agent architecture. In Prof. AIED99 Workshop on Ontologies for Intelligent educational
Systems. Retrieved from http://www.ei.sanken.
osaka-u.ac.jp/aied99/a-papers/W-Chen.pdf
Cranor, L., Langheinrich, M., Marchiori, M.,
Presler-Marshall, M., & Reagle, J. (2002). The

platform for privacy preferences 1.0 (P3P1.0)

Specification. World WideWeb Consortium
(W3C). Retrieved from http://www. w3.org/TR/
P3P/.
Dolog, P., & Schfer, M. (2005, July 24-30).
Learner modeling on the semantic Web? Workshop on Personalisation on the Semantic Web
PerSWeb05, Edinburgh, UK.
Evangelou, C. E., Karacapilidis, N., & Tzagarakis
M. (2006). On the development of knowledge
management services for collaborative decision
making. Journal of Computers, 1(6), 19-28.
Gephart, M., Marsick, V., Van Buren, M., & Spiro,
M., (1996). Learning organizations come alive.
Training & Development, 50(12), 34-45.
Gongla, P., & Rizzuto, C. R. (2001). Evolving
communities of practice: IBM Global Services
experience. IBM Systems Journal, 40(4), 842862.
Gray, B. (2004). Informal learning in an online
community of practice. Journal of Distance
Education, 19(1), 20-35.
Hoadley, C. M., & Kilner, P. G. (2005). Using technology to transform communities of practice into
knowledge-building communities. SIGGROUP
Bulletin, 25(1), 31-40.
IMS LIP. (2001). IMS learner information package
specification. The Global Learning Consortium.
Retrieved from http://www.imsglobal.org/profiles/index.html
Kirschner, P., Buckingham-Shum, S., & Carr, C.
(2003). Visualizing argumentation: Software tools
for collaborative and educational sense-making.
London: Springer Verlag.
Marsick, V. J., & Watkins, K. E. (1999). Facilitating
learning organizations: Making learning count.
Aldershot, U.K.; Brookfield, VT: Gower.
Maudet, N., & Moore, D. J. (1999). Dialogue
games for computer supported collaborative ar-

1053

Augmenting Collaboration with Personalization Services

gumentation. In Proceedings of the 1st Workshop

on Computer Supported Collaborative Argumentation (CSCA99).
Moor, A., & Aakhus, M. (2006). Argumentation
support: from technologies to tools. Communications ACM, 49(3), 93-98.
PAPI. (2000). Draft standard for learning technologyPublic and Private Information (PAPI) for
Learners (PAPI Learner). IEEE P1484.2/D7, 200011-28. Retrieved from http://edutool.com/papi
Quan-Haase, A. (2005). Trends in online learning
communities. SIGGROUP Bulletin, 25(1), 1-6.
Steeples, C., & Goodyear, P. (1999) Enabling
professional learning in distributed communities
of practice: Descriptors for multimedia objects.
Journal of Network and Computer Applications,
22, 133-145

Veerman, A. L., Andriessen, J. E., & Kanselaar,

G. (1998). Learning through computer-mediated
collaborative argumentation. Retrieved from
http://eduweb.fsw.ruu.nl/arja/PhD2.html
Vidou, G., Dieng-Kuntz, R., El Ghali, A., Evangelou, C. E., Giboin, A., Jacquemart, S., & Tifous,
A. (2006, Nov. 30-Dec. 1). Towards an ontology
for knowledge management in communities of
practice. In Proceedings of the 6th International
Conference on Practical Aspects of Knowledge
Management, (PAKM06). Vienna, Austria
Wenger, E. (1998). Communities of practice:
Learning, meaning and identity. Cambridge
University Press.
Wenger, E., & Snyder, W. (2000). Communities
of practice: The organizational frontier. Harvard
Business Review, 78, 139-145.

This work was previously published in InternatIonal Journal of Web-based LearnIng and TeachIng TechnologIes, Vol. 2, Issue
3, edited by L. Esnault, pp. 77-89, copyright 2007 by IGI Publishing, formerly known as Idea Group Publishing (an imprint
of IGI Global).

1054

1055

Chapter 3.3

Culture and Technology:

A Mutual-Shaping Approach
Thomas Herdin
University of Salzburg, Austria
Wolfgang Hofkirchner
University of Salzburg, Austria
Ursula Maier-Rabler
University of Salzburg, Austria

Abstract
The aim of this chapter is to discuss the mutual
influence between culture and technology on a
broad inter- and transcultural level. Especially,
how does information culture shape the meaning
of information, communication, and knowledge,
and consequently, the design, spread, and usage
of ICTs in certain societies? Vice versa, we are
interested in the ways in which the spread and
usage of ICTs affect the predominating culture.
We aim for a model that incorporates cultural as
well as technological factors in order to provide a
basis for future ICT research that goes beyond both
technological determinism and social constructivism. We believe that new technologies indeed can
contribute to more justice in the world in terms of

access to knowledge and wealth, if sociocultural

factors are taken into account more seriously.
Current developments in the context of the UN
World Summit on the Information Society raise
awareness in this direction. At the same time, we
are well aware that the loose notion and imprecise
definition of the concept of culture allows for the
exploitation of the term in empty political and
techno-economical policies. Culture degenerates
to an indispensable buzzword in the current ICT
debate. This chapter is an attempt to introduce
the concept of culture into the socioresponsible
ICT research on equal terms with technology,
economy, and society.

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Culture and Technology

THE RELATIONSHIP BETWEEN

CULTURE AND TECHNOLOGY
How can technology be defined? Technology
often is considered a means to a particular end,
the means being artificially created, not natural,
and something that is not directly necessary for
the individual or the end user; it serves, rather, to
fulfill the need to produce something that is later
to be consumed. However, we use the term in a
broader sense. We regard technology as being
more than just the sum of such artefacts, which are
merely the crystallized, concrete manifestations
of human behavioral patterns. A method is the
how, the way in which a goal is reached and which
involves the use of means. A means is a medium
in that it mediates between the starting point and
the desired result, regardless of what sort of action
is involved. Thus, one could speak of social technology (e.g., psychotherapy) as a technology and
not merely of technology as something used for
(material) production in a society. So, technology
also includes the know-how involved in the use
and application of the artefacts. In short, technology embraces the ways and means of acting in
pursuit of a goal (Hofkirchner, 1999).
How can culture be defined? Using the same
analogy for technology, it could be understood to
be an equally artefact-based concept, which is not
a means to an end but rather an end in itself. That
is to say, it is not in itself an essential of life, but
rather something that represents a human desire
(i.e., what makes humans distinct from other living
beings). Here, too, there is a notion that culture is
not only the result of a process but also this very
process as it moves toward the goal; that is to say,
culture is a characteristic of goal-oriented actions
(i.e., the striving toward goals as well as the goals
themselves) (Hofkirchner, 1999). It is this notion
of culture that we refer to in this chapter.
Are there imaginable connections between
culture and technology? The two ideal-typical
extreme positions are well-known, each making

1056

a single direction of determination (Hofkirchner,

1999).
The first position can be referred to as technological determinism, which postulates the total,
or at least dominating, influence of technology
on culture. Technology is supposed to develop
more or less on its own, pushing social development along as it goes. This may be interpreted
positively or negatively. An uncritical opinion
of Marxist origin saw social advancement as an
inevitable result of technical achievements, just
as the ideology of the bourgeoisie justified the
progress of the technically possible as socially
desirable. This view is opposed entirely by fundamentalists who hold technological development
responsible for the loss of important values in
society. Neither philosophy accepts the possibility
of technological development being influenced in
any way. Both ignore the fact that there would be
no such development if multinational corporations
and national governments were to stop investing
in research and development; if there were no
economic, military, or political interests to divert
their resources into these areas; and if there were
no values, morals, or norms that underlay these
economic, military, or political interests. The fact
that on a micro-level there are countless thousands
of engineers constantly involved in technology
design, and that on a macro-level managers and
politicians decide which technological options
are realized, supports the second theorysocial
constructivismthat technology is constructed
deliberately to be a part of society. According
to this view, the interests of those groups that
dominate the genesis of technology finally are
embodied in the technology, which in itself cannot be neutral. Here again, both a critical and an
approving variant may be distinguished. While the
one bemoans the inability of existing technology to
pursue ethically justified, socially acceptable, and
peaceful and environmentally sound objectives,
the other sees the existing economic, democratic,
and human rights structures as the best guarantee

Culture and Technology

of developing optimal technological options. Both

versions neglect the inherent dynamism within
technological development.
Do the two theoriestechnological determinism and social constructivismtogether give a
realistic view of the relationship between technology and culture? This would mean that two
equally matched factorsthe technical and the
culturalwould not be complete without the other.
Furthermore, one also might break away from
strict determinism and grant each side a measure
of independence, thus denying that one side totally
dominates the other. But would we then have a
workable proposition to discuss, or would we be
reduced to the assumption that one factor partly
influences the other but is itself partly influenced
by its counterpart? This is a superficial answer.
Is it not rather the case that the actions we are
talking about, whose dependence on mediating
factors we want to stress if we are talking about
technology and whose immersion in value judgments we wish to highlight when we are discussing
culture, not only have an individual character, but
rather, through the availability of technological
methods and cultural values on the part of society, acquire a deeply societal nature? The use of
technology makes every action no longer unique
to any individual person. Technology is based on
cooperation, be it in the application of special
methods, the implementation of these in specific
social areas, their invention and development, or
in any situation in which the skills and knowledge
of other members of society are required. The
same holds true for convictions, value judgments,
instructions, standards, behavioral patterns, and
the like. These are just as much a part of the
context of life of the individual, and they promote
certain technological methods but discourage
others. Technology makes every technologically
mediated action into a socially determined one,
and its use is a human characteristic. Technological development is part of cultural development;
this means that technology is part of culture, and
so their relationship to each other is one of part

and whole. Culture is the all-embracing factor

in this context.
In each part-whole relationship, the parts are
the necessary preconditions for the emergence
of the whole but are not the sufficient condition
for the complete determination of the result. The
whole arises from the parts but then exerts control
over them in the form of downward causation;
the parts are no longer independent of each other
as separate entities but are dominated by the
whole. The relation of part and whole in regard
to technology and culture is, therefore, as follows:
technology has the meaning, the purpose, and
the task of functioning as means and method for
solving social problems. Social interests, cultural
values, norms, and morals are thus in the origin
and manifestation of technology in its invention,
diffusion, and application in the entire process
of its development, as its reason for existence.
This, however, is insufficient to enslave technology completely. Technology is ambivalent;
sometimes it appears to resist our intentions by
wholly or partly failing to do what is wanted of
it, and other times it not only fulfills our expectations but goes on to do other useful tasks that
originally had not been anticipated. Technology
represents potential for the realization of social
goals. These technologically realizable goals may
correspond to pre-existing goals within society;
the practical attainment of these by technological
means, however, may cause them to change, at
least slightly. It is, of course, also possible that the
intended goals may differ from those that can be
reached with technological support. In this case,
new technology may be developed in order to
meet the requirements, or the requirements may,
as it were, be adapted to fit the reality of what is
technically possible. Realizable goals, therefore,
do not always exist at the start of the process but
may be discovered as options made available by
technology. Whether society decides to pursue
these goals on the grounds that they are possible
is no longer a question of technology but rather of
social decision making (Hofkirchner, 1994).

1057

Culture and Technology

To conclude, we consider the relationship of

technology and culture to be dialectic. A relationship is usually called dialectic if, first, the sides
of the relation are opposed to each other; second,
both sides depend on each other; and third, they
form a relation that is asymmetrical (Hofkirchner,
2004). A part-whole relationship is dialectic since
part and whole represents opposites, the whole
depends on the parts as well as the parts on the
whole, and parts and whole build up a hierarchy
in which the different levels cannot be replaced
by each other.
Considering this notion of the relationship
between technology and culture on a broad and
general level, the following section attempts to
add further thought in the context of information and communication technologies (ICTs) and
culture. The specific meaning of information and
communication for different societies, which is
predominantly the result of a special culture, determines the meaning and, therefore, the spread
and usage of ICTs. Vice versa, ICTs have been
developed and will be developed in the future in
certain (information and communication) cultures, which leads to the functions and practices
of use we are facing when we implement ICTs.

INFORMATION AND
COMMUNICATION CULTURES
When referring to information and communication cultures, we address the basic significance
of having access to information and knowledge
and the practices of communication and cooperation in a specific society. The most important
consideration involves the relationship between
those who have access to information that has a
profound effect on the distribution of power of
control over flows of information within society.
It is assumed that within societies with a strong
hierarchical structure, the flow and dissemination of public information is restricted to just a

1058

few people, while in more liberal societies, there

is a far broader basis for direct access to public
information. Furthermore, more hierarchically
structured societies are less likely to be expected
to adapt to the Internet than liberal societies with
a flatter hierarchy (Maier-Rabler, 1995, 2002).
The general attitude toward access to information, toward transparency of structures and
processes, and toward empowerment and freedom
of expression pertaining to a specific society or
state is deeply rooted in traditions and practices of
social and cultural conditions. The cultural-social
framework of a society is formed mainly by the
political-social system, by the legislative system,
and particularly by the predominating ethic and
religious values. As a result of these diverse
dimensions, a continuum between the poles of
information-friendly vs. information-restrictive
cultures emerges (Maier-Rabler & Sutterltti,
1992; Maier-Rabler, 1995).
Information-friendly societies foster the development of knowledge throughout all groups of a
society by providing equal and universal access to
all available public information. In informationfriendly societies, people have access to public
information, freedom of speech is guaranteed to
individuals and institutions, and the concept of
universal access is understood as the equitable
and affordable access to information infrastructure and to information and knowledge essential
to collective and individual human development
for all citizens. In information-friendly societies,
curiosity is encouraged by education systems, and
skills for information retrieval are taught rather
than just being fed information. Questions count
more than answers, and students are encouraged
to research instead of memorize given information
(Maier-Rabler, 2002).
The political system in information-friendly
cultures is likely to be in a form of communicative democracy within a developed system of
civil society. Direct democratic participation is a
living practice enjoyed by all social groups. The

Culture and Technology

legal system is likely to be an information-rich,

case-based system in which access to information
is vital for legal practice.
The economic system in an informationfriendly environment strongly depends on access
to information and its dissemination to shareholders and customers. Wealth and success are highly
valued, and information on turnovers, revenues,
and profits are publicly available. Informationfriendly societies experience a great expansion
in their limitations, especially through the new
information and communication technologies. At
the same time, it has become clear that without
a capable citizenship and without capable institutions, unintended and even unwanted consequences take place. What is more, the current
crises of the stock markets have been due to access
to information that neither has been audited nor
controlled. On a political level, we face a threat
to well-established forms of representative democratic systems through populist political trends.
New ways of direct democratic participation turn
into the opposite, if utilized by people who have
not had the chance to acquire the needed skills.
However, in information-friendly societies, the
chances to implement successful programs to
provide equal chances for all members of society
to acquire capabilities (in the context of ICT) are
higher than in information-restrictive societies.
If we turn to information-restrictive societies,
however, we see that they are characterized by
a strong hierarchical order throughout society,
leading to fewer chances for social, economic,
and cultural movement. In such environments,
people obtain access to relevant public information when needed, whereby the information is
predominantly defined by the authorities or other
higher-ranking institutions or persons within the
respective hierarchical system. In such societies,
people are accustomed to information simply being provided and not having to actively retrieve
it. This attitude characterizes the relationship
between citizens and authorities, customers and
businesses, the public and the media, and students

and teacher. The education system in informationrestrictive cultures does not encourage curiosity
or question-based learning. The right answer
is the measure of success. What is right and what
is wrong again are defined by authorities in the
education system. People are not trained to address
their environments and to pose questions critically.
These answer-oriented societies are an obstacle
for the optimal utilization of new information and
communication technologies. Digital communication networks such as the Internet work best
with a question-oriented approach that leads to a
variety of plausible answers in different contexts.
Expecting the right and only answer (as people in
information-restrictive societies are trained) leads
to predictable disappointments and, therefore, less
motivation to get involved in new media.
In information-restrictive cultures, the flow of
information between authorities and citizens as
well as between businesses and customers follows
the push principle, whereby authorities and businesses decide which information is being passed
on. In such cultures, the Internet is perceived
merely as a new and additional (mass) medium
to transfer information to a mass audience. Consequently, a huge amount of information and
communication capacities of the Internet simply
are left unused. As there are not any geographical,
national, or cultural borders within digital communication networks, information and applications
from information-friendly cultural environments
compete with those from information-restrictive
cultures on a global stage.
We assume that information-friendly cultures
provide a competitive advantage for their members
in the global information society.

THE HUMAN-CENTERED AND

CULTURALLY SENSITIVE ICT
ADOPTION PROCESS
This chapter aims toward a better understanding of ICT adoption processes being dependent

1059

Culture and Technology

Figure 1. Model of a human-centric and culturally sensitive ICT adoption process

p
le
cro
Ma

technology

vel

vel
ro le Capabilities
Mic

Access

human
Skills

Cognition
ICT adoption process

el
lev
i c ro

vel
o le
a cr

technology

human
Skills

Access
Cognition

Capabilities

ICT adoption process

vel
o le
a cr

el
lev
ro
Capabilities
Mic

technology
Access

human
Skills

Cognition
ICT adoption process

from different information and communication

cultures. This process, in most societies, is driven
by predominantly techno-economic e-policies
that are still striving to overcome the Digital
Divide and to foster economic growth by means
of ICTs on the macro-level of state policy. This
approach has been criticized by various authors
in recent years (Preston, 2004; van Dijk, 2005;
Warschauer, 2002).
Most critics have in common the need to turn
away from techno-deterministic viewpoints to human-centered and culturally sensitive approaches.
This also can be characterized as a shift from
building infrastructures to creating identities,
or from bridging the digital divide to closing the
knowledge gap. This means putting the individual
in the center of the adoption process of technology; therefore, cognitive, cultural, and social
factors must be considered in order to achieve a
comprehensive understanding.
Following Mansell (2001) and Garnham (1997),
we suggest adopting a rights-based capabilities
approach in the ICT adoption process to ensure
that people have the possibilities to make informed

1060

decisions about the specific ways in which they

want to make use of ICTs. Acquiring those capabilities first demands awareness processes on
an individual cognitive level. Only when people
understand the individual and social implications of ICTs will they be able to make informed
choices about their specific usage patterns. The
stage when people shift from technology-driven
skills to culturally embedded understanding is the
stage that brings the ICT adoption process from
the macro-level of state e-policy to the micro-level
of the individualan indispensable precondition
to bring about the skilled user.
This process requires socially or culturally
motivated individuals on the one hand and governments who want to offer a set of alternative choices
for their citizens to allow them to achieve whatever
new-media-lifestyle they want on the other.
As we have already mentioned, the development of these adoption processes depends
strongly on the predominating information and
communication culture in a given society. In
information-friendly environments, people have
a greater chance of developing capabilities in the

Culture and Technology

context of ICT and, therefore, making informed

decisions based on the degree of their involvement with new information and communication
technologies.
The following model aims to visualize two
dimensions of the ICT adoption process: (1) the
stages from access to capabilities and (2) the
helical transformation of adoption processes (p)
as a result of the mutual relation between technology and culture. Every culturally embedded
adoption process leaves the new capable user on
an advanced stage that itself is the ground for the
access step to technology.

Model of a Human-Centered
and Culturally Sensitive ICT
Adoption Process
The adoption process, which also can be considered the major stage for targeted ePolicy measures,
starts with the problems of technology-determined
access. We need access to technology in order to
make experiences and to trigger the following
steps. Unfortunately, many processes get stuck
in the access stage; If they build it, they will
come could be the motive for access-only strategies. Most countries favor this access-dominated
strategy, which is predominantly in the interest of
the technology industry and, therefore, an industry
policy measurement.
The critique of the access-only strategy led
to a human-oriented enhancement of the same
strategy. People need to have adequate skills in
order to use the accessed technology. At first
glance, this could solve the problemnot only
provide people with technology but also train
them to use it. Similar to the access stage, the
skills stage also is geared predominantly to the
interest of the technology industry; in this case,
the big international or global software monopolists. Acquiring skills means dealing with a given
technology. The creative potential of people in the
context of technology is not addressed (National
Research Council, 2004).

A further step has to be taken in order to

involve the individual in the process of adopting
new information and communication technologies. People must know why they should make
use of ICTs and not only how to use them. On the
cognitive level, the awareness of technology in
the specific cultural sphere has to be raised. Here,
there is a cultural translation of technology. Only
when people understand the diverse patterns of
different practices of ICT usage will they be able
to make the informed choices as preconditions
for gaining capabilities. And only the capable
user will provide the basis for economic growth
and competitiveness for which most countries,
regions, and cultures are striving.
The capable user is the point of departure for
the next iteration of the ICT adoption process (p).
Capable users have different demands for access
to new technology and also represent a different
level for skills training. Such qualified users, who
differ in terms of cultural and social backgrounds,
represent the input into p, and so forth.

DIGITAL CULTURES
Cultural Shifts: Transculturality
In recent decades, the concept of interculturality
has been very popular and influential in regard
to the fairly young discipline of intercultural
communication (Leeds-Hurwitz, 1998). In this
context, communication was understood to be
an action taking place between countries that
were perceived as self-contained units. In this
traditional definition, cultures are seen as types of
autonomous islands that are virtually completely
closed-off, which Beck (1997) called metaphorically the container theory of society (p. 49). But
modern societies are very diverse entities. They
contain and incorporate many elements of different origins, and the boundaries between foreign
and indigenous cultures get blurred and finally
become untraceable. Tsagarousianou (2004) sug-

1061

Culture and Technology

gests that diasporas should not be seen as given

communities, a logical, albeit deterritorialized,
extension of an ethnic or national group, but
as imagined communities, continuously reconstructed and reinvented (p. 52). Welsch (1999)
developed a new approach of connected cultures,
which he called transculturality. This approach
emerged due to cultures being interconnected and
similar lifestyles merging and being assimilated.
Cultures cannot be perceived as homogenous units
anymore, because they are complex and diverse
in themselves. Cultures today are extremely
interconnected and entangled with each other.
Lifestyles no longer end at the borders of national
cultures, but go beyond these, are found in the
same way in other cultures. The way of life for
an economist, an academic or a journalist is no
longer German or French, but rather European or
global in tone (Welsch, 1999, 197f.).
This also can be observed in the Internet
community. People from different countries use a
sort of transcultural ideological language in chat
rooms and are united by common interests. Even
though they come from very different parts of
the world, they have more in common with each
other than they have with some members of their
respective national communities. The mutuality
derived from their similar interests prevails over
the mutuality derived from nationality.
Enhancing Welschs (1999) concept of transculturality, we consider that this concept needs
a more focused perspective on the permeability
between global and local cultures, which means
that transculturality allows the individual to switch
between different identities according to current
needs, feelings, interests, and demands. People
want to belong to a certain group and want to be
identified as a member of such a group; they do
not want to constantly act, think, and live on a
global level. The identity of the self cannot exist
only on a global level, and therefore, the search
for identity, collective or individual, ascribed or
constructed, becomes the fundamental source

1062

of social meaning. Yet identity is becoming

the main, and sometimes the only, source of
meaning in an historical period characterized
by widespread destructuring of organizations,
delegitimation of institutions, fading away of
major social movements, and ephemeral cultural
expressions (Castells, 2001, p. 3).

LINKING CULTURE,
KNOWLEDGE, AND ICTS
At this point, we introduce the extended concept
of culture, which is intertwined with the concept
of knowledge with the aim to discuss the correlation between culture, knowledge, and the role of
ICTs. This endeavor eventually should lead to an
approach that allows us to connect the complex
concept of cultures with its impact on various
spheres of our respective lives and, therefore, on
our identity. Therefore, the term digital culture
will be used to describe the model of mutual influence between culture and technology, which
we use as a fundamental framework to develop a
new understanding of the use of ICTs. This model
aims at an understanding of cultural differences
in handling information to guarantee a beneficial
development of society.
If the concept of transculturality is introduced
into the notion of knowledge, there is a rapid
increase of global knowledge. ICTs allow direct
communication between vast numbers of people
with different cultural backgrounds but do not
automatically distribute access to knowledge
equally. In fact, many citizens cannot gain access
to global knowledge or even local knowledge
other than to their own knowledge because of
their low economic status (digital divide) and
their low educational levels (cultural divide).
These divides create groups of haves or have-nots,
communication-rich or communication-poor,
winners or losers in the globalization process.
Concerning identities, these divides determine

Culture and Technology

Figure 2.
CULTURE

The
The spread
spread and
and
usage
usage ofofICTs
ICTs
affect
culture
affect culture
Push

DigitalCulture

Pull

Culturaldivide
divide
Cultural

Culture
shapesthe
the
Culture shapes
spread
spread and usage
usage
of ICTs
ICTs
of

Digital divide
Digital

TECHNOLOGY

the different opportunities of switching identity

levels. However, the more people are capable of
assuming different identities, both on a local and
global level, the more they are capable of gaining
advantages in the digital culture.
To avoid getting caught in this gap-trap and
to guarantee a sort of mutual benefit, we have to
find a way to reconcile some aspects of knowledge
production and acquirement, which means that
global knowledge has to be incorporated locally
(pull factor) in order to allow people to benefit
from global knowledge on a local level. Also, local knowledge has to be introduced into the cycle
of global knowledge production (push factor) in
order to make sure that there is an awareness of the
existence of this local identity in a global society.
Thus, developments in this global society can be
influenced with regard to local positions. We face
the challenging task of creating a permeable flow
of communication that allows for global and local
knowledge bases to interact.

COMBINING ICTS AND CULTURE

TO A NEW APPROACH
As already mentioned, technology and culture influence each other and are mutually dependent on
each other. It is, however, important to discuss the
terms culture and technology and the respective

The spread and

usage of ICTs
affect culture

points of view. It is equally important to demonstrate that culture and technology influence each
other by using the term digital culture.
Drawing upon these basic insights, we will
discuss the dialectic of shaping, diffusion, and
usage of ICTs in societies and different cultural
knowledge bases along the following dimensions:
content, distribution, and context.

DIGITAL CONTENT CULTURE

This indicator refers to the concept of knowledge production or, in other words, how data are
converted into knowledge. According to Willke
(2004), one has to distinguish between data and
knowledge, even though knowledge management
often is mistaken for data preparation and data
exchange.
In fact, data are nothing but raw material for
knowledge, and in the age of ICT, getting ahold
of useful data is not difficult. What is difficult,
however, is reducing and filtering huge amounts
of potentially useful data and converting them
into information first by putting them into a broad
context that adds relevance to them; knowledge
is gained by putting information into a practical
context and modifying or creating a practical due
to it in order to make the information practically
useful (Willke, 2004).

1063

Culture and Technology

ICTs, like the Internet, can transport and make

available huge amounts of data and information.
The content to be distributed is taken from this
basic range of knowledge. In a metaphorical sense,
the Internet can be linked to a sea of information,
all of which is useful in principle. Yet, to get ahold
of the invaluable essence of relevant information,
we have to siphon off all irrelevant information.
The focus is on translating data and information
into helpful insights that can be used to improve
real-life situations by adding practical relevance
to the data.
To guarantee the success of knowledge
transfer and the adaptation of new knowledge,
a transdisciplinary approach, in addition to an
interdisciplinary approach, has to be adapted.
This means that different scientific approaches
are used, but an effort also is made to involve the
local community in the process. In that way, one
can ensure that the goals are adapted to the local
culture, which increases the likelihood of locals
accepting them.
There are three main topics that have to be
discussed: knowledge management, learning
strategies, and educational approaches.

DIGITAL DISTRIBUTION CULTURE

The second dimension illustrates the relationship
between the channel that is used for information
transport and the dissemination of knowledge
the pull/push strategy.
The first aspect is the communication code: if
a message is to be transported, it has to be converted into numbers, text/words, and/or pictures
(Willke, 2004). There are limits to the amounts
and kinds of information that can be transported
in a certain channel. This depends on the type of
channel as well as the respective circumstances
(e.g., legal, technical, environmental, infrastructural) that, in fact, might influence the usage of
the channel.

1064

If we distinguish between explicit and tacit

(i.e., structured and unstructured) knowledge, we
can see how difficult it is to distribute knowledge.
While explicit knowledge (represented in documents, databases, products, and processes) is easy
to transfer, tacit knowledge is more dependent
on action, context and personal experience, which
makes it difficult to formalize and communicate
(Martin, 2003, p. 44).
The next aspect can be observed in culturally
influenced communication rituals. Each and every
one of our actions is influenced by culture (i.e.,
norms, values, beliefs), and by performing these
actions repeatedly, we permanently reinforce
our cultural understanding. A similar cycle of
conditioning can be found in technology. We
develop technology by drawing upon our cultural
understanding. We then use this technology on
a daily basis and, thereby, cause it to impact our
identity (reinforcement).
This development can be observed with the
personal computer. The term personal already
indicates that this technology was invented in a
very individualistic culture. The more time we
spend interacting with computers/technology in
this way, the more our patterns of thought and conveying knowledge are assimilated to those used
by computers/technology. Our way of thinking
becomes more abstract, and knowledge is reduced
to mere words and graphics, which lead to future
inventions being more abstract as a logical result.
The term digital culture means that we shape our
ICTs and are shaped by them in reverse.
The same applies to the Internet, whose basic
technology was developed in the academic-military information culture in California in the late
1950s and early 1960s. This implies a certain
practice of converting data into knowledge using
the Internet as a practical source for information.
In similar information cultures, it is clearer how
one can use this kind of data. But people from
a different cultural setting who have a different
concept of knowledge acquisition might not be

Culture and Technology

Table 1.
Dimensions

Digital Divide

Cultural Divide

Digital Culture

Content

Data, Information
knowing that

Knowledge
knowing how

Data, Information,
Knowledge
knowing why

Distribution

Channels limited to
technical possibilities

Inadequacy between
text and channel

Sharing and dissemination

of knowledge

Context

Limited to technical
Connectivity

Skills
Realization
Application

Inclusion
Awareness
Capabilities

able to make adequate use of the Internet. They

might not be familiar with the work processes
dominating the information culture within which
the Internet was developed. Therefore, it could
lead to difficulties to connect to and make use of
the Internet. Besides, the way the Internet is used
might not cohere with their cultural behavior.

DIGITAL CONTEXT CULTURE

There are factors that influence culture and technology on a meta-level. The central questions are:
What culture do we live in? What culture do we
work in? What culture do we act in?
An important indicator is knowledge as a commodity or as a free public good. First, costs are
linked to the question of whether knowledge can
be circulated freely or whether it should be treated
as property (IPRintellectual property rights;
DRMdigital rights management). Second, costs
refer to the investment and maintenance of infrastructure (hardware, software, bandwidth), and
finally, we have to calculate the costs for educating
people to use ICTs successfully and to develop the
ability to convert data into knowledge.
Another important indicator deals with power,
which can be explained by using the game theory.
It seems that most political and economic decisions
are based on the zero-sum game theory, which

means that any gain for one player represents an

equal loss for the other. We have to face the fact
that empowering people also means a loss of power
for the powerful to some extent. The powerful
create new competitors by empowering other
people, societies and/or subcultures. This is not so
much about unlimited development as it is about
creating a situation of equal opportunities.
Content, distribution, and context are influenced by technology as well as culture. It is
not enough to focus only on the digital divide
but also on the cultural divide, and by using the
concept of digital culture, we can develop a kind
of empathy with the goal that we have to create
inclusion and development as a central value, if
we really want to change from a segregated to an
inclusive society.

THE ONE AND THE MANY

In this respect, it is worth discussing the relationship of the one and the many. Due to global
challenges that endanger the species as a whole
and that must be met by a single set of intelligently coordinated actions, the partitions of
humankind are on the brink of forming a unit
on a planetary scale, and many cultures are on
the brink of forming one culture. The awareness
of this required delicate relationship between the

1065

Culture and Technology

one and the many may serve as a normative idea

that guides the measures to be taken to advance
world society.
The question is how one of the many relates to
another one and how the many relate to the oneness that is made up of the manifold. Is the world
society to become the common denominator of
the various identities? Or is one of the many the
only one? Or are the many merely summands of
the individual? Or do the many participate in a
one that goes beyond them?
The reductionist way of thinking in intercultural discourse is called universalism. Cultural universalism reduces the variety of different cultural
identities to what they have in common. Identities
are homogenized by a sort of melting pot that
was named McWorld (Barber, 2001). Modernism
(i.e., the pursuit of human rights, democracy, and
capitalism based on the same mode of metabolism
carried out by the same technology everywhere)
is universalisticshimmering between a claim to
liberalism and pompous imperialistic behavior as
it is witnessed by its adversaries. In either case, it
gets rid of the richness of cultural identities, the
many are reduced to a shallow one, and there is
no diversity in the unity.
A second strand in intercultural discourse
revolves around the way of thinking that overuses projection. It may be called particularism
or totalitarianism. Cultural particularism or
totalitarianism extrapolates what separates one
cultural identity from the rest and construes an
imaginary common. It also leads to homogenization. The melting pot in this case, however,
was named Jihad (Barber, 2001), because it is
the anti-modern fundamentalism that may be a
good example for imposing a certain one out of
the many on the rest of them. Here, a culture that
is accredited with very specific social relations is
raised to the level of the ideal, which is to serve
as a model for all other cultures to copy. Thus, a
specific form is built up to be the general norm.
Inasmuch as it is something particular that is
raised in this manner, it concerns particularism.

1066

Inasmuch as it reaches the status of the general

norm, it concerns totalitarianism. This results
also in unity without diversity.
A third way to conceive intercultural discourse
is relativism. Cultural relativism rests on the figure
of dissociation. By denying any commonality of
different cultural identities, it yields fragmentation. The many fall apart. These concepts of
multiculturalism and separatism suit postmodern
thoughts. Here, each of the many cultures is seen
as something with the right to exist and remain
free from external interference. Each special case
is made into a norm in its own right. Inasmuch
as it is one of many that is made into a norm, we
may speak of pluralism. Inasmuch as every special
case is treated thus, we must, however, speak of
indifferentism. Relativism does not claim general
validity and does not wish to unify anything or
anyone. The postmodernist form leaves differences as they are. World society would simply
be diversity without unity.
None of these three options can satisfy. Either
the one is regarded as the necessary and sufficient
condition for the many, the many are considered
necessary and sufficient for the one, or one and
many are deemed independent.
Cultural thinking that reconciles the one and
the many is achievable only on the basis of an
integration and differentiation way of thinking. It
integrates the differences of the manifold cultural
identities and differentiates the common as well.
Welsch (1999) coined the term transculturalism
and notions of glocalization (Robertson, 1992)
or new mestizaje (a term coined by John Francis
Burke in Reconciling Cultural Diversity With a
Democratic Community: Mestizaje as Opposer to
the Usual Suspects in Wieviorka (2003), which
are useful in this context.
The process of emergence of a new sustainable world society may be sketched in terms of
dialectics. Diversity is not abolished but rather
sublated and leads in an evolutionary leap to a
unity through diversity, which, in turn, enables and
constrains diversity in order to make it diversity

Culture and Technology

in unity, which then builds the new base for unity

through diversity. World culture is located on the
macro-level; the partitions of world culture that are
located on the micro-level take care of the world
culture in order to preserve humanity.

application to communications. The Public, 4(4),

25-34.

CONCLUSION

Hofkirchner, W. (1999). Does electronic networking entail a new stage of cultural evolution? In
P. Fleissner & J. C. Nyiri (Eds.), Cyberspace: A
new battlefield for human interests? Philosophy of
culture and the politics of electronic networking
(vol. II) (pp. 3-22). Innsbruck, ron, Budapest:
Studienverlag.

Starting with a critique of both techno-deterministic and social-constructive approaches toward the
relationship between technology and culture, we
argue for a dialectical, mutual-shaping approach.
Especially in the context of information and
communication technologies (ICTs) and society,
this dialectical relationship between culture and
technology is important. To strive for the capable
user, cultural dimensions have to be incorporated
into a model that transfers the spread and usage of
technology on the one hand and the social shaping
of technology on the other. The concept of digital
culture represents a framework that embraces the
techno-cultural dimensions of content, distribution, and context. This framework provides an
applicable instrument that allows addressing the
important questions in the context of technology
and society, such as equal knowledge distribution,
provision of capabilities, and social inclusion.

References
Barber, B. (2001). Jihad vs. McWorld. New York:
Ballantine.
Beck, U. (1997). Was ist Globalisierung? Frankfurt am Main: Suhrkamp.
Castells, M. (2001). The information age: Economy, society and culture: The rise of the network
society (vol. 1). Oxford: Blackwell.
Garnham, N. (1997). Amartya sens capabilities approach to the evaluation of welfare: Its

Hofkirchner, W. (1994). On the philosophy of design and assessment of technology. In S. Katsikides

(Ed.), Informatics, organization and society (pp.
38-46). Vienna-Munich: Oldenbourg.

Hofkirchner, W. (2004). Unity through diversity:

DialecticsSystems thinkingSemiotics. In H.
Arlt (Ed.), The unifying aspects of cultures (CDROM). Wien: INST.
Leeds-Hurwitz, W. (1998). Notes in the history of
intercultural communication: The foreign service
institute and the mandate for intercultural training.
In J. N. Martin, T. K. Nakayama, & L. A. Flores
(Eds.), Readings in cultural contexts (pp. 15-29).
Mountain View, CA: Mayfield.
Maier-Rabler, U. (1995). Die neuen Informationsund Kommunikationstechnologien als
gesellschaftspolitische Herausforderung. Informatik Forum, 9(4), 157-168.
Maier-Rabler, U. (2002). Cultural aspects and digital divide in Europe. Medien Journal, 3, 14-32.
Maier-Rabler, U., & Sutterltti, E. (1992). Pressestatistik im internationalen Vergleich. Endbericht
des Teilprojekts Pressestatistik und Datenkoordination im Rahmen des Forschungsprogramms
konomie und Zukunft der Printmedien. Salzburg: Research Report.
Mansell, R. (2001). New media and the power
of networks [Inaugural Professorial Lecture,
Dixons Chair in New Media and the Internet].
London School of Economics and Political Sci-

1067

Culture and Technology

ence. Retrieved February 25, 2006, from www.lse.

ac.uk/collections/media@lse/pdf/rmlecture.pdf
Martin, B. (2003). Knowledge management and
local government: Some emerging trends. Asia
Pacific Management Review, 8(1), 43-57.
National Research Council. (2004). Beyond
productivity. Information technology, innovation, and creativity. Washington, DC: National
Academy Press.
Preston, P. (2004). European Union ICT policies:
Neglected social and cultural dimensions. In J.
Servaes (Ed.), The European information society
(pp. 33-58). Bristol: Intellect Books.
Robertson, R. (1992). Globalization. London:
Sage.
Tsagarousianou, R. (2004). Rethinking the
concept of diaspora: Mobility, connectivity and
communication in a globalised world. Westminster Papers in Communication and Culture, 1(1),
52-66.

van Dijk, J. (2005). The deepening divide. Inequality in the information society. Thousand
Oaks, CA: Sage.
Warschauer, M. (2002, July 1). Reconceptualizing
the digital divide. First Monday, Peer Reviewed
Journal on the Internet, 7(7). Retrieved from
http://www.firstmonday.org/issues/issue7_7/
warschauer/index.html
Welsch, W. (1999). TransculturalityThe puzzling form of cultures today. In M. Featherstone,
& S. Lash (Eds.), Spaces of culture: City, nation,
world (pp.194-213). London: Sage. Retrieved
February 2, 2006, from http://www2.uni-jena.
de/welsch/Papers/transcultSociety.html
Wieviorka, M. (2003). Kulturelle differenzen
und kollektive identitten. Hamburg: Hamburger
Edition.
Willke, H. (2004). Einfhrung in das systemische
wissensmanagement. Heidelberg: Carl-AuerSysteme.

This work was previously published in Information Technology Ethics: Cultural Perspectives, edited by S. Hongladarom and
C. Ess, pp. 54-67, copyright 2007 by Information Science Reference, formerly known as Idea Group Reference (an imprint of
IGI Global).

1068

1069

Chapter 3.4

Importance of Interface Agent

Characteristics from End-User
Perspective
Alexander Serenko
Lakehead University, Canada

ABSTRACT
This article reports on an empirical investigation of
user perceptions of the importance of several characteristics of interface agents. Interface agents are
software entities that are incorporated into various
computer applications, including electronic mail
systems. As evidenced by the growing body of
empirical studies and the increasing number of
interface agent-based applications on the software
market, there is a strong need for the development
of this technology. According to a meta-review of
agent-related literature by Dehn and van Mulken
(2000), there are several characteristics of interface agents that require special attention from
agent developers. However, prior to this study,
the importance of these characteristics from the
end-user perspective remained unclear. In order to
identify the significance of these characteristics, a
group of actual users of an e-mail interface agent
was surveyed. The results indicate that information accuracy and the degree of the usefulness of
an agent are the most salient factors, followed by
user comfortability with an agent, the extent of
user enjoyment, and visual attractiveness of an

agent. The implications of the findings for both

theory and practice are discussed.

Introduction
To create an artificial being has been a dream of
men since the birth of science. Professor Hobby
(William Hurt) in Artificial Intelligence (Spielberg, 2002)
For thousands of years, people have thought of
someone doing basic tasks for them. That could be
a robot, a cyborg, or a well-trained pet. Not until
the beginning of the 21st century did it become
possible. Now, with the recent development of
telecommunications networks and computer technologies, a new type of software application plays
the role of virtual assistants that potentially may
alleviate some of the problems associated with the
employment of software systems. This class of applications often is referred to as intelligent agents,
software agents, avatars, or interface agents. As
demonstrated by the growing body of academic
literature and by the increasing number of agentbased software applications on the market, there is

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Importance of Interface Agent Characteristics from End-User Perspective

increased interest in the creation of such software

entities. In this article, these software systems are
labeled as interface agents.
Interface agents emerged from the recent developments in the field of intelligent agents. The
idea of software agents was first introduced by
John McCarthy (1956, 1958) and later coined by
the MIT Lincoln Laboratory computer scientist
Oliver Selfridge. In the 1980s, this concept was
explored by agent visionaries such as Marvin
Minsky and Alan Kay and further utilized in
the recent classic works of Pattie Maes, Nicolas
Negroponte, Jeffrey Bradshaw, Hyacinth Nwana,
and Divine Ndumu. The past few years have witnessed the rapid development of prototypes and
working models of intelligent agents, many of
which already are incorporated in end-user commercial applications. A number of recent studies
demonstrate the fruitfulness and viability of using
agent-based technologies in various areas; for
example, in automatic negotiation (Castro-Schez
et al., 2004; Fatima et al., 2005), natural-language
customer support services (Lester et al., 2004),
education (Takacs, 2005), and user notification
systems (Horvitz et al., 2003). Some academics
have shifted their research from human-agent
interaction to human-agent cooperation (Rickel
& Johnson, 2000; Rickel et al., 2002) and manmachine symbiosis (Klein et al., 2004; Lesh et
al., 2004; Lesh et al., 1999), when the human
user and the software agent collaborate toward
achieving shared goals.
In terms of this article, an interface agent is
defined as an autonomous (i.e., independent),
continuous (i.e., long-lived), reactive (i.e., it
monitors an external environment and reacts to
any changes), and collaborative (i.e., it cooperates
with other software processes or agents) software
entity that exhibits strong visual or audio presence
in the computer interface and that communicates
with a user directly (i.e., by bypassing intermediaries) (Detlor, 2004; Lieberman & Selker, 2003;
Serenko & Detlor, 2004). Interface agents draw
their strength from the naturalness of the living-

1070

organism metaphor in terms of both cognitive

accessibility and communication style (Laurel,
1997, p. 68). Typically, interface agents are personalizable and implemented in the form of humanlike
or cartoonlike animated characters, electronic
figures, graphical user interfaces, textual boxes,
or any other visual components (Godoy et al.,
2004; Schiaffinoa & Amandi, 2004).
Having the available agent technology is insufficient; it also should be accepted and utilized
appropriately by its target users. For the past 10
years, there have been various attempts to understand what people like or dislike in interface
agents and why they adopt or reject them. The
goal of this stream of research is to develop a
valid, complete list of characteristics that interface agents should possess that would warrant the
end-user acceptance of this technology.
By performing a meta-analysis of the human-computer interaction literature, Dehn and
van Mulken (2000) presented a comprehensive
yet exhaustive list of characteristics of interface
agents that potentially may influence the human-interface agent interaction process. Most
of these characteristics are drawn from various
independent investigations conducted in laboratory settings. At the same time, no study reports
how real-life users value the characteristics of
an interface agent-based technology. In order to
bridge that void, the present investigation attempts
to solicit and to analyze the opinions of interface
agent users on several key characteristics of the
technology. It is assumed that this information
potentially may improve the quality of the technology and the way it is delivered to the customer.
For example, if agent manufacturers could know
what interface agent characteristics are more or
less important for users, they would be able to
concentrate their short-term efforts to improve
positive user perceptions of these characteristics.
This, in turn, might increase user satisfaction with
agent-based technology and accelerate the rate of
innovation diffusion.

Importance of Interface Agent Characteristics from End-User Perspective

As such, Dehn and van Mulken (2000) classified the various characteristics of interface
agents (e.g., the users subjective experience of
the system, the users behavior while interacting with the system, and the outcome of the
interaction). Each category includes several factors. However, it is not viable to investigate the
importance of these characteristics applied to all
types of interface agents in a single project. Since
interface agents may be incorporated in the form
of personal secretaries, Internet guides, electronic
commerce assistants, or educators, a separate
study is required for each kind of interface agents.
It is believed that interface agents embedded in
different types of software environments may
require certain system-specific features and facets.
For example, users who work with an interface
agent that facilitates online shopping may look for
effectiveness and efficiency. In contrast, people
who employ an interface agent as entertainers
may emphasize the aspect of enjoyment over that
of effectiveness or efficiency.
With respect to the present study, interface
agents for electronic mail were chosen for two
reasons. First, e-mail is an important telecom-

munications medium that is utilized heavily by

both individuals and organizations. However,
todays e-mail systems provide inadequate support for constantly changing user needs, fail to
convey ambiguous content and human emotions,
overload people with continually growing flows
of unstructured information, and exhibit an
inefficient direct manipulation interface. As a
result, many individuals feel frustrated utilizing
e-mail. The use of interface agents is a potential
solution to the currently challenging task of email management. Second, the software market
presents several versions of interface agents that
have been delivered to end users. Currently, most
other types of interface agents have been realized
in the form of pilot studies, working prototypes,
or beta versions. This identifies the opportunity
to reach the actual users of this technology and
to poll them directly. It is for these reasons that
interface agents for e-mail were selected.
A review of the general characteristics of interface agents presented by Dehn and van Mulken
(2000) allowed the identification of several factors
that were believed to be applicable to the e-mail
environment. Table 1 offers a list of these char-

Table 1. Characteristics of interface agents

N

Characteristics
With respect to interface agents for e-mail, it is important for users:

to believe that an interface agents appearance should correspond to its level of

intelligence.
to believe that the information provided by an interface agent is accurate.

to like the appearance of an interface agent.

to feel comfortable with an interface agent.

to perceive an interface agent useful.

to perceive an interface agent enjoyable.

to perceive all interactions with an interface agent as natural.

to avoid being distracted by an interface agent while engaged in important tasks.

1071

Importance of Interface Agent Characteristics from End-User Perspective

acteristics. However, little is known about how

important these characteristics are for the actual
users of e-mail interface agents. As noted by Dehn
and van Mulken (2000), the results of the empirical
studies that identified these characteristics appear
to be mixed and inconsistent.
To bridge this void and to rank the importance
of the previous characteristics, this study polled
the actual users of e-mail interface agents. It was
believed that the end users who have utilized this
technology for a long period of time may present valid and reliable information that will be of
interest to agent researchers and developers. The
following research question was proposed:
How important are the characteristics of e-mail
interface agents identified in Table 1 from the
end-user perspective?

Methodology and Results

In order to answer the studys research question, a survey of current and past users of an
interface agent-based application for e-mail was
conducted. Despite the extensive work underway

in the incorporation of interface agents in e-mail

applications, most previous studies and projects
have been realized in forms of conceptual discussions, preliminary empirical investigations, and
pilot systems (Bergman et al., 2002; Dabbish et
al., 2005; Florea & Moldovanu, 1996; Griss et al.,
2002; Gruen et al., 1999; Lashkari et al., 1994;
Maes, 1994; Voss, 2004) rather than in end-user
products. E-mail notification applications are
one of the first commercial systems that utilize
interface agent technologies in the electronic
mail environment. This type of interface agents
was chosen to conduct a user survey. Out of all
commercially available interface agent systems
for e-mail, Blind Bat Software was chosen randomly by the researcher, the executives of the
company were approached, and agreement was
reached. The list of customers who potentially
might serve as the studys participants was sent
to the researcher. Figure 1 presents a screenshot
of the software product.
In order to poll e-mail agent users on their perceptions of the importance of the characteristics
of interface agents, a survey instrument was designed. The questionnaire provided basic instructions, a definition of an interface agent for e-mail,

Figure 1. E-mail announcer developed by Blind Bat Software (http://www.blindbat.com)

1072

Importance of Interface Agent Characteristics from End-User Perspective

and several screenshots of the agent developed

by Blind Bat. Users were asked to indicate their
opinion on perceptions of the importance of agent
characteristics outlined in Table 1. Particularly,
the question stated, Based on your experience
with interface agents for e-mail, how important
is it for you? After this, eight statements were
provided, measured on a seven-point Likert-type
scale ranging from totally unimportant to very
important. In addition, demographic information
was solicited. The data for this investigation were
collected as part of a larger project conducted by
Serenko (2005).
By utilizing the total design method (Dillman, 1999), the four-phase survey process was
developed. As such, all respondents were e-mailed
an initial request to participate in the study and
then three follow-up reminders. Fifty-nine usable
responses were obtained. An acceptable response
rate was achieved. Note that the actual response
rate may not be revealed as per the nondisclosure
agreement with Blind Bat Software.
Eighty percent of the surveyed users were male,
and 20% were female. Over 65% of all users were
between 31 and 50 years old, and the 46-to-50age category was the most frequent user group.

Over one-half of the respondents were occupied

in the information technology sector; most of
them were well-educated and financially welloff and demonstrated a high degree of personal
innovativeness in the domain of IT. According to
Rogers (2003), this group of people corresponds
to innovators, who constitute 2.5% of all people
that adopt a particular product or service.
Recall that respondents were asked to rate their
perceptions of the importance of eight characteristics of interface agents on a seven-point Likerttype scale. The purpose was to understand what
characteristics were more or less imperative from
the end users point of view. Figure 2 visualizes
the results, and Table 2 presents the list of questions sorted by the mean.
To analyze whether there were differences
in these eight means, the ANOVA test was conducted. The goal of this statistical method is to
determine the existence of differences among
several population means (Aczel, 1996). This
technique is an extension of the two-sample t
test. The results demonstrated that there was a
high degree of confidence that at least some of
the means differed from one another (F = 12.846,
d.f. between = 7, d.f., within = 456, significance

Figure 2. User perceptions of the importance of interface agent characteristics

7.00
6.00
5.00

5.34

5.47

natural
interactions

nondistraction

5.78

5.86

5.90

6.05

6.28

attractiveness

enjoyment

comfortability

usefulness

info accuracy

4.22

4.00
3.00
2.00
1.00
0.00

appearance vs
intelligence

Dimension

1073

Importance of Interface Agent Characteristics from End-User Perspective

Table 2. User perceptions of the importance of interface agent characteristics

Based on your experience with interface agents for e-mail,
how important is it for you:
to believe that the information provided by an interface agent
is accurate?
to perceive an interface agent as useful?
to feel comfortable with an interface agent?
to perceive an interface agent as enjoyable?
to like the appearance of an interface agent?
to avoid being distracted by an interface agent while engaged
in important tasks?
to perceive all interactions with an interface agent as natural?
to believe that an interface agents appearance should
correspond to its level of intelligence?

level = 0.000). To measure the practical value

of the detected differences, the effect size was
calculated as the ratio of sum of squares between
the sum of squares total. The effect size was very
strong (2 = 0.16).
After it was determined that differences existed
among the means, the Tukey Honestly Significant
Difference test was done by using SPSS. The
Tukey post hoc test is a statistical method of
pairwise comparisons of the population means.
It allows the comparison of every possible pair
of means using a selected single level of significance. The test yielded a matrix where asterisks
(*) indicated significantly different group means
at an alpha level of 0.1. Table 3 presents the results
of mean comparisons.
Based on these results, several statistically significant differences in item means were observed.
Overall, the means of the questions positioned
on the left-hand side and right-hand side of Figure 2 strongly differed from one another. This
demonstrated a strong degree of confidence that
respondents were able to distinguish among the
questions and that the results presented in Figure
2 were statistically sound.

1074

Mean
6.28

Std
dev
1.04

6.05
5.90
5.86
5.78
5.47

1.13
1.10
1.13
1.17
1.74

5.34
4.22

1.36
1.86

Discussion and Conclusion

Recall that the purpose of this study was to obtain
strong empirical evidence on the importance of
interface agent characteristics to bridge the gap in
the human-agent interaction literature. The results
of the survey showed that trust in an agent (i.e.,
information accuracy) as well as an agents utility
(i.e., the persona effect) were the most important
factors from the end users point of view. They
were followed by the degree of conformability
and enjoyment with an agent.
First, agent users believed that the accuracy
of any information provided by an agent was the
most critical factor. This finding is consistent with
prior research that points out the importance of
trustworthiness in human-agent interaction (Bickmore & Cassell, 2005; Bickmore & Picard, 2005;
Hertzum et al., 2002). Indeed, in order to delegate
tasks to an agent, a person must believe that the
agent will perform them accurately and report
back the true rather than the desirable state.
Second, respondents indicated the significance
of an agents usefulness. This, again, is consistent
with prior empirical research and speculations on
the importance of the persona effect in agents. The
persona effect emerges when an interface agent
adds the positive perceptions of usefulness, ease of

Importance of Interface Agent Characteristics from End-User Perspective

Table 3. The Tukey test

(I)
CHARACTERISTIC

1
appearance corresponds to the
level of intelligence

2
information accuracy

3
attractiveness

4
comfortability

5
usefulness

6
enjoyment

7
natural interactions

8
little distraction

(J) CHARACTERISTIC
2
3
4
5
6
7
8
1
3
4
5
6
7
8
1
2
4
5
6
7
8
1
2
3
5
6
7
8
1
2
3
4
6
7
8
1
2
3
4
5
7
8
1
2
3
4
5
6
8
1
2
3
4
5
6
7

Mean Difference
(I-J)
-2.05(*)
-1.55(*)
-1.67(*)
-1.83(*)
-1.64(*)
-1.12(*)
-1.24(*)
2.05(*)
.50
.38
.22
.41
.93(*)
.81(*)
1.55(*)
-.50
-.12
-.28
-.09
.43
.31
1.67(*)
-.38
.12
-.16
.03
.55
.43
1.83(*)
-.22
.28
.16
.19
.71(*)
.59
1.64(*)
-.41
.09
-.03
-.19
.52
.40
1.12(*)
-.93(*)
-.43
-.55
-.71(*)
-.52
-.12
1.24(*)
-.81(*)
-.31
-.43
-.59
-.40
.12

Sig.
.000
.000
.000
.000
.000
.000
.000
.000
.486
.800
.986
.719
.006
.028
.000
.486
1.000
.956
1.000
.674
.920
.000
.800
1.000
.999
1.000
.353
.674
.000
.986
.956
.999
.995
.092
.275
.000
.719
1.000
1.000
.995
.440
.761
.000
.006
.674
.353
.092
.440
1.000
.000
.028
.920
.674
.275
.761
1.000

1075

Importance of Interface Agent Characteristics from End-User Perspective

use, or enjoyment with an existing system. The key

outcome of the persona effect is the improvement
of existing software applications by embedding
interface agents. By emphasizing the importance
of an agents usefulness, subjects demonstrated
that value-added services were the key factors
influencing their adoption decisions.
Third, perceptions of the importance of comfortability and enjoyment with an agent were also
high. The extent to which a user feels comfortable
employing an agent partially corresponds to the
ease of use of the agent.
Fourth, items pertaining to non-distraction
and the naturalness of interactions received lower
scores. Prior work suggests that a user should perceive all interactions with an agent to be natural,
and the agent is not supposed to disrupt current
user activities. However, this item received a
lower score compared with information accuracy,
usefulness, comfortability, and enjoyment.
Finally, in contrast to prior research, respondents stated that the appearance of an agent
should not necessarily correspond to its level of
intelligence. Two assumptions may explain this
contradiction. First, highly innovative individuals
might wish to utilize an agent that looks maximally intelligent, regardless of its actual degree
of intelligence. Second, if users were not satisfied
with the agents appearance, they easily might
install another one, given that there is a variety of
cartoon or humanlike agent characters available
on the Web. Thus, end users had control over the
interface of an agent that reduced their perception
of the importance of the agents appearance.
These findings are important for both theory
and practice. With respect to theory, the investigation discovered some discrepancies between
the view of agent researchers and the opinion
of real-life users. With regard to practice, it is
suggested that agent designers begin emphasizing the more important characteristics of e-mail
interface agents in their products. In the short
term, they need to concentrate their efforts on
the development of interface agents that pro-

1076

vide accurate and reliable information and are

perceived to be really useful by the end users.
After the issues of information accuracy and
usefulness are addressed, agent developers may
attempt to improve several other characteristics
of interface agents. They may improve the degree
of user comfortability with the software, increase
user enjoyment, and advance the visual appeal of
an agent. In the long term, agent manufacturers
may want to decrease the degree of an agents
intrusiveness and facilitate the naturalness of
human-agent interactions. However, it is unlikely
that they will need to create an interface agent
whose appearance would correspond to its level
of intelligence. Instead, they should offer a variety
of agent interfaces and leave it up to the end users
to decide which one to utilize.
Overall, this investigation is one of the first
documented attempts to explore the importance
of interface agent characteristics by polling the
actual users of this technology. The author hopes
that other researchers will continue to explore this
field that will lead to the creation of really useful
interface agents.

References
Aczel, A. D. (1996). Complete business statistics.
Chicago: Irwin.
Bergman, R., Griss, M., & Staelin, C. (2002). A
personal email assistant [technical report HPL2002-236]. Hewlett-Packard Company.
Bickmore, T. W., & Cassell, J. (2005). Social
dialogue with embodied conversational agents. In
J. van Kuppevelt, L. Dybkjaer, & N. O. Bernsen
(Eds.), Advances in natural, multimodal dialogue
systems. New York: Kluwer Academic.
Bickmore, T. W., & Picard, R. W. (2005). Establishing and maintaining long-term human-computer
relationships. ACM Transactions on Computer-

Importance of Interface Agent Characteristics from End-User Perspective

Human Interaction, 12(2), 293-327.

Computing Systems, Pittsburgh, Pennsylvania.

Castro-Schez, J. J., Jennings, N. R., Luo, X., &

Shadbolt, N. (2004). Acquiring domain knowledge for negotiating agents: A case study. International Journal of Human-Computer Studies,
61(1), 3-31.

Hertzum, M., Andersen, H. H. K., Andersen,

V., & Hansen, C. B. (2002). Trust in information sources: Seeking information from people,
documents, and virtual agents. Interacting with
Computers, 14(5), 575-599.

Dabbish, L. A., Kraut, R. E., Fussell, S., & Kiesler,

S. (2005). Understanding email use: Predicting
action on a message. Proceedings of the SIGCHI
Conference on Human Factors and Computing
Systems, Portland, Oregon.

Horvitz, E., Kadie, C., Paek, T., & Hovel, D.

(2003). Attentive user interfaces: Models of attention in computing and communication: From
principles to applications. Communications of the
ACM, 46(3), 52-59.

Dehn, D. M., & van Mulken, S. (2000). The

impact of animated interface agents: A review
of empirical research. International Journal of
Human-Computer Studies, 52(1), 1-22.

Klein, G., Woods, D. D., Bradshaw, J. M., Hoffman, R. R., & Feltovich, P. J. (2004). Ten challenges for making automation a team player in
joint human-agent activity. IEEE Transactions on
Intelligent Systems, 19(6), 91-95.

Detlor, B. (2004). Towards knowledge portals:

From human issues to intelligent agents. Dordrecht, The Netherlands: Kluwer Academic
Publishers.
Dillman, D. A. (1999). Mail and Internet surveys:
The tailored design method (2nd ed.). New York:
John Wiley & Sons.
Fatima, S., Wooldridge, M., & Jennings, N.R.
(2005). A comparative study of game theoretic and
evolutionary models for software agents. Artificial
Intelligence Review, 23(2), 187-205.
Florea, A. M., & Moldovanu, A. (1996). An intelligent email agent. Proceedings of the 4th Romanian
Conference on Open Systems, Bucharest.
Godoy, D., Schiaffino, S., & Amandi, A. (2004).
Interface agents personalizing Web-based tasks.
Cognitive Systems Research, 5(3), 207-222.
Griss, M. et al. (2002). CoolAgent: Intelligent
digital assistants for mobile professionalsPhase
1 retrospective [HP Laboratories report HPL2002-55(R1)]. Hewlett-Packard Company.
Gruen, D., Sidner, C., Boettner, C., & Rich, C.
(1999). A collaborative assistant for email. Proceedings of the Conference on Human Factors and

Lashkari, Y., Metral, M., & Maes, P. (1994).

Collaborative interface agents. Proceedings of
the Twelfth National Conference on Artificial
Intelligence, Seattle, Washington.
Laurel, B. (1997). Interface agents: Metaphors
with character. In J. M. Bradshaw (Ed.), Software agents (pp. 67-78). Menlo Park, CA: AAAI
Press.
Lesh, N., Marks, J., Rich, C., & Sidner, C.L. (2004).
Man-computer symbiosis revisited: Achieving
natural communication and collaboration with
computers. The Institute of Electronics, Information and Communication Engineers Transactions
on Electronics, E87-C(12).
Lesh, N., Rich, C., & Sidner, C. L. (1999). Using
plan recognition in human-computer collaboration. Proceedings of the Seventh International
Conference on User Modelling, Banff, Canada.
Lester, J., Branting, K., & Mott, B. (2004). Conversational agents. In M. P. Singh (Ed.), Practical
handbook of Internet computing (pp. 10/11-10/17).
Baton Rouge: Chapman & Hall/CRC Press.
Lieberman, H., & Selker, T. (2003). Agents for the

1077

Importance of Interface Agent Characteristics from End-User Perspective

user interface. In J. M. Bradshaw (Ed.), Handbook

of agent technology. MIT Press.
Maes, P. (1994). Agents that reduce work and
information overload. Communications of the
ACM, 37(7), 31-40.
McCarthy, J. (1956). Inversion of functions defined by turing machines: Princeton University
Press.

terface agent interaction: Personalization issues.

International Journal of Human-Computer Studies, 60(1), 129-148.
Serenko, A. (2005). User adoption of interface
agents for email. Doctoral dissertation, Hamilton,
Canada: McMaster University.
Serenko, A., & Detlor, B. (2004). Intelligent agents
as innovations. AI & Society, 18(4), 364-381.

McCarthy, J. (1958). Programs with common

sense. Proceedings of the Teddington Conference
on the Mechanization of Thought Processes,
London.

Spielberg, S. (Director/Producer), Curtis, B. (Producer), Harlan, J. (Producer), Kennedy, K (Producer), Parkes, W. F. (Producer). (2002). Artificial
intelligence [Motion picture]. DreamWorks.

Rickel, J., & Johnson, W. L. (2000). Task-oriented

collaboration with embodied agents in virtual
worlds. In J. Cassell, J. Sullivan, S. Prevost, & E.
Churchill (Eds.), Embodied conversational agents
(pp. 95-122). Boston: MIT Press.

Takacs, B. (2005). Special education and rehabilitation: Teaching and healing with interactive
graphics. IEEE Computer Graphics and Applications, 25(5), 40-48.

Rickel, J. et al. (2002). Toward a new generation

of virtual humans for interactive experiences.
IEEE Intelligent Systems, 17(4), 32-38.
Rogers, E. M. (2003). Diffusion of innovations
(5th ed.). New York: Free Press.

Voss, I. (2004). The semantic of episodes in communication with the anthropomorphic interface
agent MAX. Proceedings of the 9th International
Conference on Intelligent User Interface, Funchal,
Madeira, Portugal.

Schiaffinoa, S., & Amandi, A. (2004). UserIn-

This work was previously published in International Journal of Intelligent Information Technologies, Vol. 2, Issue 2, edited
by V. Sugumaran, pp. 49-60, copyright 2006 by IGI Publishing, formerly known as Idea Group Publishing (an imprint of IGI
Global).

1078

1079

Chapter 3.5

Web Mining for Public

E-Services Personalization
Penelope Markellou
University of Patras, Greece
Angeliki Panayiotaki
University of Patras, Greece
Athanasios Tsakalidis
University of Patras, Greece

INTRODUCTION
Over the last decade, we have witnessed an explosive growth in the information available on the
Web. Today, Web browsers provide easy access
to myriad sources of text and multimedia data.
Search engines index more than a billion pages
and finding the desired information is not an easy
task. This profusion of resources has prompted the
need for developing automatic mining techniques
on Web, thereby giving rise to the term Web
mining (Pal, Talwar, & Mitra, 2002).
Web mining is the application of data mining
techniques on the Web for discovering useful patterns and can be divided into three basic categories:
Web content mining, Web structure mining, and
Web usage mining. Web content mining includes
techniques for assisting users in locating Web
documents (i.e., pages) that meet certain criteria,

while Web structure mining relates to discovering information based on the Web site structure
data (the data depicting the Web site map). Web
usage mining focuses on analyzing Web access
logs and other sources of information regarding
user interactions within the Web site in order to
capture, understand and model their behavioral
patterns and profiles and thereby improve their
experience with the Web site.
As citizens requirements and needs change
continuously, traditional information searching,
and fulfillment of various tasks result to the loss of
valuable time spent in identifying the responsible
actor (public authority) and waiting in queues.
At the same time, the percentage of users who
acquaint with the Internet has been remarkably
increased (Internet World Stats, 2005). These two
facts motivate many governmental organizations
to proceed with the provision of e-services via

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Web Mining for Public E-Services Personalization

their Web sites. The ease and speed with which

business transactions can be carried out over the
Web has been a key driving force in the rapid
growth and popularity of e-government, e-commerce, and e-business applications.
In this framework, the Web is emerging as the
appropriate environment for business transactions
and user-organization interactions. However,
since it is a large collection of semi-structured and
structured information sources, Web users often
suffer from information overload. Personalization
is considered as a popular solution in order to
alleviate this problem and to customize the Web
environment to users (Eirinaki & Vazirgiannis,
2003). Web personalization can be described, as
any action that makes the Web experience of a
user personalized to his or her needs and wishes.
Principal elements of Web personalization include
modeling of Web objects (pages) and subjects
(users), categorization of objects and subjects,
matching between and across objects and/or
subjects, and determination of the set of actions
to be recommended for personalization.
In the remainder of this article, we present
the way an e-government application can deploy
Web mining techniques in order to support intelligent and personalized interactions with citizens.
Specifically, we describe the tasks that typically
comprise this process, illustrate the future trends,
and discuss the open issues in the field.

lowing four tasks (Etzioni, 1996):

BACKGROUND
The close relation between Web mining and Web
personalization has become the stimulus for
significant research work in the area (Borges &
Levene, 1999; Cooley, 2000; Kosala & Blockeel,
2000; Madria, Bhowmick, Ng, & Lim, 1999).
Web mining is a complete process and involves
specific primary data mining tasks, namely data
collection, data reprocessing, pattern discovery,
and knowledge post-processing. Therefore, Web
mining can be viewed as consisting of the fol-

1080

Information RetrievalIR (Resource

Discovery): It deals with automatic retrieval
of all relevant documents, while at the same
time ensuring that the non relevant ones are
fetched as few as possible. The IR process
mainly deals with document representation, indexing, and searching. The process
of retrieving the data that is either online
or offline from the text sources available
on the Web such as electronic newsletters, newsgroups, text contents of HTML
documents obtained by removing HTML
tags, and also the manual selection of Web
resources. Here are also included text resources that originally were not accessible
from the Web but are accessible now, such
as online texts made for search purposes
only, text databases, and so forth.
Information ExtractionIE (Selection
and Pre-Processing): Once the documents
have been retrieved in the IR process, the
challenge is to automatically extract knowledge and other required information without
human interaction. IE is the task of identifying specific fragments of a single document
that constitute its core semantic content and
transforming them into useful information.
These transformations could be either a kind
of pre-processing such as removing stop
words, stemming, etc. or a pre-processing
aimed at obtaining the desired representation such as finding phrases in the training
corpus, transforming the presentation to
relational or first-order logic form, and so
forth.
Generalization (Pattern Recognition and
Machine Learning): Discover general patterns at individual Web sites or across multiple sites. Machine learning or data mining
techniques are used for the generalization.
Most of the machine learning systems,
deployed on the Web, learn more about the

Web Mining for Public E-Services Personalization

users interest than the Web itself.

Analysis (Validation and Interpretation):
A data driven problem, which presumes
that there is sufficient data available, so
that potentially useful information can be
extracted and analyzed. Humans also play an
important role in the information or knowledge discovery process on the Web, since
the Web is an interactive medium. This is
especially important for validation and/or interpretation but under Etzionis view (1996)
of Web mining, manual (interactive, query
triggered) knowledge discovery is excluded
and thus the focus is placed on automatic
data-triggered knowledge discovery.

Web mining refers to the overall process of

discovering potentially useful and previously
unknown information, knowledge, and patterns
from Web data. In this sense, it implicitly covers
the standard process of knowledge discovery
in databases (KDD) and can be considered as
a KDD extension applied to the Web (Markellos, Markellou, Rigou, & Sirmakessis, 2004a).
Specifically, Web mining can be categorized into
three areas of interest based on which part of the
Web is mined:

Web Content Mining: Focuses on the

discovery/retrieval of useful information
from Web contents/data/documents. Web
content data consist of unstructured data
(free texts), semi-structured data (HTML
documents) and more structured data (data
in tables, DB generated HTML pages)
Web Structure Mining: Focuses on the
structure of the hyperlinks within the Web
as a whole (inter-document) with the purpose
of discovering its underlying link structure.
Web structure data consist of the Web site
structure itself
Web Usage Mining: Mines the secondary
data derived from Web surfers sessions or
behaviors and focuses on techniques that

could predict user behavior while the user

interacts with the Web (Cooley, 2000). Web
usage data can be server access logs, proxy
server logs, browser logs, user profiles, registration data, user sessions or transactions,
cookies, user queries, bookmark data, mouse
clicks and scrolls, and any other data as the
result of interactions
Recently, Web usage mining (Srivastava,
Cooley, Deshpande, & Tan, 2000) has been
proposed as an underlying approach for Web
personalization (Mobasher, Cooley, & Srivastava,
2000). The goal of Web usage mining is to capture
and model the behavioral patterns and profiles of
users interacting with a Web site. The discovered
patterns are usually represented as collections of
pages or items that are frequently accessed by
groups of users with common needs or interests.
Such patterns can be used to better understand
behavioral characteristics of visitors or user segments, improve the organization and structure of
the site, and create a personalized experience for
visitors by providing dynamic recommendations.
In particular, techniques such as clustering, association rule mining, and navigational pattern
mining that rely on online pattern discovery from
user transactions can be used to improve the scalability of collaborative filtering when dealing with
clickstream and e-government data.

WEB MINING TECHNIQUES IN

E-PUBLIC SERVICES
For the implementation and successful operation
of e-government, the proper design, which will
be the basis in order to receive a series of strategic, administrative, and operational benefits, is
necessary. The application of e-government in
the public domain can be gradually performed in
fourteen (14) levels (Markellou, Panayiotaki, &
Tsakalidis, 2003). This allows the unobstructed
flow of information from/to the public sector and

1081

Web Mining for Public E-Services Personalization

gives the possibility not only to the citizens but

also to the enterprises (private sector) to acquire
better access in the services that state provides.
One of these levels is the upgrade of portal with
applications adjusted to every user, where Web
mining techniques may be applied to improve
access to information through the provided eservices (Markellou et al., 2004a; Markellou,
Rigou, & Sirmakessis, 2004b).
Specifically, the deployment of Web mining in
the e-government domain relates to the analysis of
citizen behavior and the production of adequate
adaptations. For example, given a specific citizen,
the presentation of required information from
an e-government portal can be tailored to meet
individual needs and preferences by providing
personal recommendations on topics relative to
those already visited. This process is typically
based on a solid user model, which holds up-todate information on dynamically changing citizen
behavior. This enables on-the-fly portal content
assembly, addressing exactly what the citizen
needs to know without wasting time on topics the
user is already proficient or not interested in. The
flowchart of this procedure is shortly illustrated
in Figure 1. E-government application constructs
users profiles integrating various sources of data,
pre-processes the data and applies Web mining
techniques to provide the users with personalized
Web experiences.

In order to personalize an e-government site,

the system should be able to distinguish between
different users or groups of users. This process is
called user profiling and its objective is to create
an information base that contains the preferences,
characteristics, and activities of the users. In the
Web domain, user profiling has been developed
significantly, since Internet technologies provide
easier means of collecting data about the users
of a Web site, which in the case of e-government
sites are citizens that must be satisfied by the
provided services. A user profile can be either
static, when the information it contains is never
or rarely altered (e.g., demographic information),
or dynamic when the user profiles data change
frequently. Such information is obtained either
explicitly (e.g., preferences, background, etc.)
using online registration forms and questionnaires, or implicitly, by recording the navigational
behavior and other users actions from server logs,
cookies, and so forth. User profiling is extremely
useful to G2C (government-to-consumer) and
G2B (government-to-business) applications. For
example, a public authority, such as a Ministry
of Finance and Economy, can customize its
information/services concerning chemical and
customs procedures to the relevant actors that
import chemical goods. Using this aspect, the
actors are facilitated in completing the necessary

Figure 1. Application of Web mining techniques in e-government domain

Personalized W eb
experiences

E-government
information a nd
services

1082

Web Mining for Public E-Services Personalization

procedures without getting lost in the maze of

information provided through a multi-field covering e-government Web site.
Another technique that can be used is the one
of clustering. Page clustering identifies groups of
pages that seem to be conceptually related according to the users perception. User clustering
results in groups of users that seem to behave
similarly when navigating through a Web site.
Such knowledge can be used in e-government in
order to perform public services segmentation.
Classification technique can be applied after
clustering in order to assign a new user to the
defined groups. It uses features with high discriminative ability for defining the various profiles, for example, the profile of an active citizen
may include the following values: Sex=male,
34<=Age<=40, Job=worker, Education= basic,
MaritalStatus=marital, NumberOfChildren=4,
and so forth. This knowledge can be used in applying personalization to e-government services
and better supporting the needs of the users providing the right information, to the right people,
at the right time.
For discovering relations between different
types of available information in an e-government environment association rules can be applied. This technique may identify correlations
between pages/users/services or other types of
items, not directly connected and reveal previously unknown associations between groups of
such items with specific similarities. The form of
an association rule can be 65% of citizens that
their MaritalStatus=marital search e-government
portal for information about LifeEpisodes=having
a baby or 40% of the citizens who accessed help
desk asked about online filling-in of tax returns
and VAT. For example, the last rule may indicate
that this information is not easily accessible or
explanatory enough and requires redesign tasks
from the portal technical group.
An extension of the previous technique comprises the sequential pattern discovery that can
be used for revealing patterns of co-occurrence,

which incorporates the notion of time. For example a pattern may be a Web page or a set of
pages accessed immediately after another set of
pages: 55% of new businesses who apply for a
certain certificate will use the certificate within
15 days or Given the transactions of a citizens
who has not apply for any information/services
during the last 3 months, find all citizens with a
similar behavior.
Finally, as search engines often appear as a
helpful tool at e-government, personalized Web
search systems may be used to enhance their functionality. In order to incorporate user preferences
into search engines, three major approaches are
proposed (Shahabi & Chen, 2003):

Personalized Page Importance: Modern

search engines employ the importance
scores of pages for ranking the search results, as well as traditional text matching
techniques.
Query Refinement: A process composed
of three steps: obtaining user profiles from
user, query modification and refinement.
Personalized Metasearch Systems:
Metasearch systems could improve the
retrieval rate by merging various ranked
results from multiple search engines into
one final ranked list.

FUTURE TRENDS
On the road to enhance an e-government application and treat each user individually, personalization plays a central role. The benefits for both
public authorities and citizens are significant
when it really works. However, several issues still
remain unclear. First of all, determining and delivering personalization is a data intensive task and
requires numerous processing steps. This usually
causes intolerably long response times, which in
turn may lead to site abandonment. To avoid this
constrain, parts of the process can be executed

1083

Web Mining for Public E-Services Personalization

offline or special algorithms and structures can

be used to guarantee fast online operation.
Another challenge is to ensure personalization
accuracy. It is true that unsuccessful recommendations can slow down the process, confuse, and
disorientate users. It is preferable not to deliver
any recommendations than deliver a set of useless
or harmful ones. Apart from that, personalization
should be delivered in the appropriate way (avoiding user intrusion and loss of concentration) and
not deprive users control over the whole process.
Moreover, as e-government sites are dynamic
environments, issues concerning the content or
structure updating e.g. newly added topics, pages,
services, etc. can be taken into consideration.
Last but not least, privacy violation during
the user profiling process should be encountered
(Volokh, 2000). Many users are reluctant to giving away personal information either implicitly
as mentioned before, or explicitly, being hesitant
to visit Web sites that use cookies (if they are
aware of their existence) or avoiding to disclose
personal data in registration forms. In both cases,
the user loses anonymity and is aware that all
of his actions will be recorded and used, often
without his consent. Additionally, even if a user
has agreed to supply personal information to a
site, through cookie technology such information
can be exchanged between sites, resulting to its
disclosure without his permission. Although
the new technologies and products for protecting users privacy on computers and networks
are becoming increasingly popular, none can
guarantee absolutely secure communications.
Electronic privacy issues in the foreseeable future
will become highly crucial and intense (Markellos et al., 2004b).

CONCLUSION
Governments enhance their attempt to offer efficient, advanced and modern services to their users
(citizens and businesses) based on information and

1084

communication technologies and especially the

Web. The remarkable acceptance of this powerful
tool has changed the way of conducting various
activities and offers citizens, businesses and public
authorities limitless options and opportunities.
However, the emerging problem to deal with is
the way an e-government can provide its users
with the right information and service according
to their specific needs and preferences. To this
direction, Web mining and personalization are
used for supporting tailored Web experiences.
These techniques appear as the most promising
for the future, since they help to establish one-toone relationships between users and governments,
improve the performance of provided information
and services, increase users satisfaction and promote e-loyalty. On the other hand, governments
take advantage of them, as long as they save costs
(e.g., transactions, communication, task management, etc.), improve response times, automate
various processes, provide alternative channels
of cooperation and communication, and upgrade
and modern their profile and image.
Many research and commercial approaches,
initiatives and tools are available, based on Web
site structure and contents, users navigation, behavior and transaction history, server log files, and
so forth. However, personalization requires rich
data in order to provide successful output. This
is not always feasible, since many users are often
negative towards the idea of being stereotyped.
Moreover, individuals privacy has been put in
jeopardy by the tasks of recording their activities
and saving the appropriate data into their profiles.
Summarizing, governments should work hard in
the direction of providing the legal framework
for ensuring the protection of users privacy and
also eliminating the possibility of misuse their
personal information.

Web Mining for Public E-Services Personalization

REFERENCES
Borges, J., & Levene, M. (1999, August 15). Data
mining of user navigation patterns. Proceedings
of the WEBKDD99 Workshop on Web Usage
Analysis and User Profiling, San Diego, CA (pp.
31-36).
Cooley, R. (2000). Web usage mining: Discovery
and application of interesting patterns from Web
data. PhD Thesis, Department of Computer Science, University of Minnesota.
Eirinaki, M., & Vazirgiannis, M. (2003). Web
mining for Web personalization. ACM Transactions on Internet Technology, 3(1), 1-27.
Etzioni, O. (1996). The world wide Web: Quagmire or Gold Mine. Communications of ACM,
39(11), 65-68.
Internet World Stats. (2005). Internet Usage StatisticsThe Big Picture. World Internet Users and
Population Stats. Retrieved June 25, 2005, from
http://www.internetworldstats.com/stats.htm
Kosala, R., & Blockeel, H. (2000). Web mining research: A survey. SIGKDD Explorations:
Newsletter of the Special Interest Group (SIG)
on Knowledge Discovery & Data Mining, ACM,
2(1), 1-15.

Conference on the Social and Ethical Impacts of

Information and Communication Technologies,
Challenges for the Citizen of the Information
Society, Syros, Greece.
Markellou, P., Panayiotaki, A., & Tsakalidis, A.
(2003, June 3-6). E-government and applications
levels: Technology at citizen service. Proceedings
of IADIS International Conference, E-Society,
Lisbon, Portugal (pp. 849-854).
Markellou, P., Rigou, M., & Sirmakessis, S.
(2004a). Mining for Web personalization. In A.
Scime (Ed.), Web mining: Applications and techniques (pp. 27-48). Idea Group Publishing.
Markellou, P., Rigou, M., & Sirmakessis, S.
(2004b). Web personalization for e-marketing
intelligence. In S. Krishnamurthy (Ed.), Contemporary research in e-marketing: Volume 1 (pp.
232-250). Idea Group Publishing.
Mobasher, B., Cooley, R., & Srivastava, J. (2000).
Automatic personalization based on Web usage
mining. Communications of the ACM, 43(8),
142-151.
Pal, S. K., Talwar, V., & Mitra, P. (2002). Web
mining in soft computing framework: Relevance,
state of the art and future directions. IEEE Transactions on Neural Networks, 13(5), 1163-1177.

Madria, S. K., Bhowmick, S. S., Ng, W. K., &

Lim, E. P. (1999, August 30-September 1). Research issues in Web data mining. Proceedings
of Data Warehousing and Knowledge Discovery,
Florence, Italy (pp. 303-312).

Shahabi, C., & Chen, Y. S. (2003, September 2224). Web information personalization: Challenges
and approaches. Proceedings of Databases in
Networked Information Systems: The 3r d International Workshop, Japan (pp. 5-15).

Markellos, K., Markellou, P., Rigou, M., & Sirmakessis, S. (2004a). Web mining: Past, present,
and future. In S. Sirmakessis (Ed.), Text mining
and applications (pp. 25-35). Berlin; Heidelberg,
Germany: Springer Verlag.

Srivastava, J., Cooley, R., Deshpande, M., & Tan,

P. N. (2000). Web usage mining: Discovery and
applications of usage patterns from Web data.
SIGKDD Explorations: Newsletter of the Special
Interest Group (SIG) on Knowledge Discovery &
Data Mining, ACM, 1(2), 12-23.

Markellos, K., Markellou, P., Rigou, M., Sirmakessis, S., & Tsakalidis, A. (2004b, April 14-16).
Web personalization and the privacy concern.
Proceedings of the 7t h ETHICOMP International

Volokh, E. (2000). Personalization and privacy.

Communications of the ACM, 43(8), 84-88.

1085

Web Mining for Public E-Services Personalization

Key Terms
Clickstream: It is a record of a users activity on the Internet, including every Web site and
every page of every Web site that the user visits,
how long the user was on a page or site, in what
order the pages were visited, any newsgroups
that the user participates in and even the e-mail
addresses of mail that the user sends and receives.
Both ISPs and individual Web sites are capable
of tracking a users clickstream.
Cookie: The data sent by a Web server to a
Web client, stored locally by the client and sent
back to the server on subsequent requests. In other
words, a cookie is simply an HTTP header that
consists of a text-only string, which is inserted into
the memory of a browser. It is used to uniquely
identify a user during Web interactions within
a site and contains data parameters that allow
the remote HTML server to keep a record of the
user identity, and what actions she/he takes at the
remote Web site.
Data Mining: The application of specific
algorithms for extracting patterns (models) from
data.

Knowledge Discovery in Databases (KDD):

The nontrivial process of identifying valid, novel,
potentially useful, and ultimately understandable
patterns in data.
Server Log: Web servers maintain log files
listing every request made to the server. With log
file analysis tools, its possible to get a good idea
of where visitors are coming from, how often
they return, and how they navigate through a site.
Using cookies enables Webmasters to log even
more detailed information about how individual
users are accessing a site.
Web Mining: The discovery and analysis of
useful information from the Web.
Web Personalization: It is the process of customizing a Web site to the needs of specific users,
taking advantage of the knowledge acquired from
the analysis of the users navigational behavior
(usage data) in correlation with other information
collected in the Web context, namely structure,
content and user profile data. Due to the explosive
growth of the Web, the domain of Web personalization has gained great momentum both in the
research and the commercial area.
Web Usage Mining: The application of data
mining techniques to Web clickstream data in
order to extract usage patterns.

This work was previously published in Encyclopedia of Digital Government, edited by A. Anttiroiko and M. Malkia, pp. 1629-1634,
copyright 2007 by Information Science Reference, formerly known as Idea Group Reference (an imprint of IGI Global).

1086

1087

Chapter 3.6

Context-Aware Service
Discovery in Ubiquitous
Computing
Huaqun Guo
Institute for Infocomm Research and National University of Singapore, Singapore
Daqing Zhang
Institute for Infocomm Research, Singapore
Lek-Heng Ngoh
Institute for Infocomm Research, A*STAR, Singapore
Song Zheng
Institute for Infocomm Research, Singapore
Wai-Choong Wong
National University of Singapore, Singapore

INTRODUCTION
The decreasing cost of networking technology
and network-enabled devices is driving the large
scale deployment of such networks and devices
so as to offer many new and innovative services
to users in ubiquitous computing. For example,
when you carry your mobile laptop or personal
digital assistant (PDA) around, or drive on the
road, various services have been made available,
ranging from finding a local printer to print a

file, to instantaneously knowing about the traffic

situation from traffic-cameras and other sensors
along a highway.
To achieve the above, every participating network-enabled end-device must solve an interesting technical problem, i.e., to locate a particular
network service or device out of hundreds of
thousands of accessible services and devices.
Such service advertising and discovery is important as mobile devices and mobile wireless
devices proliferate on networks. For this reason,
a service discovery and advertising protocol is an

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Context-Aware Service Discovery in Ubiquitous Computing

important tool to help these devices find services

on the network wherever they connect, and to let
other network users know about the services they
are offering.
Context-aware service discovery, on the other
hand, would help users to find services that are
most appropriate based on fast-changing client conditions, such as location. For example,
most laptops are statically configured to print
to dedicated office printers. With the help of the
context-awareness, a laptop could find the nearest
accessible printer attached to the network that the
laptop is currently plugged into.

SERVICE DISCOVERY PROTOCOLS

As new services are made available, clients would
be able to obtain services based on a set of clientdefined keywords or attributes. These attributes
would allow the client to narrow the scope of the
search so that only those services that pertain
to its needs would be discovered. Furthermore,
the client would be able to automatically start
interacting with the newly discovered services
without any programming (Intel).
For the past few years, competing industries
and standards communities have been hotly pursuing the technologies for service discovery. Espeak (HP, 2002), UDDI (Universal Description,
Discovery, and Integration), Suns Jini (Arnold,
ed. 2001), Microsofts Universal Plug and Play
(UPnP), IBMs Salutation (Salutation Architecture
Specification), Service Location Protocol (SLP) of
IETF (Guttman, 1999), and Bluetooths Service
Discovery Protocol (SDP) (Helal, 2002) are among
the front-runners in this new race. E-speak and
UDDI are designed specifically for discovering
Web services. On the other hand, Jini, UpnP, and
Salutation are geared toward services furnished
by hardware devices such as printers, faxes, etc.
We elaborate each of these in detail next.

1088

E-speak is an open software platform

designed by HP (Hewlett-Packard Co.) to

facilitate the delivery of e-services (electronic services) over the Internet. Based on
Extensible Markup Language (XML), the
E-speak Service Framework Specification
makes it possible for e-services to advertise,
discover, negotiate, and form contracts, learn
each others interfaces and protocols, and
invoke each others services, all without
human intervention.
UDDI protocol is one of the major building
blocks required for successful Web services.
UDDI creates a standard interoperable
platform that enables companies and applications to quickly, easily, and dynamically
find and use Web services over the Internet.
UDDI is a cross-industry effort driven by
major platform and software providers, as
well as marketplace operators and e-business
leaders within the OASIS (Organization for
the Advancement of Structured Information
Standards) consortium.
Jini is a distributed service discovery and
advertisement architecture that relies on
mobile code and leverages the platform
independent of Java language. Jini entities
consist of services, lookup servers that
catalog available services, and clients that
require services. All service advertisements
and requests go through the lookup servers.
To register service availability or to discover services, a service or client must first
locate one or more lookup servers by using
a multicast request protocol. This request
protocol terminates with the invocation of
a unicast discovery protocol, which clients
and services use to communicate with a
specific lookup server. A lookup server can
use the multicast announcement protocol to
announce its presence on the network. When
a lookup server makes an announcement,
clients and services that have registered
interest in receiving announcements of new
lookup services are notified.

Context-Aware Service Discovery in Ubiquitous Computing

UPnP is considered at some levels to be a

natural extension of Microsoft Plug and Play
to the networking scenario. The UPnP specification describes device addressing, service
advertisement and discovery, device control,
event, and presentation. AutoIP (Troll, 2000)
protocol is used to allow devices to dynamically claim IP addresses in the absence of
a DHCP server. UPnP uses simple service
discovery protocol (SSDP) for service discovery. SSDP uses HTTP over multicast
(HTTPMU) and unicast UDP (HTTPU) to
announce a devices presence to others and
discover other devices and services. Each
service can be searched by any of its three
associated IDsservice type, service name,
and location. An UPnP service description includes a list of actions to which the
service responds and a list of variables that
model the services state at runtime. After
discovery, a control point sends a message
to the services specified control object URL
(Uniform Resource Locator) according to
the Simple Object Access Protocol (SOAP).
The device or service returns action-specific
values. A control point can also subscribe to
receive event notifications which are updates
of services state in a format of General
Event Notification Architecture (GENA).
Salutation is an architecture for service
advertisement, discovery, and invocation
among devices and services of dissimilar
capabilities. The architecture is composed of
three fundamental components: functional
units, Salutation managers, and transport
managers. In Salutation, the concept of a
service is broken down into a collection of
functional units with each unit representing
some essential feature. A service description
is then a collection of functional unit descriptions, each having a collection of attribute
records (name, value, etc.). These records
can be queried and matched against during
the service discovery process. The Saluta-

tion manager contains a registry to keep

information about services, and a client can
register or un-register itself. An ensemble of
Salutation managers can coordinate with one
another to exchange registration information, even with those on different transport
media. Communication between Salutation
managers is based on remote procedure call
(RPC). A discovery request is sent to the local
Salutation manager, which in turn will be
directed to other Salutation managers. This
cooperation among Salutation managers
forms a conceptually similar lookup service
to Jini. One difference, though, is that it is
distributed over the network. The transport
manager isolates the implementation of the
Salutation managers from particular transport-layer protocols and hence gives Salutation network transport independence.
SLP is an Internet Engineering Task Force
(IETF) standard for decentralized, lightweight, and extensible service discovery. It
uses service URLs, which define the service
type and address for a particular service. For
example, service:printer:lpr://hostname is
the service URL for a line printer service
available at hostname. Based on the service
URL, users (or applications) can browse
available services in their domain and select
and use the one they want. SLP establishes
a framework for resource discovery that
includes three agents that operate on
behalf of the network-based software: User
Agents (UA), Service Agents (SA), and Directory Agents (DA). UAs perform service
discovery on behalf of client software. SAs
broadcast the advertisement of the location
and attributes on behalf of services. As a
centralized service information repository,
the DA caches advertisements from SAs and
processes discovery queries from UAs.
Bluetooth SDP is specific only to Bluetooth
devices (Specification of the Bluetooth
System). It defines a very simple service

1089

Context-Aware Service Discovery in Ubiquitous Computing

discovery mechanism for peer-to-peer

type networks. It does not provide access
to services, brokering of services, service
advertisements, or service registration, and
there is no event notification when services
become unavailable (Helal, 2002). Since
there is no centralized directory for service
registration, the attributes about a service are
stored in the local service discovery server
in each mobile device. Clients can search for
available services either by service classes,
which correspond to unique 128-bit identifiers and uniquely identify types of devices,
or by matching attributes. The server replies
whether a service is available on the queried
platform only and a higher level protocol
must provide specific actions to utilize the
service.
All of the above mentioned discovery mechanisms have their pros and cons and are valuable

within their own infrastructures. The comparison

of service discovery protocols is shown in Table
1. Most of them promise similar functionalities
which are targeted at reducing configuration
hassles, improving device cooperation, and
automating discovery of required services. However, due to the lack of a widely adopted service
discovery mechanisms, these different products
have different APIs and do not work together. In
addition, current service discovery frameworks
do not consider important context information.
Therefore, none of them offer a universal solution
to context-aware service discovery.

CONTEXT-AWARE SERVICE
DISCOVERY
The basic areas we need to address in order to
create a context-aware service discovery architecture are:

Table 1. Comparison of service discovery protocols

Developer

E-Speak

UDDI

Jini

UPnP

Salutation

SLP

Bluetooth SDP

HP

OASIS
standards
consortium

Sun

Microsoft

Salutation
consortium

IETF

Bluetooth SIG

Hardware
devices

Hardware
devices

Hardware
devices

Only Bluetooth
devices

Service type

Web
services

Web
services

Hardware
devices
with JVM
(Java Virtual
Machine)

Central cache
repository

Yes

Yes

Yes (lookup
server)

No

Yes

Yes

No

Operation w/o
directory

Lookup table
required

No

Yes

Yes

Yes

Network
transport

TCP/IP

TCP/IP

Independent

TCP/IP

Independent

TCP/IP

Bluetooth
communication

Programming
language

XML

Independent

Java

Independent

Independent

Independent

Independent

OS and
platform

Independent

Dependent

JVM

Dependent

Dependent

Dependent

Dependent

1090

Context-Aware Service Discovery in Ubiquitous Computing

1.

2.

3.

Interoperability: Our proposed architecture can interoperate with and provide

value added functionality to existing service
discovery mechanisms.
Context-awareness: It will acquire and
search for context, and carry out context
reasoning to find the most appropriate services for users/applications.
Security, privacy, and trust: All of this
must be done in a manner that provides a
sufficient level of security, privacy, and trust
to meet the customers needs.

Based on the above requirements, we propose

a context-aware service discovery architecture
as shown in Figure 1. In order to interoperate
with existing service discovery mechanisms,
we adopt OSGi (Open Service Gateway Initia-

tive) architecture (Marples, 2001; Dobrev, 2002;

Kim, 2001) to accommodate the different service
collaboration.
Founded in March 1999, the OSGi creates a
set of open specifications for the delivery of a
wide array of services to end users. The OSGi
mission is to enable the deployment of services
over wide area networks to local networks and
devices (Marples, 2001). The key benefits of OSGi
are the following:

Platform independent. Based on Java technologies, the OSGi gateway can run on most
operating systems.
Multiple network technology support. Home
network technologies, such as Jini and UPnP,
facilitate device and service discovery in the
OSGi gateway (Dobrev, 2002).

Figure 1. Context-aware service discovery architecture

Users/Applications

C ontext-A war e Ser vice Discover y

Appropriate Ser vices
Secur ity/Pr ivacy/T r ust

Knowledge Base

C ontext R easoning
Ser vices R epr esentation

OSGi
W r apper

W r apper

W r apper

1091

Context-Aware Service Discovery in Ubiquitous Computing

Multiple services support and service

collaboration. Multiple applications from
different service providers are able to run
on a single service platform and provide
functionality to other services.

Interoperability
The benefits of OSGi provide a feasible interoperation platform for different service discovery
mechanisms. Each of the existing service discovery mechanisms has a wrapper to communicate
with the OSGi. Wrappers transform the functions
of those entities into the form of services and
publish those services in the OSGi framework
service registry. In the OSGi framework, entities discovered by existing service discovery
protocol are represented in the form of services
in a standardized way. As the functions of every
entity are described using common convention,
entities can thus understand each other and collaborate to achieve a certain goal.

Context-Awareness
Context information plays an important role in
making the physical spaces smart. Users and
applications often need to be aware of their surrounding context and adapt their behaviors to
context changes. Context-awareness involves
the use of context to provide relevant services to
the user, where relevancy depends on the users
task (Dey, 2000).
An appropriate model should address different
characteristics of context information, such as
dependency and uncertainty. Context acquisition
is closely coupled with sensors to acquire context
data from physical or virtual sensors. In our earlier work, we have proposed an ontology-based
context model to describe context information
in a semantic way, which exhibits features such
as expressiveness, extensibility, ease of sharing
and reuse, and logic reasoning support (Zhang,
2006). Context search provides users and applica-

1092

tions both synchronous context query service and

asynchronous context event notification service. A
context processing and management engine called
Semantic Space has also been implemented to
support context reasoning (Wang, 2004). Context
reasoning uses context to carry out logical reasoning to provide the most appropriate services to
the users/applications.

Security, Privacy, and Trust

In the service discovery, the interaction and information exchange between clients and services
must be secure, private, and trustworthy. OSGi
defines SSL (Secure Sockets Layer) service for
secure network communication and enforces
policy-based fine-grained control mechanisms
to permit local resource access. The relevant
techniques will be exploited and the research
works in this area are still ongoing as shown in
orange color in Figure 1. We will elaborate more
in the future trends.

FUTURE TRENDS
The pervasive availability of embedded devices
in the environment imposes significant technical
and research challenges in service cooperation
and interaction. One aspect that captures more
attention at the present is security, privacy, and
trust which are explained briefly next.
Security includes the three main properties of
confidentiality, integrity, and availability (Stajano,
2002). Confidentiality is concerned with protecting the information/service from unauthorized
access; Integrity is concerned with protecting the
information/service from unauthorized changes;
and Availability is concerned with ensuring that
the information/service remains accessible.
Privacy is the claim of individuals, groups,
or institutions to determine for themselves when,
how, and to what extent information is communicated to others. Privacy is about protecting

Context-Aware Service Discovery in Ubiquitous Computing

users personal information. Privacy control

relates not only to the process of setting rules
and enforcing them, but also to the way privacy
is managed/controlled adaptively according to
changes in the degree of disclosure of personal
information or user mobility from one space to
another (Cook, 2005).
Trust management is to develop a trust
specification that can be analyzed and evaluated
before appropriate interactions/transactions really
start. As clients and services often interact and
collaborate with each other in an ad-hoc manner, those clients and services may come from
unfamiliar administrative domains and therefore
be completely unknown to each other. To safely
take advantage of all the possibilities, it is essential to assess the confidence level of involved
parties, estimate the likely behavior of entities,
and recommend a certain level for interactions,
which is so called trust management. Trust
management involves the trust model, the trust
and recommendation specifications for different
entities from different certified authorities, and
risk thresholds. The trust management service
uses the trust model to compute the trust value
for each interaction. The trust value indicates the
degree of confidence and is used as a parameter to
determine the access level of services or degree
of information interaction or exchange.
From the above elaboration, we may predict
that the future trends for context-aware service discovery will involve addressing the issues related to
security, privacy, and trust. The techniques related
to authentication, authorization, cryptography,
and trust management will be exploited for the
interaction and information exchange between
clients and services.

CONCLUSION
This paper surveys the existing service discovery
protocols, such as E-speak, UDDI, Jini, UPnP,
Salutation, SLP, and SDP, and discusses the dif-

ferent mechanisms. All of the above mentioned

mechanisms have their pros and cons, but none
of them offer a universal solution to contextaware service discovery. Next, we identified the
requirements of service discovery mechanisms
and described a context-aware service discovery architecture to support the interoperability
and context-awareness which are required in
ubiquitous computing. Finally, the future trends
for context-aware service discovery will involve
addressing the issues related to security, privacy,
and trust.

REFERENCES
Arnold, K. (Ed.). (2001). The Jini specifications
(2nd edition). Addison-Wesley.
Cook, D. J., & Das, S.K. (2005). Smart environments technologies, protocols and application.
NJ: John Wiley & Sons.
Dey, A.K. (2001). Understanding and using context. Personal and ubiquitous computing. Special
issues on Situated Interaction and Ubiquitous
Computing.
Dey, A.K., & Abowd, G.D. (2000). Towards a
better understanding of context and contextawareness. CHI 2000 Workshop on the What,
Who, Where, When and How of Context-Awareness, April.
Dobrev, P., et al. (2002). Device and service
discovery in home networks with OSGi. IEEE
Communication Magazine, 86-92.
Guttman, E., Perkins, C.,Veizades, J., & Day,
M. (1999). Service location protocol Version 2.
IETF RFC 2608, June. http://www.rfc-editor.
org/rfc/rfc2608.txt
Helal, S. (2002). Standards for service discovery
and delivery. IEEE Pervasive Computing, July/
September.

1093

Context-Aware Service Discovery in Ubiquitous Computing

HP. (2002). E-speak. Retrieved from http://

searchcio.techtarget.com/sDefinition/0,,sid19_
gci503994,00.html
Intel, Dynamic Service Discovery Executive
White Paper (UDDI Enhancement Proposal).
Retrieved from http://www.intel.com/cd/ids/developer/asmo-na/eng/20409.htm?page=1
Kim, J.-H., Yae, S.-S., & Ramakrishna R.S. (2001).
Context-aware application framework based on
open service gateway. International Conference
on Info-tech and Info-net (pp. 200-204).
Marples, D., & Kriens, P. (2001). The open services
gateway initiative: An introductory overview.
IEEE Communication Magazine, December,
110-114.
Salutation Architecture Specification. http://www.
salutation.org/specordr.htm
Sen, R., Handorean, R., Roman, G.C., & Gill, C.
(2005). Service oriented computing imperatives in
ad hoc wireless settings, service oriented software
engineering: Challenges and practices. Hershey,
PA: Idea Group Publishing.
Specification of the Bluetooth System. http://www.
bluetooth.com/developer/specification/specification.asp
Stajano, F. (2002). Security for ubiquitous computing. Wiley.
Troll, R. (2000). Automatically choosing an IP
address in an ad-hoc IPv4 network. IETF Draft
Draft-ietf-dhc-ipv4-autoconfig-05.txt. http://ftp.
gnus.org/internet-drafts/draft-ietf-dhc-ipv4-autoconfig-05.txt
UDDI (Universal Description, Discovery and
Integration). http://www.uddi.org/
UPnP, Universal Plug and Play specification v1.0.
http://www.upnp.org/
Wang, X.H., Zhang, D.Q., Dong, J.S., Chin, C.Y.,
& Hettiarachchi, S.R.. (2004). Semantic space: A

1094

semantic Web infrastructure for smart spaces.

IEEE Pervasive Computing, 3(2).
Zhang, D.Q., Cheng, H.S., Gu, T., Zhu, M.L.,
Guo, H.Q., & Leong, M.K.. (2006). A scalable and
trusted middleware for smart spaces. Accepted in
Fourth International Conference On Smart homes
and health Telematics (ICOST 2006), Northern
Ireland, UK.

Key TERMS
Client: An application that is interested in or
requires some other application to perform some
type of work for the client (Intel).
Context: Any information that can be used to
characterize the situation of an entity. An entity
is a person, place, or object that is considered
relevant to the interaction between a user and an
application, including the user and the application
themselves (Dey, 2001).
Context-Awareness: To use context to provide
relevant services to the user, where relevancy
depends on the users task (Dey, 2000).A service
is a component or application that performs the
work on behalf of a requesting application or
client (Intel).
Service Advertisement: Is responsible for advertising a given service description on a directory
service or directly to other hosts in the network.
The effectiveness of an advertisement is measured
as a combination of the extent of its outreach and
the specificity of information it provides up front
about a service (Sen, 2005).
Service Description: Is responsible for
describing a service and the type of context
information in a comprehensive, unambiguous
manner that is machine interpretable to facilitate
automation and human readable to facilitate rapid
formulation by users (Sen, 2005).

Context-Aware Service Discovery in Ubiquitous Computing

Service Discovery: Is the keystone and carries

out three main functions. It formulates a request,
which is a description of the needs of a service
consumer. This request is formatted in a similar
manner to the service description. It also provides
a matching function that pairs requests to services
with similar descriptions, so that the service which
can best fulfill service consumers needs on peer
devices is selected. Finally, it provides a mechanism for the service consumer to communicate
with the service provider (Sen, 2005).
Service Invocation: Is responsible for facilitating the use of a service. Its functions include
transmitting commands from the service consumer to the service provider and receiving results. It

is also responsible for maintaining the connection

between the consumer and the provider for the
duration of their interaction. A good invocation
mechanism abstracts communication details from
the service consumer and, in the case of network
failure, redirects requests to another provider or
gracefully terminates (Sen, 2005).
Ubiquitous Computing (ubicomp, or sometimes ubiqcomp): Integrates computation into the
environment, and enables people to move around
and interact with information and computing naturally and casually. One of the goals of ubiquitous
computing is to enable devices to sense changes
in their environment and to automatically adapt
and act based on these changes according to user
needs and preferences.

This work was previously published in Encyclopedia of Internet Technologies and Applications, edited by M. Freire and M.
Pereira, pp. 119-125, copyright 2008 by Information Science Reference, formerly known as Idea Group Reference (an imprint
of IGI Global).

1095

1096

Chapter 3.7

Ontological Engineering in
Pervasive Computing
Environments
Athanasios Tsounis
University of Athens, Greece
Christos Anagnostopoulos
University of Athens, Greece
Stathes Hadjiefthymiades
University of Athens, Greece
Izambo Karali
University of Athens, Greece

Abstract
Pervasive computing is a broad and compelling
research topic in computer science that focuses
on the applications of technology to assist users
in everyday life situations. It seeks to provide proactive and self-tuning environments and devices
to seamlessly augment a persons knowledge and
decision making ability, while requiring as little
direct user interaction as possible. Its vision is
the creation of an environment saturated with
seamlessly integrated devices with computing
and communication capabilities. The realisation

of this vision requires that a very large number

of devices and software components interoperate
seamlessly. As these devices and the associated
software will pervade everyday life, an increasing number of software and hardware providers
will deploy functionality in pervasive computing
environments (PCE). That poses a very important
interoperability issue, as it cannot be assumed that
the various hardware and software components
share common communication and data schemes.
We argue that the use of Semantic Web technologies, namely the ontologies, present a intriguing
way of resolving such issues and, therefore, their

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Ontological Engineering in Pervasive Computing Environments

application in the deployment of PCE is a highly

important research issue.

Introduction
The vision of pervasive computing presents many
technical issues, such as scaling-up of connectivity requirements, heterogeneity of processors and
access networks and poor application portability over embedded processors. These issues are
currently being addressed by the research community; however the most serious challenges are
not technological but structural, as embedded
processors and sensors in everyday products
imply an explosion in the number and type of
organisations that need to be involved in achieving
seamless interoperability (OSullivan, 2003). In a
typical pervasive computing environment (PCE)
there will be numerous devices with computing
capabilities that need to interoperate (Nakajima,
2003). These devices might be of different vendors
and may operate based on different protocols.
Therefore, the key issue in deploying a PCE is
achieving application level interoperability. The
complexity of such a venture is considerable. It is
extremely difficult to reach agreements when the
players involved expand from all the hardware and
software providers (e.g., IBM, HP, Microsoft) to
all the organisations that will equip their products
with computing and communication capabilities
(e.g., coffee machines, refrigerators). Therefore,
we cannot rely on shared a priori knowledge based
on commonly accepted standards to resolve the
issue. Instead, software components must adapt
to their environment at runtime to integrate their
functionality with other software components
seamlessly. An intriguing way of resolving this
issue is the use of semantics, namely the use of
Semantic Web technologies such as ontologies.
In this manner, software entities provide semantically enriched specifications of the services
that they provide and the way they should be
invoked. Moreover, the data that are exchanged

are also semantically enriched, enabling the entities to reason and make effective decisions. This
is particularly important for the description of
contextual information, which is of main interest
in a PCE. As context we identify any information
that is, directly or indirectly, associated with any
entity in the environment.
The novelty of the Semantic Web is that the
data are required to be not only machine readable
but also machine understandable, as opposed to
todays Web which was mainly designed for human interpretation and use. According to Tim
Berners-Lee, the Director of World Wide Web
Consortium, the Semantic Webs goal is to be
a unifying system which will (like the Web for
human communication) be as un-restraining as
possible so that the complexity of reality can
be described (Berners-Lee, 2001). With the
realisation of a Semantic Web it would be easy
to deploy a wide range of services that would be
almost impossible to manage in the current Web.
Semantics enable developers to create powerful
tools for complex service creation, description,
discovery, and composition. The application
areas of the Semantic Web extend from knowledge repositories to e-commerce and from user
profiling to PCE.
New standards are being developed as a first
step in realising the Semantic Web. The Resource
Description Framework (RDF), which is a Web
mark-up language that provides basic ontological primitives, has been developed by the W3C
(Beckett, 2004). RDF is a language for representing meta-information about resources in the
World Wide Web. However, by generalising the
concept of a Web resource, RDF can also be
used to represent information about things that
can be identified on the Web, by means of URIs.
The DARPA Agent Markup Language + Ontology Inference Layer (DAML+OIL) extends RDF
with a much richer set of modelling primitives
(Rapoza, 2000). The DAML+OIL have been
submitted to W3C as a starting point for the Web
Ontology Working Group and led to the creation

1097

Ontological Engineering in Pervasive Computing Environments

of the standard for Web ontologies, namely the

Web Ontology Language (OWL) (McGuinness,
2004; Noy et al., 2001).
In this chapter, we survey the research that
has been carried out regarding ontology and
knowledge engineering and try to map the key
findings to the requirements of a PCE. We present standards and tools that are being used for
the development of ontologies. Furthermore,
we discuss efforts regarding the management of
multiple, overlapping, and evolving ontologies
and the semantic mappings among them. We
also depict how ontologies can help in searching mechanisms and in profiling users and data
sources. Moreover, due to the diverse nature of
PCE using interoperability standards is not a sufficiently scalable approach. Hence, we show the
important role that ontologies can play in achieving service interoperability, a key requirement
in PCE. Finally, we present the efforts that have
been carried out linking the pervasive computing
paradigm with ontologies and illustrate with an
example our vision that brings these two worlds
together in terms of semantics usage.
The rest of this chapter is structured as follows: In section 2, we outline our vision of a
service oriented PCE and some arguments in the
use of the ontology paradigm. Section 3 deals
with ontological engineering aspects by providing definitions related to ontologies, presenting
current Semantic Web standards for developing
ontologies and describing the work that has been
carried out regarding their management and manipulation. Section 4 introduces issues related to
how ontologies might be used to provide semantic
searching mechanisms related to information retrieval. In this section also, we give an overview of
ontology-based profiling. In section 5, we portray
how ontologies assist in providing semantic based
service interoperability and how ontologies may be
useful in a PCE. Section 6 outlines some several
examples of semantic-based service interoperability through ontologies. Finally in section 7,
we provide our conclusions and outline future
research directions.
1098

A Service-Oriented Pervasive
Computing Environment
Service-oriented architectures focus on application-level interoperability, in terms of well-established service ontologies (Martin, 2003; Fensel,
2002). This is achieved by means of well-defined
interfaces to various software components. In this
fashion, rapid integration of existing functionality
is achieved in the design and implementation of
new applications. Object Management Groups
(OMG) Common Object Request Broker Architecture (CORBA) (Orfali, 1998) is an example of
such widespread service-oriented architecture.
One of the most important issues in a PCE is the
establishment of ad-hoc relationships between
applications or between applications and devices.
As a result, the use of a common mechanism,
which will be able to provide interoperability in
a dynamic fashion, is essential. Furthermore, the
separation of the applications functionality from
the adopted communication schemes is of considerable importance, as it enables system developers to distinguish between service functionality
(e.g., functional and no-functional context-aware
service features, as referred in McIlraith et al.,
2001) and system functionality. Hence, the deployment of a PCE reduces to the problem of ad-hoc
service discovery, composition (e.g., WSFL,
service composition context-aware paradigm) and
execution with minimum a-priori knowledge of
the services functionality. While OWL-S is based
on the combination of OWL, WSDL and SOAP,
WSMO uses F-Logic and XML-based features
of Web Services (WS). Based on the previous
languages, significant research has been devoted
to semantic WS (e.g., ODE SWS [Gmez et al.,
2004], METEOR [Aggarwa et al., 2004]).
However, interoperability is a difficult task to
accomplish in a highly heterogeneous and volatile
environment. Therefore, for service-oriented architectures to be applicable a well-defined mechanism is required to describe the communication
schemes which are employed, provide data with

Ontological Engineering in Pervasive Computing Environments

semantics and describe the profiles and the policies

of the entities that constitute the PCE. Furthermore, ontologies may be employed for defining
entity-specific policies that govern the context
access, usage, and manipulation. They may also
provide abstract descriptions of the physical and
software components (i.e., conceptual modelling
via ontology paradigm) that comprise the PCE.
Hence, considerable flexibility in deploying such
environments is provided.
Accurate and expressive context models are
required to represent context in an efficient and
effective way, as the manipulation of the context will be based on its model. The model will
determine the variety of the actions that may be
applied on it, as well as the accuracy of the results.
Moreover, context knowledge and manipulation
enable the system to be context aware and proactive and employ more efficient decision making
mechanisms. Therefore, we consider the context
representation to be of great importance, since
formal context modelling (in terms of developing an object-centred or frame-based knowledge
manipulation system) achieves multirepresentation knowledge, by rendering heterogeneous
nature of PCEs. Moreover, model-mapping and
context multirepresentation techniques should
be employed to transform the whole context or
part of it from one form (i.e., representation formulae or dissimilar semantic conceptual model)
to another (e.g., [Ding et al., 2002]). As long as
the appropriate mappings exist, the entities may
exchange contextual information while preserving
their own context model (i.e., localised ontologies
[Bouquet, 2002]). Additionally, inference engines
(e.g., rule-based systems [Friedman, 2005]) may
be used to deduct previously unknown derivative
context from primitive contextual information. In
this fashion, a pervasive computing system is able
to produce knowledge and provide proactivity,
assisting the user in an unobstructive way. Last
but not least, the profile and policy management
for the descriptions of the physical and software
entities and the definition of the rules that govern

the context access, usage, and manipulation may

be based on the ontological engineering paradigm.
Therefore, we envisage an environment where all
entities interact with services through abstract
interfaces and argue that the use of Semantic
Web technologies is useful in achieving semantic
based service interoperability in PCEs.

Ontological Engineering
The word ontology comes from philosophy, where
it means a systematic explanation of being. In
the last decade, this word has become relevant
to the knowledge engineering community. The
authors in Guarino (1995) propose the words
Ontology (with capital o) and ontology to
refer to the philosophical and knowledge engineering concepts, respectively. There are a lot of
relevant definitions, but we keep only two terms
that are relevant, to some extent, to the pervasive
computing paradigm. In Neches et al. (1991) an
ontology is defined as follows:
An ontology defines the basic terms and relations
comprising the vocabulary of a topic area as well
as the rules for combining terms and relations to
define extensions to the vocabulary.
Such descriptive definition informs us what
to do to build an ontology by giving some vague
guidelines: This ontology description denotes how
to identify basic terms and relations between terms
and how to identify rules combining terms and
their relationships. Through such a definition one
can deduce that ontology does not include only
terms that are explicitly defined in, but also the
knowledge that can be inferred from it. In this
point, the authors in Guarino (1995) consider an
ontology as:
A logical theory, which gives an explicit, partial account of a conceptualisation, where conceptualisation is basically the idea of the world that a person

1099

Ontological Engineering in Pervasive Computing Environments

or a group of people can have. Consecutively, one

may consider that the ontology distinguishes terms
that are mainly taxonomies from dissimilar human
senses and models a specific domain in a deeper
way by providing well-documented restrictions on
such domain semantics.
In the pervasive computing area it can be
claimed that ontology is used to provide a vocabulary with explicitly defined and machine
comprehensible meaning. In multi-agent systems,
ontologies are essential for providing a common
vocabulary and a shared perception of domain
knowledge to allow communication in an open
environment. An ontology provides an explicit
specification of the structure of a certain domain and includes a vocabulary for referring to
the subject area, and a set of logical statements
expressing the constraints existing in the domain
and restricting the interpretation of the vocabulary.
Thus, the developing of ontologies, especially for
pervasive environments, provides the means by
which distributed software components share
common semantics of the terms being used for
communication and knowledge representation.

Ontology Languages for the

Semantic Web
As already mentioned, several standards for Web
ontologies have emerged. We will provide a brief
description of the most widely used standards,
namely RDF, DAML+OIL and OWL.
The RDF is a language for representing
information about resources in the WWW. It is
particularly intended for representing metadata
about Web resources, such as the title, author,
and modification date of a Web page. However,
by generalising the concept of a Web resource,
RDF can also be used to represent information
about things that can be identified on the Web,
even when they cannot be directly retrieved on
the Web. Examples include information about
items available from online shopping facilities

1100

(e.g., information about specifications, prices, and

availability). RDF is intended for situations in
which this information needs to be processed by
applications, rather than only being displayed to
people. RDF provides a common framework for
expressing this contextual information that it can
be exchanged between applications without loss
of meaning. It represents a simple graph model
and uses a well-documented XML schema for
datatypes by identifying all elements by means
of URIs. The structure of any expression in RDF
is a triplet (Figure 1), consisting of a subject, an
object, and a predicate (also called property).
RDF provides a way to express simple statements about resources, using named properties and
values. However, RDF user communities also need
the ability to define the vocabularies (i.e., terms or
taxonomies in ontology community) they intend
to use in those statements, specifically, to indicate
that they are describing specific kinds or classes
of resources, and will use specific properties in
describing those resources. RDF itself provides
no means for defining such application-specific
classes and properties. Instead, such classes and
properties are described as an RDF vocabulary,
using extensions to RDF provided by the RDF
Schema (RDFS) (Brickley, 2004). The RDFS
provides the facilities needed to describe such
classes and properties, and to indicate which
classes and properties are to be used together. In
other words, RDFS provides a type system for
RDF. The RDFS type system is somehow similar

Figure 1. RDF triplet

Subject

Predicate

Object

rd f:S u bje ct
rd f:O b je ct
rd f:P re d ica te

The SPO conceptual model of RDF

Ontological Engineering in Pervasive Computing Environments

to the type systems of object-oriented programming languages, such as Java.

DAML+OIL is also a semantic markup language for Web resources. It builds upon earlier
W3C standards such as RDF and RDFS, and
extends these languages with richer modelling
primitives (e.g., in terms of class subsupmtion or
object properties). It adds the familiar ontological
primitives of object-oriented and frame-based
systems, and the formality of a very expressive
description logic (i.e., an ALC Description Logic
language).
OWL is a semantic markup language for
publishing and sharing ontologies on the World
Wide Web. OWL is intended to be used when
the information contained in documents needs
to be processed by applications, as opposed to
situations where the content only needs to be
presented to humans. OWL can not only be used
to explicitly represent the meaning of terms in
vocabularies and the relationships between those
terms, but also may serve implicitly representing
inferred terms or whole taxonomies by means
of logical reasoners (i.e., applications, such as
RACER, FaCT, DLP (Haarslev, 1999), based on
Description Logics axioms that provide inferred
knowledge with respect to certain First Order
Logic algorithms, such as Structural Subsumption and Tableau algorithms). OWL has more
facilities for expressing meaning and semantics
than XML, RDF, and RDFS, and thus OWL goes
beyond these languages in its ability to represent
machine interpretable content on the Web. OWL is
an enhanced revision of DAML+OIL Web ontology language. OWL provides three increasingly
expressive sublanguages designed for use by
specific communities of implementers and ontology developers. The OWL class (i.e., Concepts
and conceptual formulas in Description Logics)
expressions are built with Knowledge Representation primitives that are properties. These will be
organised in two groups. The first group includes
primitives defined for OWL Lite while the second

group includes primitives defined for OWL DL

and OWL Full.
In the previous paragraph, we discussed a set of
concept models (i.e., interpretations in the ABox
that satisfy the TBox formulas) that are decidable
and satisfiable, but with restricted expressiveness
due to representation of an sublanguage of the
ALC DL. In this context, there is no disjunction
and negation concept formulas and the property
restrictions (owl:Restriction) of objects (i.e., the
instances of concepts in the ABox) are set to
cardinality (owl:cardinality).
The latter group stands for fully expressive
OWL languages, but the OWL DL is the only
sublanguage that can be considered as a machine comprehensible ALCN Description Logic
language for the Semantic Web. The OWL Full
is not a decidable language in terms of Description Logics, since the ABox and the TBox may
overlap (e.g., an instance of the ABox can be
interpreted as a concrete concept of the TBox)
according to their concept definitions and there
is no such a reasoner (i.e., DL algorithm) that can
infer whether an interpretation expressed in OWL
Full is satisfiable or not. The OWL DL is a DL
language enabling the definitions of conjunction,
disjunction, and negation for any interpretation of
the concept models and deals with certain value
(i.e., owl:allValuesFrom, R.C in DL semantics),
existential (owl:someValuesFrom, R.C in DL
semantics) and number (e.g., owl:maxCardinality,
n in DL semantics) restrictions with arbitrarily
cardinalities of their fillers (i.e., instances that
satisfy Roles in a Knowledge Base ABox).
Due to the dynamic nature of the contextual
information in PCEs and the vague, temporal or
probabilistic knowledge representation, there have
been many extensions in OWL DL language in
order to cope with such models (e.g., imprecise
information representation, probabilistic knowledge concept definitions, temporal information
representation). Such OWL DL languages deal
with fuzzy context modelling (e.g., the proposed

1101

Ontological Engineering in Pervasive Computing Environments

[Stracia, 1998]), temporal contextual concepts

(e.g., the proposed [Artale, 1998]) and probabilistic
knowledge representation in terms of Bayesian
Networks (e.g., the proposed P-OWL language).
Such OWL extensions are based on novelty extensions of the classical DLs (e.g., the OWL is a
kind of a SHIN language) with the use of Modal
Logics (Elgesem, 1997) that provide different
world assumptions in well-defined DL Tboxes.
Moreover, in order to deduct new knowledge
by reasoning over contextual facts (i.e., set of
instances, such as the Abox interpretations),
certain extensions in the ontological community
by representing rules have been proposed and
standardised. The SWRL (Horrocks et al., 2003)
is a language for a Semantic Web Rule Language
based on a combination of the OWL DL and OWL
Lite sublanguages with the Unary/Binary Datalog
RuleML sublanguages of the Rule Markup Language (Boley, 2001). Such language extends the
set of OWL axioms to include Horn-like clauses.
Hence, Horn-like clauses can be combined with
an OWL knowledge base by defining rules for the
ABox instances reasoning. A high-level abstract
syntax is provided that extends the OWL abstract.
An extension of the OWL model-theoretic semantics is also given to provide a formal meaning for
OWL ontologies including rules written in this
abstract syntax.

Managing the Ontology

Taking into account the distributed nature of a
PCE, several key challenges emerge concerning
the use of ontologies. One has to be able to deal
with multiple and distributed ontologies (i.e.,
distributed knowledge representation in terms of
ontological modularity of taxonomies) to enable
reuse, maintainability, evolution, and interoperability. Furthermore, support is needed in managing the evolution of multiple and distributed
ontologies, in order to ensure consistency. In
fact, the ability of logical reasoners to link independent ontology modules, to allow them to be

1102

separately maintained, extended, and re-used is

one of their most powerful features. In PCE we
encounter large and distributed ontologies (i.e.,
in terms of number of concepts), representing
semantically similar contexts (e.g., equivalent
concepts or roles).
In highly distributed systems such as the Semantic Web, modularity naturally exists in terms
of conceptual modelling. Large ontologies (e.g.,
medical or biological ontologies such as GALEN
(Rector, 2003; Rector et al., 1999) may be split in
smaller and reusable modules in order to meet the
needs of ontology maintenance and distributed
reasoning. Requirements for managing ontologies in such a distributed environment include
the following:

The semantic loose coupling, which refers to

the conceptualisation sets of ontologies that
are irrelevant (in terms of concept perception).
The self-standing or self-containment of
concepts is the idea of defining concepts
that enable the automatic inference of
other, classified concepts without having
to access contextual information in other
ontologies.
The ontology consistency, which performs
integrity checks to assess whether relevant
knowledge in other systems has changed,
and, updates the self-standing ontological
module, if needed.

In Maedche (2003), the authors present a

framework for managing multiple and distributed
ontologies in a Semantic Web environment. They
describe a conceptual modelling framework for
single, multiple, and distributed ontologies, which
renders feasible the realisation of ontology-based
systems using well-established technologies such
as relational databases. The authors make use of
the object-oriented paradigm and extend it with
simple deductive features. In their approach,
information is organised in Ontology-Instance

Ontological Engineering in Pervasive Computing Environments

models, which contain both ontology entities

and instances. They introduce the single ontology consistency and the dependent ontology
consistency definitions and propose an evolution strategy that ensures the consistency of the
evolving ontologies by unambiguously defining
the way in which elementary changes will be
resolved.
In another approach, discussed in Rector
(2003), the key is modularity. Modularity is
acknowledged as a key requirement for large ontologies in order to achieve reuse, maintainability,
and evolution. Rector, motivated by the fact that
mechanisms for normalisation are standard for
databases, proposes the concept of normalisation for ontologies. He proposes a two-step normalisation implemented using OWL or related
description logic-based formalisms. For the first
step, namely the ontological normalisation, he
makes use of the analysis in Guarino (2000). For
the second step (implementation normalisation)
he proposes an approach based on decomposing
the ontology into independent disjoint skeleton
taxonomies restricted to simple trees, which can,
then, be recombined using definitions and axioms
to explicitly represent the relationships between
them. The main requirement, though, for implementation normalisation is that the modules to be
reused can be identified and separated from the
whole, so as to evolve independently.
In Guarino (1999), the authors introduce the
OntoSeek, a system that employs an ontology
server for ontology management. The server
provides an interface for applications willing
to access or manipulate an ontology data model
(i.e., a generic graph data structure), and facilities maintaining a persistent Lexical Conceptual
Graph (LCG) database. End users and resource
encoders can access the server to update the
LCG database encoded in a markup language,
(e.g., XML). The main function of the ontology
server is to create, edit, evaluate, publish, maintain, and reuse ontologies. Particularly important
is the ability to support the collaborative works
through the Web.

Semantic Mappings Among

Ontologies
Clearly, it would be desirable in a PCE that the
various services are capable of interoperating
with existing data sources and consumers. It is
very difficult to build a consensus about the terminologies and structures that should be used. In
order to introduce the significance of the ontology
matching issue in the computing paradigm, we
support the Semantic Web to make information
on the World Wide Web more accessible using
machine-readable metadata. Since the ontological
development deals with conceptualising specific
domains, the notion of context modelling could
be imperative by using specific domain ontologies. Thus, an ontology is contextualised when its
concepts are kept local (i.e., self-standing concepts
that are not shared with other ontologies) and are
associated only with the concepts of other ontologies, through well-defined explicit mappings (or
implicit mappings, through distributed reasoning
tasks, such as Compilation of Implied Knowledge
with Queries).
An architecture is needed for building a
Web of data by allowing incremental addition
of sources, where each new source maps to the
sources considered most convenient. Therefore,
information processing across ontologies is only
possible when semantic mappings between the
ontologies exist. Technological infrastructure
for semantic interoperability between semantically autonomous communities (i.e., ontological
development that is compliant with the requirements in subsection 3.2) must be based on the
capability of representing local ontologies and
contextual mappings between them, rather than
on the attempt to create a global and possibly
shared conceptualisation. Manually finding such
contextual mappings is tedious, error prone, and
clearly not possible on a large scale. This is what
researchers call semantic heterogeneity in the
Semantic Web paradigm, namely, a situation in
which different meanings use the same concept

1103

Ontological Engineering in Pervasive Computing Environments

to mean the different things and use different

granularity to describe the same domain forming
explicit different perspectives of the world. In this
context, the lack of uniformity prompts for the
meaning negotiation among different agents, since
they do not understand each others as they use
languages with heterogeneous semantics.
A lot of work has been performed during the
last years in this direction, which resulted in the
development of systems that provide mappings
semi-automatically (Doan, 2003; Halevy, 2003;
Fowler et al., 1999; Chalupsky, 2000). In this way,
given the semantic mappings between different
ontologies, one may match them in order to be
able to reason about a specific domain.
Given two ontologies, the ontology-matching
problem is to find semantic mappings between
them. The simplest type of mapping is the one-toone mapping between the elements of the different
ontologies, namely each element of one ontology
is associated with at most one element of the second ontology and vice versa. Mappings between
different types of elements are possible, meaning

that a property may be mapped to a class. More

complex mappings are possible, such that a class
or an instance may map to a union of classes or
instances. In general, a mapping could be specified as a query that transforms instances, classes
and properties from one ontology into instances,
classes, and properties of another.
To illustrate the above, suppose that we want
to provide mappings between two WWW sites
that sell books. In Figure 2, the ontology-matching
problem between an author-centric ontology and
a book-centric ontology is depicted. A matchmaking process must provide semantic mappings
between the nodes in the two ontologies in an
automatic manner.
In Doan (2003), the authors describe GLUE,
a system that employs machine-learning techniques to determine such mappings. Given two
ontologies, for each concept in one ontology the
system finds the most similar concept in the other
ontology. A key feature of the GLUE is that it
uses multiple learning strategies, each of which
exploits well a different type of information either

Figure 2. Ontology matching problem

A u tho r
hasName

P u b lish e r

writes

N am e

publishes

B o ok

isTitled

T itle

A u th o r-c e n tric O n to lo g y

B o ok
hasAuthor
A u tho r
owl:Class
xmlSchemaDatatype
owl:DatatypeProperty
owl:ObjectProperty

1104

hasName
N am e

isPublished
isTitled
T itle

B o o k -c e n tric O n to lo g y

P u b lish e r

Ontological Engineering in Pervasive Computing Environments

in the data instances or in the taxonomic structure

of the ontologies.
The first issue to address such conceptual
matching is the notion of similarity between two
concepts. The approach in Doan (2003) is based
on the observation that many practical measures
of similarity can be defined based solely on the
joint probability distribution of the concepts involved. The second challenge is that of computing
the joint distribution of any two given concepts A
and B. Specifically, for any two concepts A and
B, the joint distribution consists of p(aAB),
p(AB), p(AB), p(AB), where
a term such as p(a:AB), is the probability that an
instance in the domain belongs to concept A but
not to concept B. GLUE, also, uses a multi-strategy
learning approach employing a set of learners and
then combining their predictions using a metalearner. Furthermore, it exploits domain specific
constraints using general heuristic algorithms to
improve the matching accuracy.
In another approach, described in Halevy
(2003), the authors provide a mediating schema
between RDF and XML data sources. Most of
these sources provide XML representations of
the data and not RDF; XML and the applications
that depend on it rely not only on the domain
structure of the data, but also on its document
structure. In order to provide interoperability
between such sources, one must map between
both their domain structures and their document
structures. Second, data management practitioners often prefer to exchange data through local
point-to-point data translations, rather than mapping to common mediated schemas or ontologies.
Therefore, the authors built the Piazza system,
which addresses these challenges. Piazza offers a
language for mediating between data sources on
the Semantic Web, which maps both the domain
structure and document structure. Piazza also
enables interoperation of XML data with RDF
data that is accompanied by rich OWL ontologies. They proposed a language for mediating
between nodes that allows mapping simple forms

of domain structure and rich document structure.

The language is based on Xquery (Fernndez et al.
2005), the emerging standard for querying XML.
They also showed that this language could map
between nodes containing RDF data and nodes
containing XML data.
In Fowler et al. (1999), the authors propose
InfoSleuth, a system that can support construction
of complex ontologies from smaller component
ontologies so that tools tailored for one component ontology can be used in many application
domains. Examples of reused ontologies include
units of measure, chemistry knowledge, and
geographic metadata. The mapping is explicitly
specified among these ontologies as relationships
between terms in one ontology and related terms
in other ontologies.
In another approach, the OntoMorph system
of the Information Sciences Institute (ISI) of
the University of Southern California aims to
facilitate ontology merging and the rapid generation of knowledge base translators (Chalupsky,
2000). It combines two powerful mechanisms to
describe Knowledge Base transformations. The
first of these mechanisms is syntactic rewriting
via pattern-directed rewrite rules that allow the
concise specification of sentence-level transformations based on pattern matching, and the second
mechanism involves semantic rewriting which
modulates syntactic rewriting via semantic models
and logical inference. The integration of ontologies can be based on any mixture of syntactic and
semantic criteria.

Ontology-Based Profiling
and Information Retrieval
In a typical PCE there exists a vast amount of
contextual information available for retrieval. The
traditional solution to the problem of information
retrieval employs keyword-based searching techniques. This solution is inadequate and, therefore,
new searching mechanisms must be developed.

1105

Ontological Engineering in Pervasive Computing Environments

The context should be modelled in a semantically

enriched manner, namely by means of ontologies,
and searching techniques that make use of this
semantic information must be applied in order
to provide superior results. Furthermore, PCEs
usually integrate sensor networks, which differ
considerably from current networked and embedded systems. They combine the large scale and
distributed nature of networked systems with the
extreme energy constraints and physically coupled
nature of embedded control systems. Sensors,
however, produce raw data. Ontologies could be
used to transform the raw data to a semantically
enriched representation that could be queried by
an advanced searching mechanism.
The proliferation of Semantic Web technologies, such as ontologies, has led to the development
of more sophisticated semantic query mechanisms. Semantic querying techniques will exploit
the semantics of content to provide results that
are superior to those delivered by contemporary
techniques, which rely mostly on the lexical and
structural properties of a document. There are
a number of proposals for querying data (e.g.,
Anyanwu et al., 2003) expressed by means of
ontologies, but most of them are constrained in
such a way that one is able to form queries only
of the type Get all elements that are related to
element A through a relationship R where R is,
typically, specified as a path expression. Recently,
research is being conducted in order to be able to
express queries of the type How is element A
related to element B?
In Khan (2004), the authors propose a framework for querying databases of a specific context
using an ontology-based index structure. The traditional solution employs keyword-based search.
The only documents retrieved are those containing
user specified keywords. But many documents
convey desired semantic information without
containing these keywords. One can overcome
this problem by indexing documents according
to context and meaning rather than keywords,
although this requires a method of converting

1106

words to meanings and the creation of a meaning-based index structure. In the discussed work,
the authors have solved the problem of an index
structure through the design and implementation
of a concept-based model using domain dependent
ontologies. In other words, the ontology provides
index terms/concepts that can be used to match
with user requests. Furthermore, the generation of
a database query takes place after the keywords
in the user request are matched to concepts in
the ontology. To achieve this, they employ a
proprietary method for query expansion and SQL
query generation.
In Wang et al. (2003), a system for classifying text documents based on ontologies
for representing the domain specific concept
hierarchy (an example is depicted in Figure 3)
is presented. Many researchers have shown that
similarity-based classification algorithms, such
as k-nearest neighbour and centroid-based classification, are very effective for large document
collections. However, these effective classification
algorithms still suffer disadvantages from high
dimensionality that greatly limit their practical
performance. Finding the nearest neighbours in
high dimensional space is very difficult because
the majority of points in high dimensional space
are almost equi-distant from all other points. The
k-nearest neighbour classifier is an instance-based
classifier. This means that a training dataset of high
quality is particularly important. An ideal training
document set for each particular category will
cover all the important terms, and their possible
distribution in this category. With such a training set, a classifier can find the true distribution
model of the target domain. Otherwise, a text
that uses only some keywords out of a training
set may be assigned to the wrong category. In
practice, however, establishing such a training
set is usually infeasible. Actually, a perfect training set can never be expected. By searching the
concept hierarchy defined by a domain specific
ontology, a more precise distribution model for a
predefined classification task can be determined.

Ontological Engineering in Pervasive Computing Environments

Figure 3. Domain specific concept hierarchy

C om pu te r
S cie n ce

A rtificia l
In te llige n ce

N e u ra l
N e tw o rk s

G e n e tic
A lg o rithm s

The experiments that were carried out indicated

that, by using this approach, the size of the feature
sets can be effectively reduced and the accuracy
of the classifiers can be increased.
An extension of the ontological reasoning, in
terms of information retrieval, is discussed in
Maedche et al. (2001). In this work, the authors,
in order to define and exploit the semantics of
domain ontologies, perform the context retrieval
process by clustering the pre-classified concepts
with several similarity measures. The metrics of
the semantic distances among concepts, which
are referred to as similarity measures, are the
Taxonomy similarity, the Relation similarity, and
the Attribute similarity. Such measures form the
input to the hierarchical clustering algorithm. In
this context, one may deduce about new relations
among concepts, something that enriches the
notion of the distributed domain ontologies in
PCE. Furthermore, the classification process, with
respect to the context retrieval, is not feasible with
undefined relations among conceptual formulas
(i.e., primitive concept, defined concepts, and
complex concept definitions).

D a ta ba se
S yste m s

D e d u ctive
D a ta ba se s

R a tio na l
D a ta ba se s

Moreover, such new relations may denote

new activities (i.e., constraints), behaviours (i.e.,
axioms), and roles on entities or concepts in ontologies. In the pervasive computing paradigm we
encounter a large number of interacting entities.
The rules that govern that interaction are complex
and should be based, among others, on the profiles
of entities. According to the Cambridge Dictionary
(1980), a profile is a portrait, drawing, diagram,
etc., of a side view of the subject. A model of an
entitys characteristics, activity, behaviour and its
role in the PCE is the entitys profile. Based on
this knowledge, complex mechanisms manage
interaction of entities in the most efficient and
effective way.
Several methods have been proposed for using
ontologies for user and data sources profiling, as
well as for describing communication mechanisms. Hence, complex queries over data sources
may be exploited and auto-learning systems may
propose content sources to users based on their
preferences.
In Middleton et al. (2004) the authors explore an ontological approach to user profiling
in the context of a recommender system. Their

1107

Ontological Engineering in Pervasive Computing Environments

system, called Foxtrot, addresses the problem of

recommending online research papers to academic researchers. Their ontological approach
to user profiling is a novel idea to recommender
systems. Researchers need to be able to search
the system for specific research papers and have
interesting papers recommended. Foxtrot uses a
research-paper-topic ontology to represent the
research interests of its users. A class is defined
for each research topic and is-a relationships
are defined where appropriate. Their ontology
is based on the CORA (McCallum et al., 2000)
digital library, as it classifies computer science
topics and has example papers for each class.
Research papers are represented as term vectors,
with term frequency/total number of terms used
for a terms weight; terms represent single words
in a papers text. Research papers in the central
database are classified by an IBk (Kibler, 1991)
classifier, a k-nearest neighbour type classifier,
which is boosted by the AdaBoostM1 algorithm
(Freund, 1996).
In Parkhomenko et al. (2003), the authors present an approach that uses ontologies to set up a
peer profile containing all the data, necessary for
peer-to-peer (P2P) interoperability (Milojicic et
al., 2002). The use of this profile can help address
important issues present in current P2P networks,
such as security, resource aggregation, group
management, etc. They also consider applications
of peer profiling for Semantic Web built on P2P
networks, such as an improved semantic search
for resources not explicitly published on the Web
but available in a P2P system. They developed
the ontology-based peer profile in RDF format
and demonstrated its manifold benefits for peer
communication and knowledge discovery in both
P2P networks and Semantic Web.
In another approach regarding P2P networks,
the authors in (Nejdl et al., 2003), focus on peer
profiling so as to enable complex user queries
and routing algorithms. RDF-based P2P networks
allow complex and extendable descriptions of
resources, and they provide complex query fa-

1108

cilities against these metadata instead of simple

keyword-based searches. Furthermore, they are
capable of supporting sophisticated routing and
clustering strategies based on the metadata schemas, attributes and ontologies. Especially helpful
in this context is the RDF functionality to uniquely
identify schemas, attributes and ontologies. The
resulting routing indices can be built using dynamic frequency counting algorithms and support
local mediation and transformation rules.

Semantic-Based Service
Interoperability Through
Ontologies in Pervasive
Environments
Although, the Web was once just a content repository, it is, now, evolving into a provider of services.
Web-accessible programs, databases, sensors,
and a variety of other physical devices realise
such services. In the next decades, computers
will most likely be ubiquitous and most devices
will have some sort of computing functionality.
Furthermore, the proliferation of intranets, ad-hoc,
and mobile networks sharpen the need for service
interoperation. However, the problem of service
interoperability arises because todays Web is designed primarily for human use. Nevertheless, an
increased automation of services interoperation,
primarily in business-to-business and e-commerce applications, is being noted. Generally,
such interoperation is realised through proprietary
APIs that incorporate hard-coded functionality
in order to retrieve information from Web data
sources. Ontologies can prove very helpful in
the direction of service description for automatic
service discovery, composition, interoperation,
and execution.
In McIlraith et al. (2001), the authors present
an agent technology based on reusable generic
procedures and customising user constraints
that exploits and showcases WS markup. To
realise their vision of Semantic WS, they cre-

Ontological Engineering in Pervasive Computing Environments

ated semantic markup of WS that makes them

machine understandable and easy to use. They
also developed an agent technology that exploits
this semantic markup to support automated WS
composition and interoperability. Driving the development of their markup and agent technology
are the automation tasks that semantic markup of
WS will enable, in particular, service discovery,
execution, and composition and interoperation.
Automatic WS discovery involves automatically
locating WS that provide a particular service
and adhere to requested properties. Automatic
WS execution involves a computer program or
agent automatically executing an identified WS.
Automatic WS composition and interoperation
involves the automatic selection, composition,
and interoperation of appropriate WS to perform
some task, given a high-level description of the
tasks objective. They use ontologies to encode
the classes and subclasses of concepts and relations pertaining to services and user constraints.
The use of ontologies enables sharing common
concepts, specialisation of these concepts, and
vocabulary for reuse across multiple applications,
mapping of concepts between different ontologies, and the composition of new concepts from
multiple ontologies.
In Gibbins et al. (2003), the authors describe
the design and implementation of an ontologically
enriched WS system for situational awareness.
Specifically, they discuss the merits of using techniques from the multi-agent systems community
for separating the intentional force of messages
from their content and the implementation of
these techniques within the DAML Services
model. They identify that the world of WS may
be characterised as a world of heterogeneous
and loosely coupled distributed systems where
adaptability to ad-hoc changes in the services
offered by system components is considered
advantageous. The term loosely coupled, means
that the interactions between system components
are not rigidly specified at design time. Instead,
system components may opportunistically make

use of new services that become available during their lifetime without having been explicitly
told of their existence from the outset. The task
of searching for a system component which can
perform some given service (i.e., service discovery) is the enabling technique that makes looselycoupled systems possible, and provides a process
by which system components may find out about
new services being offered. Service descriptions
are more complex expressions, which are based
on terms from agreed vocabularies and attempt to
describe the meaning of the service, rather than
simply ascribing a name to it. An essential requirement to service discovery is service description.
A key component in the semantics-rich approach
is the ontology. In the conventional WS approach
exemplified by Web Services Definition Language
(WSDL) or even by DAML Services, the communicative intent of a message (e.g., whether it is
a request or an assertion) is not separated from the
application domain. This is quite different from
the convention from the Multi-Agent Systems
world, where there is a clear separation between
the intent of a message, which is expressed using
an Agent Communication Language (ACL), and
the application domain of the message, which is
expressed in the content of the message by means
of domain-specific ontologies. In this approach,
there is a definite separation of the intent of the
messages and the application domain.
In Medjahed et al. (2003), the authors propose
an ontology-based framework for the automatic
composition of WS. An important issue in the
automatic composition of WS is whether those
services are composable (Berners-Lee, 2001).
Composability refers to the process of checking
if WS to be composed can actually interact with
each other. A composability model is proposed
for comparing syntactic and semantic features of
WS (Figure 4). The second issue is the automatic
generation of composite services. A technique
is proposed to generate composite service descriptions while obeying the aforementioned
composability rules. Such technique uses as

1109

Ontological Engineering in Pervasive Computing Environments

Figure 4. Composability model for Web services

CCom
ompositio
positionn
LLeevvel
el

CCom
ompositio
positionn
SSoundness
oundness

SSem
emantic
antic
CCom
omposability
posability

Q
Qualitativ
ualitativee
PProperties
roperties
O
Operation
peration
SSem
emantics
antics
SServ
ervice
ice &&
O
Operation
peration LLeevvel
el

M
Messages
essages
O
Operation
peration
M
Mode
ode
BBinding
inding

input a high-level specification of the desired

composition. This specification contains the list
of operations to be performed through composition without referring to any composite service.
Based on their composability model, they propose
an approach for the automatic composition of
WS. This approach consists of four conceptually
separate phases: specification, matchmaking,
selection, and generation. The authors define an
XML language, which adopts an ontology-based
model to cater for Semantic Web-enabled WS,
for the specification of composite services. Once
specifications are provided, the next step is to
generate appropriate composition plans using a
matchmaking algorithm. The general principle
of the matchmaking algorithm is to map each
operation of the composite service to one or more
operations of existing services. At the end of the
matchmaking phase, several composition plans
may have been generated. To facilitate the selection of relevant plans, they define three parameters
for the quality of composition: ranking, relevance,
and completeness. The ranking of a composition
gives an approximation of its importance, the

1110

SSyntactic
yntactic
CCom
omposability
posability

relevance gives an approximation of the composition soundness and the completeness gives
the proportion of composite service operations
that are composable with other services. The
last phase in their approach aims at generating a
detailed description of a composite service. This
description includes the list of outsourced services, mappings between composite services and
component service operations, mappings between
messages and parameters and flow of control and
data between component services.
Let us imagine a PCE where all devices interoperate through services. In this way, a serviceoriented architecture for functionality integration
is being employed. Therefore, the problem of
deploying a PCE reduces to the problem of service description and ad-hoc service discovery,
composition, and execution.
We propose that the services/agents that
represent the various devices share a common
communication scheme, possibly by means of
an ACL or a common ontology that describes
the communication details. However, they do
not share any a priori knowledge about the se-

Ontological Engineering in Pervasive Computing Environments

mantics of the functionality that these devices

provide. As described above, through the use of
the device-specific ontologies one can achieve
runtime service discovery and execution based on
an abstract description of his objectives. In other
words, a device will query its environment, giving
the description of its objectives and employing
complex semantic querying mechanisms. After
obtaining a service reference, it will proceed with
its execution. As new devices join the PCE and
while others abandon it, the cycle of service description, discovery, and execution will continue
forever with minimum external interference.
Furthermore, the work that has been carried
out in the field of ontology management and
ontology mapping ensures that the consistency
of the ontologies is maintained as they evolve
throughout time. Furthermore, webs of complex
services are being created as devices join in or
withdraw from the PCE and as new composite
services are being created based on noncomposite
(i.e., primitive) or even other composite services.
This is only possible when one can provide semantic context mappings between the different
services and automatically produce semantic
descriptions of the composite services. Hence, a
complicated hierarchy of services may be created.
Such a hierarchy should be both manageable and
highly flexible.

Examples of Semantic-Based
Service Interoperability
Through Ontologies in PCE
To illustrate the key issues, consider the following
simple scenario (Figure 5). Bob has an important
professional meeting (i.e., human society context) in his office (OfficeA) with Alice. For this
meeting it is important that Bob has a certain
postscript file (FileA) printed (i.e., application
context). Unfortunately, the printer in his office
(PrinterA) is not capable of printing postscript
files. However, there is one such printer (PrinterB)

in another office (OfficeB), but Bob is unaware

of its existence (i.e., service registry context). In
addition, the two offices are adjacent (i.e., spatial
context). Bob is engaged in the activity of editing
(i.e., human activity-oriented context) the file and
finishes just 15 minutes before his appointment
(e.g., temporal context). The PCE, by recording
the time at which Bob finishes editing the file and
knowing that the appointment is scheduled in 15
minutes, (i.e., contextual history) infers that Bob
wants this file for the appointment (e.g., applying
deduction rules over certain contexts). Furthermore, it judges that there is sufficient time for
Bob to go to the next office and print the file and,
consequently, issues the suggestion (i.e., ambient
notification to human).
There are several types of semantically enriched contexts in this scenario. First of all the
system knows the location of Bob, as well as
the locations of the offices within the building.
It also has knowledge about the equipment that
is present at the offices and its capabilities (i.e.,
terminal profile). The system has access to his
personal agenda and monitors the activities that
he is engaged in (i.e., user profile). It is, also,
capable of associating any activity with specific
entities, which, in turn have specific attributes
(i.e., inference over context facts). Furthermore,
it employs an inference engine to deduce new
contextual information (Bob wants the files for
his appointment) and uses this information to
issue suggestions.
In another scenario, a system for observing
and assisting elderly people, who are alone at
home, may be considered. Ontologies may be
used to define simple cases (e.g., the person is
sleeping for the past fifteen hours or has been in
the bathroom for too long) in which the system
must take specific actions, wake the patient by
ringing a bell or informing the attending doctor.
Furthermore, with the addition of a rule engine,
complex conditions may be defined that were,
previously, required to be diagnosed by a trained
doctor. For instance, the system may constantly

1111

Ontological Engineering in Pervasive Computing Environments

Figure 5. An example ontology

S e m a n tics

B obs context
B obs agenda c ontext

O ffice -A

C om p lem ent c ontext

O ffice -B

L e ge nd

H istory

isAdjacent
patterns

o w l:C lass
o w l:Indiv idual
o w l:O bjectP roperty
rdfs:subC lassO f
instance

S patial

isLocatedAt

P rinterA -A pp

E ntity

A ppointm ent
T em poral

takesPlace

duringInterval
A ctivity

isEngaged

P rinterB -A pp

A pp lication
runsOn

boundsService

E ditin g

produces

A rtefact
D ev ice

actsOn

hasDuty

H um an

S ocia l

F ile-A

R egistry
S rv P rntB

A genda

P rinterA

B ob

P rinterB

A lice

M eeting

S rv P rntA

monitor the patients Electro Cardiogram (ECG)

and, based on time depended composite rules,
may decide that the patient requires immediate
medical attention.
Research has been reported in the association
of Semantic Web technologies and pervasive
computing. In Chen et al. (2004), the authors
describe a framework for an agent-based PCE.
Central to their framework is the presence of an
intelligent context broker that accepts context
related information from devices and agents in the
environment as well as from other sources, including information available on the Web describing
the space and events and activities scheduled for
it. The context broker integrates and reasons over
this information to maintain a coherent model of
the space, the devices, agents, and people in it, and
their associated services and activities. A key to
the realisation of this architecture is the use of a set
of common ontologies (e.g., specialising an Upper
Level Ontology or semantics similarity measures).
They identify ontologies as a key component for
building open and dynamically distributed PCEs

1112

in which agents and devices share contextual

information. They describe the use of OWL and
other tools for building the foundation ontology
for the Context Broker Architecture (CoBrA).
The CoBrA ontology models the basic concepts
of people, agents, place, and presentation events
in an intelligent meeting room environment. It
provides a vocabulary of terms for classes and
properties suitable for building practical systems
that model context in PCEs.
In Wang (2003), the author describes The context broker, a broker-centric agent architecture,
which aims to support context-aware computing
in intelligent spaces. In this architecture, a broker
maintains and manages a shared context model on
the behalf of a community of devices and agents.
It provides necessary common services including
ontology mediation and teamwork support, and
enforces policies for security, trust, and privacy.
A reasoning engine is responsible for reasoning
with ontological and contextual knowledge. As
ontological knowledge, the author identifies the
static knowledge derived from the underlying

Ontological Engineering in Pervasive Computing Environments

ontology model (expressed in OWL), and as contextual knowledge, the dynamic knowledge that is
inferred from acquired situational information.
In Ranganathan et al. (2003) and RanganathanGAIA (2003), the authors describe GAIA, an
infrastructure for Smart Spaces, which are ubiquitous computing environments that encompass
physical spaces. GAIA converts physical spaces
and the ubiquitous computing devices they contain
into a programmable computing system. It offers
services to manage and program a space and its
associated state. GAIA is similar to traditional
operating systems in that it manages the tasks common to all applications built for physical spaces.
Each space is self-contained, but may interact
with other spaces or modular ontologies (e.g., via
context mappings). GAIA provides core services,
including events, entity presence (e.g., devices,
users and services), discovery, and naming. By
specifying well-defined interfaces to services,
applications may be built in a generic way so that
they are able to run in arbitrary active spaces.
The core services are started through a bootstrap
protocol that starts the GAIA infrastructure.
Finally, GAIA allows application developers to
specify different behaviours of their applications
for different contexts. The use of contextualised
ontologies makes easier for developers to specify
context-sensitive behaviour.
One of the main uses of ontologies is that it
allows developers to define all the terms that can
be used in the environment. Ontological engineering allows the attachment of precise semantics
to various terms and the clear definition of the
relationships between different terms. Hence, it
prevents semantic ambiguities where different
entities in the environment have different ideas
from what a particular term means. Different entities in the environment can refer to the ontology
to get a definition of a term, in case they are not
sure (e.g., due to the imprecise conceptual modelling or the nature of the contextual information).
Furthermore, developers employ ontologies for
describing both entities (e.g., concepts) and con-

textual information (domain specific conceptual

modelling). In this manner, they can determine
whether these descriptions are valid with respect
to the schema defined by the ontology. When
a new entity is introduced into the system, its
description can be checked against the existing
ontology to see whether it is satisfiable (Roman
et al., 2002). If the description is not consistent
with the concepts described in the ontology, then
the description is faulty. Moreover, ontologies are
used to provide semantic discovery of objects and
services. Conventional object registries provide a
limited capability for object discovery and the existing discovery protocols support limited ability
to spontaneously discover entities on a network.
However, in PCEs it must be possible to discover
only the relevant entities, without knowing,
in advance, what will be relevant (Amann et al.,
2003). This process has also been termed matchmaking. Finally, a key feature of applications in
PCEs is that they are context-aware, i.e., they are
able to obtain the current context and adapt their
behaviour to different situations.

Conclusion and Future Work

The vision of pervasive computing introduces
structural issues, as in a typical ubiquitous environment there will be numerous devices with
computing capabilities that need to interoperate.
Therefore, we cannot rely on shared a priori
knowledge based on commonly accepted standards to resolve such issues. Software components
must adapt to their environment at runtime to
integrate their functionality with other software
components seamlessly. We argued that the
concept of Semantic Web, by the means of ontologies, could prove to be a satisfactory solution
to the aforementioned problem, as it enables the
system and the software providers to deal with
semantically enriched specifications of services
and semantically enriched data. Models based
on ontologies may be used to provide objects, and

1113

Ontological Engineering in Pervasive Computing Environments

mainly contextual information, with semantics.

The context representation is thought to be of the
utmost importance, as context knowledge and
manipulation enables a system to be context aware
and proactive and to employ more effective and efficient decision-making mechanisms. In addition,
profile and policy management could be based on
ontology models, in order to provide maximum
flexibility for describing entities and rules regarding resource usage and manipulation.
Moreover, we surveyed the research that has
been carried out regarding ontologies and tried
to map the key findings to the requirements of
a PCE. We have discussed efforts dealing with
the management of multiple, overlapping, and
evolving ontologies and the semantic mappings
among them. We, also, depicted how ontologies
can help in searching mechanisms and in profiling
of users and data sources. Furthermore, due to
the diverse nature of PCEs, the use of interoperability standards is not a sufficiently scalable
approach. Hence, we showed the important role
that ontologies could play in achieving service
interoperability, a key requirement in PCEs,
and depicted how one may use ontologies in the
deployment of PCE.
Research on the association of ontologies and
PCEs is rather restricted. Mainly the artificial intelligence community is carrying out the research
regarding ontologies, whilst the research regarding pervasive computing is mostly conducted
by the network community. Thus, the need of
collaboration between the two communities is
clearly visible. Nevertheless, a significant trend is
identified. More and more researchers acknowledge the necessity of incorporating ontologies
into the development of systems for supporting
PCE. In this way, the emerging standard for Web
ontologies, such as the (OWL), may prove to be a
significant factor in the road ahead.
However, there are several future research
directions that should be considered. Description logics may perform well in defining domain
vocabularies and concepts but lacks in describ-

1114

ing modal entities, such as temporal entities,

vague knowledge, and probabilistic knowledge.
Several extensions have been proposed (Baader
et al., 2003) in that direction. These extensions
include the following: the ability to predefine
specific predicates over concrete domains (e.g.,
the greater-than or less-than operators over the
domain of integers), the use of modal extensions
in order to define concepts like belief and obligation, temporal extensions for describing time
dependence, representation of uncertain and
vague knowledge (probabilistic and fuzzy logic
extensions) and nonmonotonic reasoning. These
extensions should be incorporated into the Semantic Web standards in order to achieve maximum
expressiveness. Last but not least, collaboration
between various scientific communities should be
employed, so as to see the Semantic Web reach its
full potential, pervading in everyday life.

References
Aggarwal, R., Verma, K., Miller, J., & Milnor,
W. (2004). Constraint driven Web service composition in METEOR-S. University of Georgia,
LSDIS Lab.
Amann, P., & Quirchmayr, G. (2003). Foundation
of a framework to support knowledge management in the field of context-aware and pervasive
computing. Proceedings of the Australasian
Information Security Workshop Conference on
ACSW03 (Vol. 21), Adelaide, Australia.
Anyanwu, K., & Sheth, A. (2003). -Queries:
Enabling querying for semantic associations
on the Semantic Web. Proceedings of the 12th
International Conference on World Wide Web,
Budapest, Hungary.
Artale, A., & Fraconi, E. (1998). A temporal
description logic for reasoning about actions and
plans. Journal of Artificial Intelligence Research,
9, 463-506.

Ontological Engineering in Pervasive Computing Environments

Baader, F., Kusters, R., & Wolter, F. (2003).

Extensions to description logics. In The description logics handbook (pp. 226-268). Cambridge
University Press.
Beckett, D. (2004). RDF/XML syntax specification. Resource Description Framework. W3C
Recommendation.
Berners-Lee, T. (2001). Services and semantics: Web architecture. The 10th International
World Wide Web Conference Hong Kong.
W3C Document. Retrieved from http://www.
w3.org/2001/04/30-tbl
Boley, H. (2001). RuleML in RDF. Version 0.2.
Bouquet, P., Dona, A., Serafini, L., & Zanobini,
S. (2002). ConTeXtualized local ontology specification via CTXML. Supported by the project
EDAMOK Enabling Distributed and Autonomous
Management of Knowledge, American Association for Artificial Intelligence.
Brickley, D., & Guha, R. (2004). RDF vocabulary
description language 1.0: RDF Schema. W3C
Recommendation.
Chalupsky, H. (2000). OntoMorph: A translation
system for symbolic knowledge. Proceedings
of 7th International Conference on Principles
of Knowledge Representation and Reasoning
(KR2000), Breckenridge, CO.
Chen, H., Finin, T., & Anupam, J. (2004). An
ontology for context-aware pervasive computing
environments. Special Issue on Ontologies for
Distributed Systems, Knowledge Engineering
Review.
Ding, Y., & Foo, S. (2002). Ontology research
and development. Part 2 A review of ontology
mapping and evolving. Journal of Information
Science, 28(5), 375-388.
Doan, A., Madhavan, J., Dhamankar, R., Domingos, P., & Halevy, A. (2003). Learning to match
ontologies on the Semantic Web. The VLDB

Journal The International Journal on Very

Large Data Bases, 12(4), 303-319.
Elgesem, D. (1997). The modal logic of agency.
Nordic Journal of Philosophical Logic, 2(2),
1-46.
Fensel, D., & Bussler, C. (2002). The Web service
modeling framework WSMF. Electronic Commerce: Research and Applications, 1, 113-137.
Fernndez, M., Malhotra, A., Marsh, J., Nagy,
M., & Walsh, N. (2005). Xquery 1.0 and XPath
2.0 Data Model. W3C Working Draft 4.
Fowler, J., Nodine, M., Perry, B., & Bargmeyer,
B. (1999). Agent based semantic interoperability
in Infosleuth. Sigmod Record, 28.
Freund, Y., & Schapire (1996). Experiments
with a new boosting algorithm. Proceedings of
the 13th International Conference on Machine
Learning.
Friedman, E. J. (2005). Jess, The Rule Engine
for the Java Platform. Version 7.0a6. Sandia
National Laboratories.
Gibbins, N., Harris, S., & Shadbolt, N. (2003).
Agent-based Semantic Web Services. Proceedings of the 12th International Conference on World
Wide Web, Budapest, Hungary.
Gmez, P., Cabero, G., & Penn, L. (2004). ODE
SWS: A Semantic Web service development tool.
Version 1.0.
Guarino, N. (1995). Formal ontology, conceptual
analysis and knowledge representation. International Journal Human-Computer Studies, 43(5),
625-640.
Guarino, N., Masolo, C., & Vetere, G. (1999,
May/June). OntoSeek: Content based access to
the Web. IEEE Intelligent Systems, 70-80.
Guarino, N., & Welty, C. (2000). Towards a methodology for ontology-based model engineering.
ECOOP-2000 Workshop on Model Engineering,
Cannes, France.
1115

Ontological Engineering in Pervasive Computing Environments

Haarslev, V., & Mller, R. (1999). RACER, A

Semantic Web inference engine.
Halevy, A. Y., Halevy, Z. G., Ives, P. M., Tatarinov,
I. (2003). Piazza: Data management infrastructure
for Semantic Web Applications. Proceedings of
the 12th International Conference on World Wide
Web, Budapest, Hungary.

McGuinness, D., & van Harmelen, F. (2004).

OWL Web Ontology Language overview. W3C
Recommendation.
McIlraith S., Son, T., & Zeng, H. (2001). Semantic
Web services. 1094-7167/01/2001. IEEE Intelligent
Systems.

Horrocks, I., Patel-Schneider, P., Boley, H., Tabet, S., Grosof, B., & Dean, M. (2003). SWRL: A
Semantic Web Rule Language Combining OWL
and RuleML. Version 0.5.

Medjahed B., Bouguettaya, A., & Elmagarnid, A.

(2003). Composing Web services on the Semantic
Web. The VLDB Journal The International
Journal on Very Large Data Bases,12(4), 333351.

Khan, L., McLeod, D., & Hovy, E. (2004). Retrieval effectiveness of an ontology-based model
for information selection. The VLDB Journal
The International Journal on Very Large Data
Bases,13(1), 71-85.

Middleton, S., Shadbolt, N., & De Roure, D.

(2004). Capturing interest through inference
and visualization: Ontological user profiling in
recommender systems. ACM Transactions on
Information Systems (TOIS),22(1).

Kibler, D., & Albert, M. (1991). Instance-based

learning algorithms. Machine Learning, 6, 3766.

Milojicic, D., Kalogeraki, S., Lukose, V., Nagaraja,

Pruyne, K, Richard, J., et al. (2002). Peer-to-peer
computing. HP Laboratories Technical Report.

Lei, L., & Horrocks, I. (2003). A software framework for matchmaking based on Semantic Web
technology. Proceedings of the 12th International
Conference on World Wide Web, Budapest, Hungary.

Nakajima, T. (2003). Pervasive servers: A framework for creating a society of appliances. Personal
and Ubiquitous Computing,7(3), 182-188.

Maedche, A., & Zacharias, V. (2001). Clustering

ontology-based metadata in the Semantic Web.
Supported by the project HRMore project, FZI
Research Center for Information Technologies,
University of Karlsruhe, Germany.
Maedche, A., Motik, B., & Stojanovic, L. (2003).
Managing multiple and distributed ontologies on
the Semantic Web. The VLDB Journal The
International Journal on Very Large Data
Bases,12(4), 286-302.
Martin, D. (2003). OWL-S 1.0 Release. OWLbased Web Service Ontology.
McCallum, A., Nigam, N., Rennie, K., & Seymore,
J. (2000). Automating the construction of Internet portals with machine learning. Information
Retrieval, 3(2), 127-163.
1116

Neches, R., Fikes, R., Finin, R., Gruber, T., Senator, T., & Swartout, W. (1991). Enabling technology
for knowledge sharing. AI Magazine, 12, 36-56.
Nejdl, W., Wolpers, M., Siberski, W., Schmitz, C.,
Schlosser, M., Brunkhorst, I., et al. (2003). Superpeer-based routing and clustering strategies for
RDF-based peer-to-peer network. Proceedings
of the 12th International Conference on World
Wide Web, Budapest, Hungary.
Noy, N., & McGuinness, D. (2001). Development
101: A guide to creating your first ontology. Stanford Knowledge Systems Laboratory Technical
Report KSL-01-05 and SMI-2001-0880.
Orfali, R., & Harkey, D. (1998). Client/server
programming with Java and CORBA. Wiley.
OSullivan, D., & Lewis, D. (2003). Semantically
driven service interoperability for pervasive com-

Ontological Engineering in Pervasive Computing Environments

puting. Proceedings of the 3rd ACM International

Workshop on Data Engineering for Wireless and
Mobile Access, San Diego, CA.
Parkhomenko, O., Lee, Y., & Park, E. (2003).
Ontology-driven peer profiling in peer-to-peer
enabled Semantic Web. Proceedings of the 12th
International Conference on Information and
Knowledge Management, New Orleans, LA.
Ranganathan, A., McGrath, R., Campbell, R., &
Mickunas, M. (2003). Ontologies in a pervasive
computing environment. Proceedings of the IJCAI-03 Workshop on Ontologies and Distributed
Systems (ODS 2003), Acapulco, Mexico.
Ranganathan-GAIA. (2003). An infrastructure
for context-awareness based on first order logic.
Personal and Ubiquitous Computing,7(6), 353364.
Rapoza. (2000). DAML could take search to a
new level. PC Week Labs.
Rector. (2003). Modularization of domain ontologies implemented in description logics and related
formalisms including OWL. Proceedings of the
International Conference on Knowledge Capture,
Sanibel Island, FL.

Rector, A., & Rogers, J. (1999). Ontological issues

in using a description logic to represent medical
concepts: Experience from GALEN. Appeared at
IMIA WG6 Workshop: Terminology and Natural
Language in Medicine, Phoenix, AZ.
Roman, M., Hess, C., Cerqueira, R., Ranganathan, A., Cambell, R., & Nahrstedt, K. (2002). A
middleware infrastructure for active spaces. IEEE
Personal Communications, 1(4), 74-83
Stracia, U. (1998). A fuzzy description logic.
Proceedings of the 15th National/10th Conference
on Artificial Intelligence/Innovative Applications
of Artificial Intelligence.
The Cambridge Dictionary. (1980). The Cambridge dictionary of English. Daniel Lawrence
Corporation.
Wang. (2003). The context gateway: A pervasive
computing infrastructure for context aware services. Report submitted to School of Computing,
National University of Singapore & ContextAware Dept, Institute for Infocomm Research.
Wang, B., Mckay, R., Abbass, H., & Barlow, M.
(2003). A comparative study for domain ontology
guided feature extraction. Proceedings of the 26th
Australasian Computer Science Conference on
Conference in Research and Practice in Information Technology, Adelaide, Austria.

This work was previously published in Web Semantics and Ontology, edited by D. Taniar, pp. 334-363, copyright 2006 by IGI
Publishing, formerly known as Idea Group Publishing (an imprint of IGI Global).

1117

1118

Chapter 3.8

MASACAD:

A Multi-Agent System for

Academic Advising
Mohamed Salah Hamdi
University of Qatar, Qatar

ABSTRACT
The evolution of the Internet into the Global Information Infrastructure has led to an explosion
in the amount of available information. Realizing
the vision of distributed knowledge access in
this scenario and its future evolution will need
tools to customize the information space. In this
article we present MASACAD, a multi-agent
system that learns to advise students and discuss
important problems in relationship to information
customization systems and smooth the way for
possible solutions. The main idea is to approach
information customization using a multi-agent
paradigm.

Introduction
In our previous work (Hamdi, 2005), we have
presented an e-learning system that provides a
service to a student that checks whether lecturers

are offering information that matches the profile

of the student and informs the student of the
information found. The student is registered in
many courses and seeks course materials from
the sites of the different lecturers. These diverse
resources can be physically distributed. They
also are dynamic so that course materials can be
added, updated, or deleted. The student profile,
that includes the courses currently attended and
possibly much more information, changes also
over time because the student can leave a course
or register in a new one. All of this means that
the customized presentation of information for
the student should be updated continuously as
new information becomes available. This happens
with no user intervention using an autonomous
multi-agent system.
In this article, following the same long-term
objective of providing a complete e-learning environment for students and aiming for the more
general goal of information customization, we
describe MASACAD (Multi-Agent System for

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

MASACAD

ACademic ADvising; MASACAD is also the

Arabic word for courses), a multi-agent system
that advises students by adopting a machine
learning paradigm. The decision for employing
the multi-agent approach combined with machine
learning is supported by the following general
arguments (more specific arguments are discussed
in Section 3.5).
Humans often deal with problem complexity
by organizing teams and cooperating in the search
for solutions. Group problem solving techniques
have evolved into powerful strategies in areas
where any single individual would simply be
overwhelmed. AI research has paralleled this
approach by developing the distributed problemsolving paradigm. Problem-solving systems can
distribute among them the processes needed to
accomplish a given task.
Given the amount of problems that can be
approached through distributed processing, AI
has directed significant effort towards exploring
possibilities to develop specialized problem-solving systems that can interact in their search for a
solution. One way to embody this approach is represented by the multi-agent system paradigm.
An agent is a collection of knowledge and
methods that are intended to embody a well defined functionality. Agents cooperate in solving
a problem by contributing various parts of the
problem-solving process. Agents can be modified
independently and they are well focused on their
tasks. Such subsystems are often easier to design,
develop, and validate, than their counterparts
that combine the equivalent functionality within
one system.
Because it is not possible to account for all the
aspects of multi-agent problem-solving at development time, a good problem-solver has to be able
to compensate autonomously for the dynamic and
unpredictable aspects of its execution environment
through adaptation. Adaptation is reached by letting the system learn from experience.
In the following, we start with presenting
some background material and then introduce the

problem of academic advising and argue that a

multi-agent system would be a good approach for
addressing this problem. The paper then presents
the architecture of the proposed system and discusses in detail its individual components. The
following section discusses the student evaluation
of the system and provides preliminary evidence
that it is helpful. Benefits and limitations of MASACAD, future directions of work, and related
research are discussed respectively in each of the
next sections. The final section presents a short
summary of the paper.

Background
The MASACAD system makes use of a wide
range of technologies. These are discussed briefly
in the following paragraphs.
E-Learning: Several years ago, online education was considered as an experimental approach with more disadvantages than advantages.
However, today it should be considered not only
a complementary educational resource but also
a serious alternative that competes with conventional and now classical methods. The adaptation
to the new features and services of the e-learning environment is not immediate and requires
experience, time, investment, pedagogical and
technical resources, and government or campus
administration support. At the UAE University
there exists enormous interest in the area of online education. Rigorous steps are taken towards
the creation of the technological infrastructure
(hardware, software, and communications) and
the academic infrastructure (course materials,
teacher-student communication) for the improvement of teaching and learning. MASACAD, the
academic advising system described in this article
is to be understood as a tool that uses network
technology to support learning and as part of the
e-learning environment at the university. We use
it to demonstrate the capability of exploiting the
digital infrastructure, enabled by the online mode

1119

MASACAD

of teaching and learning, to extract and infer

useful knowledge and to point out the potential
of computational intelligence in the area of intelligent Web-based education.
Information Customization: Building
software that can interact with the range and
diversity of the online resources is a challenge
and the promise of information customization
(IC) systems is becoming highly attractive. It
would be wonderful if an IC system can identify
and present the information with little or no user
intervention. The system should also be able
to update the presentation as new information
becomes available. This will release the user
from continually observing the resources. This
raises, of course, questions about robustness and
persistence of the system. An important point in
relationship with IC systems concerns therefore
their evaluation. The best test of an IC systems
ability to help a user is a user trial. In the present
work, we evaluate the IC system for academic
advising both by standard tests (performance of
the learning algorithm) and by user trials (performance of the whole system).
User Modeling: An IC system is a kind of
software that acts in accordance with a users
preferences in an environment. To realize an IC
system acting in accordance with a users preferences, user modeling is needed. User profiles are
of great importance for information extraction
and information customization since they are
essential for deciding what kind of information
is needed, where this information can be found,
how this information can be retrieved, and how
this information should be presented to the user.
User profiles will therefore have a great influence
on the solution to be adopted for implementing
an IC system. In our case they will have a strong
impact on the multi-agent system to be created.
Agent Approach: Agents can be viewed as
a new model for developing software to interact
over a network. This view has emerged because
of the predominance of networks in the world.
Information, knowledge, and electronic resources

1120

in general, are distributed across a network and

programs and methods are needed to access them
and present them in a customized manner. Using
agents adds a layer of abstraction that localizes
decisions about dealing with local peculiarities of
format, knowledge conventions, and so on, and
thus helps to understand and manage complexity.
Agents therefore should be seen as an abstraction that appears to provide a powerful way of
conceptualizing, designing, and implementing a
particularly complex class of software systems.
Multi-agent systems are systems composed of
multiple interacting agents, where each agent
is a coarse-grained computational system in its
own right. The approach of multi-agent systems
seems to be a suitable framework for developing
IC systems since many of the properties of IC
systems or requirements on these systems coincide with those required on multi-agent systems
and on agent-based systems in general. The IC
system proposed in this article for dealing with
the problem of academic advising adopts the
multi-agent paradigm.
Machine Learning: Human expertise, needed
for solving problems, should be transferred and
transformed from some source of knowledge to
a program. This transfer is usually accomplished
by a series of lengthy and intensive interviews
between a knowledge engineer, who is normally
a computer specialist, and a domain expert who
is able to articulate his expertise to some degree.
Unfortunately, the productivity of the interviewing process is typically poor for many reasons
(see Jackson, 1999). This has led researchers to
look upon the knowledge acquisition phase as
the bottleneck problem of expert systems applications (Feigenbaum, 1977). This dissatisfaction with the interview method has encouraged
some researchers to try to automate the process
of knowledge acquisition by means of machine
learning methods. Artificial neural networks are
a particular method for empirical learning. They
have proven to be equal, or superior, to other
empirical learning systems over a wide range of

MASACAD

domains, when evaluated in terms of their generalization ability (Shavlik, Mooney, & Towell,
1991). Although the almost complete ignorance
of problem-specific theory by empirical learning systems may mean that they do not address
important aspects of induction, it is interesting to
see in the following study, how domain-specific
knowledge about academic advising of students
can be employed by a domain free neural network
learning algorithm.
Web Mining: Current Web mining research
aims at creating systems that (semi) automatically tailor the content delivered to the user from
a Web site. This is usually done by mining the
Web both the contents, as well as the users
interaction (Cooley, Mobasher, & Srivastava,
1997). To mine data from the Web is therefore
different from mining data from other sources of
information. For the problem of academic advising at hand, the Web represents the main source
of information. A solution to this problem will
therefore mine the Web. The Web mining technique adopted, bases on the multi-agent paradigm
combined with machine learning and ideas from
user modeling.

The Problem of Academic

Advising
We will illustrate our ideas using MASACAD, an
example consisting of an e-learning application.
In this application the focus is on the academic
advising for students.

Academic Advising
The general goal of academic advising is to assist
students in developing educational plans which
are consistent with academic, career, and life
goals and to provide students with information
and skills needed to pursue those goals. More
specifically, advisors will assist students in the
following ways:

Guide students through the universitys

educational requirements;
Assist in scheduling the most appropriate
courses;
Introduce them to pertinent resources;
Promote leadership and campus involvement;
Assist in career development;
Assist students with the timely completion
of their degree; and
Help students find ways to make their educational experience personally relevant.

Why is a Software Assistant

Needed for Academic Advising?
In order to help the student, improve the advising process and make it easier, and overcome the
many problems that may occur, an intelligent
assistant in form of a computer program will be
of great interest. Such an intelligent assistant will
automate the advising process in the university.
It will also simplify the task of faculty, staff,
students, and professional advisors and make
it possible to save time and effort and prevent
mistakes. These benefits are added to the many
other advantages of any assistant software that
is used to solve problems that ordinarily require
human expertise.

Restriction of the General Goal

of Academic Advising
The goal of academic advising, as stated previously, is too general because many experts are
involved and because a huge amount of expertise is needed. Hence, realizing an intelligent
software assistant that is able to deal with all the
details shown previously will be too difficult, if
not impossible. In the following implementation
we will therefore restrict academic advising and
understand it as just being intended to provide
the student with an opportunity to plan programs
of study, select appropriate required and elective

1121

MASACAD

classes, and schedule classes in a way that provides

the greatest potential for academic success.
The task is still interesting and of moderate
size since when planning a program of study and
selecting classes, there are quite a lot of things
to consider such as prerequisites, course availability, effective course sequencing, work load,
and instructor-student match-ups. Later, when the
major problems are understood, improvements
and extensions can be attempted, and attempts
can be made to tackle the advising problem in a
more general framework.

Resources Needed
for Academic Advising
There are a lot of diverse resources that are
required to deal with the problem of academic
advising (see Table 1).
First of all, one needs the student profile that
includes the courses already attended, the corresponding grades, the interests of the student
concerning the courses to be attended, and perhaps
much other information. The part of the profile
consisting of the courses already attended, the
corresponding grades, and so on, is maintained
by the university administration in appropriate
databases to which the access is restricted to some
administrators. The part of the profile consisting
of the interests of the student concerning the

courses to be attended exists actually only in the

head of the student and should therefore be asked
for from the student before advising is performed.
However, attempts may be made to let the system
learn the interests, for example, by monitoring the
student or/and looking at his profile.
The second resource needed for solving the
problem of academic advising are the courses that
are offered in the semester for which advising is
needed. This information is as well maintained
by the university administration in appropriate
Web sites and is accessible for everyone.
The third resource needed for solving the
problem of academic advising is expertise. Expertise is the extensive, task-specific knowledge
acquired from training, reading, and experience.
It is the knowledge that allows experts to make
better and faster decisions than non-experts when
solving complex problems. It consists of facts,
theories, as well as rules and procedures about
a problem area.
For the problem of academic advising this type
of expertise may be referred to as the university
regulations concerning academic advising. They
consist of all the details concerning courses,
programs, and curricula such as names/codes of
courses, in which term (fall, spring, etc.) courses
are offered, prerequisites for courses, descriptions
of courses, teaching and assessment methods,
statuses (compulsory, elective, etc.) of courses,

Table 1. Information resources for the problem of academic advising

Student Profile
Interests of
the Student
E ntered by the student
A ttempts can be made
to learn them by
monitoring the user
and/or looking to his
profile
Frequently changed

1122

Courses Already
Attended
A ll information
concerning courses
already attended by the
student
S tored in a database
A ccess is restricted to
the administration
Frequently changed

Other
Information
Other useful
information
about the
student

Courses Offered

Expertise

A ll information
available about
courses offered
in the semester
for which
advising is
needed
A vailable online
and accessible
for everyone
Frequently
changed

Documented knowledge:

A ll details and regulations concerning

courses, programs, and curriculums
A vailable in many forms: Web, booklets,
A ccessible for everyone
Rarely changed

K nowledge of a human expert:

H euristics, strategies, meta-knowledge,

M ore complex than documented
knowledge

E xpertise is gained through a learning

process

MASACAD

number of credit hours of courses, definitions

and implications of curriculum terminology (e.g.,
cross-listing, interdisciplinary, etc.), established
programs, established tracks, curriculum total
credit hours and their distribution (e.g., compulsory vs. elective), titles of academic degrees
offered by the university, and so on. This kind
of information is provided by the university and
is accessible for everyone. It is published in Web
pages, booklets, and in many other different
forms such as printouts and announcements. The
sources of knowledge are many; however, the
primary source will be a human expert. A human
expert should posses more complex knowledge
than can be found in documented sources. This
kind of knowledge includes rules (heuristics) of
what to do in a given problem situation, global
strategies for solving specific problems, and
meta-knowledge.
Because, as mentioned earlier, knowledge acquisition is the major bottleneck in developing
expert systems, we will approach a solution to
gaining the expertise needed for academic advising by adopting a machine learning paradigm.

Why a Multi-Agent System?

First of all, academic advising is intended to be a
good domain and problem to test the adequacy of
the multi-agent paradigm for dealing with information customization. It provides a complex and
dynamic environment and constitutes a wonderful
experimental test-bed for investigating the issue.
Conversely, dealing effectively with the problem
of academic advising will require a multi-agent
system. The multi-agent paradigm seems to be
appropriate and even superior to other approaches
such as a traditional distributed database approach
enhanced with an intelligent user interface for the
reasons elaborated in the following.
The resources needed for academic advising
are physically distributed and dynamic:

Their content may change: It is for example

frequent that, at the beginning of a semester,
some of the offered courses are canceled,
and some other ones are added. The profile
of the student also changes frequently, for
example when the grades of the exams taken
are entered. Also the interests of the student
concerning the courses to be attended may
change. As an example, students usually
want to enroll themselves in courses that are
attended by some of their friends. Changes
to the university regulations, in contrast,
especially those concerning the advising
process like the curriculums for the different
majors, are comparatively less frequent.
Their form (structure) may change: They
may be available via a Web page, an intelligent agent, a data base, a legacy system,
and so on.
Their location may change: Existing ones
may be moved and new ones may be incorporated.

Hence, dedicating a separate intelligent agent

to each individual resource for coping with all
of its peculiarities will have many advantages
such as:

Reduction of the scope of changes: There is

no need to change the whole system when
changes concern only a specific resource.
In this case, only the concerned agent is
changed.
Easy incorporation of new resources: Only
one new agent is needed for each new resource.
Easy extension and improvement of the
system: We may think, for example, of a self
adjusting system, that is, the system itself
searches for appropriate resources and each
time a resource is identified, it is wrapped
with an appropriate agent.

1123

MASACAD

Details of the
Multi-Agent-Based Solution
MASACAD, the multi-agent system described in
the following, offers a service to a student who
needs academic advising, that is, he wants to
know which courses he should enroll himself in
for the coming semester. The service is currently
available only for computer science students. In
the current version, no attempt is made to learn
the student profile, that is, the part the profile
consisting of the interests of the student regarding the courses to be attended. The interests are
therefore asked for from the student before advising him. The focus in the current version is on
the individual agents of the system, on how they
cooperate to solve the problem, and on how one
of them, namely, the Learning Agent, learns to
perform the advising process by adopting a supervised learning solution using neural networks
(see Advising Procedure).
As a solution to the problems of network
communication, we use Bee-gent (Bonding and
Encapsulation Enhancement Agent) (Kawamura,
Hasegawa, Ohsuga, & Honiden, 2000), a communication framework based on the multi-agent
model. The Bee-gent framework is comprised of

two types of agents. Agent Wrappers are used

to agentify (i.e., providing an agent interface)
existing applications, while Mediation Agents
support inter-application coordination by handling
all communications. The mediation agents move
from the site of one application to another where
they interact with the agent wrappers. The agent
wrappers themselves manage the states of the
applications they are wrapped around, invoking
them when necessary. MASACAD consists of 3
Agent Wrappers (Learning Agent, Data Base
Agent, WWW Agent) and one Mediation Agent
(Searcher).

System Architecture of MASACAD

MASACAD consists of a mediation agent that
provides the information retrieving service
(Searcher) and of a User System, a Grading
System, and a Course Announcement System
that are technically applications, each of them
existing within an agent wrapper, that is, they
represent different agents of the advising system
(see Figure 1).
The student (user) uses a Graphical User Interface (GUI) to utilize the service of the advising
system. When the student enters his identifier and

Figure 1. Process flow through MASACAD

DB
JDBC
access

GUI
Application
GUI
Operation
Events

2.Move

Display
User
System

1.Request for info

9.Info Reply

IRS
Searcher

M obile
M ediation
A gent
8.Move

1124

Grading
System

4.Info Reply
D ata B ase A gent

IRS
Searcher

5.Move

Learning A gent

GUI: Graphical User Interface

JDBC: Java DataBase Connectivity

3.Request for info

IRS
Searcher

URL

6.Request for info

7.Info Reply

IRS: Information Retrieval Service

DB: Data Base URL: Uniform Resource Locator

URL
connection
Course
Announcement
System
W W W A gent

MASACAD

password, the agent wrapper for the User System

application creates a mediation agent (Searcher),
which migrates to the agent wrappers of the
Grading System application and of the Course
Announcement System application to retrieve the
needed information (courses successfully taken
by the student with the corresponding grades
and courses offered in the semester for which
advising is needed). After that, the student has
to enter his interests by choosing courses among
those that are offered in the semester for which
advising is needed by checking some boxes showing the course names. When the student finishes
entering his interests, he has to just click on the
Get Advice button to obtain a list of courses
sorted in descending order of importance in which
he is advised to enroll himself in.
In the following, each of the individual components of the system is be described in detail.

Grading System
The application Grading System is a data base
application for answering queries about the students and the courses they have already taken.
The agent wrapper for the application Grading
System is responsible for invoking the database
by using JDBC (Java DataBase Connectivity),
transforming requests for information about the
students and courses into queries to the database
system, collecting the results of the queries, and
finally replying these results to the mediation
agent. Table 2 summarizes the steps performed
by the agent wrapper of the application Grading System.

Table 2. Agent wrapper for Grading System

Receive request from mediation agent

Perform data base access setup

Perform information retrieval

Inform the mediation agent of the result

Course Announcement System

The application Course Announcement System
is a Web application for answering queries about
the courses that are expected to be offered in the
semester for which advising is needed. The agent
wrapper for the application Course Announcement System takes on the invocation of the Web
application. It is responsible for initiating communication with the Web application, transforming
requests for information about the courses into
queries to the Web application, collecting the
results of the queries, and finally replying these
results to the mediation agent. Table 3 summarizes
the steps performed by the agent wrapper of the
application Course Announcement System.

Mediation Agent
The mediation agent realizes services by interacting with the agent wrappers on the basis
of conversations (sending and receiving XML
[eXtensible Markup Language] messages). When
the mediation agent migrates, it carries its own
program, data and current state. Frequency of
communication is reduced compared to a purely
message-based system and network loads are decreased largely because communication links can
be disconnected after the launch of the mediation
agent. Processing efficiency is improved because
the mediation agent communicates with the applications locally. The behavior of the mediation
agent is described in Table 4.

User System
The agent wrapper for the User System application creates a mediation agent (Searcher), which
migrates to the agent wrappers of the applications
Grading System and Course Announcement
System to retrieve the needed information.
After that, the GUI Application is contacted
to give the opportunity to the student to express
his interests. The Advising Procedure (see next

1125

MASACAD

Table 3. Agent wrapper for Course Announcement System

Receive request from mediation agent

Initiate communication with the URL

Prepare data to be sent to the Web server (this data consists of the request for information communicated
by the mediation agent represented in the appropriate way as expected by the Web application)

Send data to the Web server (the CGI (Common Gateway Interface) program on the server site reads this
information, performs the appropriate actions, and then sends information back to the agent wrapper
via the same URL)

Read data from the URL

Extract information from this data; i.e., extract course identifiers of offered courses

Inform the mediation agent of the result

section) is then started and results are sent back to

the GUI Application. The agent wrapper for the
User System application also monitors changes
in the offered courses and in the student profile
and alerts the user by sending an e-mail. Table
5 summarizes the steps performed by the agent
wrapper of the application User System.

The Advising Procedure

The Advising Procedure is focused on advising
computer science students. Computer science
students, in order to complete their degree, must
accomplish during their study a total of 132 credit
hours by choosing at each term courses among the
85 courses that are specified by the curriculum.
Some of these courses are compulsory, others are
elective. The choice should of course occur according to the university regulations and in a way
that provides the greatest potential for academic
success as seen by a human academic advisor.
Taking into account the adequacy of the
machine learning approach for gaining human
expertise added to the availability of experience
with advising students makes the adoption of a
paradigm of supervised learning from examples
obvious. Back-propagation is the best-known
example of a supervised neural network training
algorithm Patterson (1996) and is therefore used to

1126

implement the Advising Procedure. The known

information (input variables) consists of the profile
of the student and of the offered courses grouped
in a suitable way. The unknown information (output variables) consists of the advice expected by
the student. In order for the network to be able to
infer the unknown information, prior training is
needed. Training will integrate the expertise in
academic advising into the network.

Settings of the Neural Network Solution

For the problem of academic advising, information
about 250 Computer Science students in different stages of study was collected, which means
that 250 examples were available for the learning
procedures. The 250 examples were used to form
three sets: a training set containing 150 examples,
a selection set consisting of 50 examples, and a
test set consisting of 50 examples. In an attempt
to reduce the number of parameters that should
be tuned, we decided to use two hidden layers.
To implement the network, a slightly modified
version of the package developed by Mark Watson
(2005) was used.
Each one of the 250 examples consists of a pair
of input-output vectors. The input vector summarizes all the information needed for advising
a particular student. It consists of 85 components

MASACAD

Table 4. Mediation agent

Receive request for information retrieval from agent wrapper for User System
Move to agent wrapper for Grading System
Judge the migration result:

If failure: end

If success:

Request for retrieval of (course, grade) pairs for the given student ID

Receive the reply

Move to agent wrapper for Course Announcement System

Judge the migration result:
If failure: end
If success:

Request for retrieval of IDs of offered courses for the given semester, college, and subject

Receive the reply

Move to agent wrapper for User System

Judge the migration result:

If failure: end

If success:

Report retrieved results to agent wrapper for User System

End

Table 5. Agent wrapper for User System

When GUI event occurs (student enters ID, Password, and Email address the first time, or asks again for
advice by clicking the button Get Advice Again) or when periodic monitoring of changes in student profile
and offered courses is due:
Create mediation agent
Request mediation agent to perform information retrieval
Receive results from mediation agent
If information retrieval was initiated by a GUI event then:

Output information concerning the offered courses to the GUI (the GUI Application represents
this information appropriately (boxes that can be checked by the student))

When GUI event occurs (student clicks on Get Advice button):

Prepare input (input vector stored in a text file) for Advising Procedure

Invoke Advising Procedure

Process results (the results are found in a text file) and output them to GUI

Else (periodic monitoring): compare retrieved information with old version and alert the user
via Email in the case that something has changed

each of them containing a number for one of the 85

courses in the curriculum. The number consists of
the grade earned if the student has already passed
the given course. Otherwise, the number encodes
the availability of the course in the semester for
which advising is needed and the interest of the

student in the course. The output vector encodes

the final decision concerning the courses in which
the student actually enrolled himself in based on
the advice of the academic advisor. This happens
by assigning priority values to each of the 85
courses in the curriculum.

1127

MASACAD

Learning Phase
The aim of the learning phase was to determine
the most suitable values for the learning rate, the
size of the network, and the number of training
cycles that are needed for the convergence of the
network.

Figure 2. Determination of the best network size

(learning period = 1000000 cycles)
Le arnin g rate

= 0 . 01

A v er a ge er r or
800

700

Learning Rate: To determine the most suitable learning rate, experiments with 10 network
configurations 85-X-X-85 with X in {3, 5, 7, 9,
20, 30, 50, 85, 100, 120} were performed (85-XX-85 represents a network with 85 input neurons,
X neurons in each of the two hidden layers, and
85 output neurons). For each of these networks,
experiments with the five learning rates: 0.01,
0.25, 0.50, 0.75, and 1.0 were conducted. In each
of these 50 experiments, the network was allowed
to learn for a period of 1000000 cycles. After
each epoch of 50000 cycles the average selection
error for the 50 pairs from the selection set was
calculated. From the 50 experiments it was clear
that the learning rate 0.01 is the most suitable
one for this application because it produces the
smallest selection error in most of the cases and
more importantly, it causes the selection error
to decrease continuously which forebodes the
convergence of the network.
Network Size: Figure 2 shows the average
selection error for the 50 pairs from the selection
set plotted as a function of the number of training
cycles for the 10 different network configurations
85-X-X-85 with X in {3, 5, 7, 9, 20, 30, 50, 85,
100, 120}. In all 10 cases the network was allowed
to learn for a period of 1000000 of cycles and
the learning rate was set to 0.01. The network
configurations 85-X-X-85 with X in {50, 85, 100,
120} seem to work best in combination with the
learning rate 0.01. To determine which one of
them is more suitable, a longer learning period
is needed. The results for a learning period of
10,000,000 cycles are illustrated in Figure 3. The
configurations 85-100-100-85 and 85-120-120-85
cause the selection error to decrease continuously

1128

600

3
5

9
7

500

20
30

400

50

300
85

100 120

200

100

0
0

200000

400000

600000

800000

1000000

1200000

Tr a ining cy c les

and reach a very satisfactory level. Both networks

have similar performances. However, following
the standard scientific precept that, all else being
equal, a simple model is always preferable to a
complex one, we can also select a smaller network
in preference to a larger one with a negligible
improvement in selection error. The network
configuration 85-100-100-85 is therefore retained
as the best network configuration (in terms of
selection error).
Number of Training Cycles: After a given
number of training cycles, when the selection
error and consequently the performance reaches
a satisfactory level and when the improvement
becomes negligible, learning can be stopped and
the network producing the smallest selection error can be retained as a solution for the problem
of academic advising. In our case, this happens
with the learning rate 0.01 and the network configuration 85-100-100-85 after a training period
of 8000000 cycles.

MASACAD

Figure 3. Determination of the best network

size
Le arnin g pe rio d

= 10 000 000

A v er a ge er r or
700

600

500
50

400

300

85
200

100

100
120

0
0

2000000

4000000

6000000

8000000

10000000

12000000

Tr a ining cy c les

Testing Phase
The final model was tested with the test set data.
For 42 of the 50 test cases (84 percent) the networks
actual output was exactly the same as the target
output, that is, the network suggested the same
courses in the same order as specified by the test
examples. In the remaining eight test cases, the
target courses were always present at the beginning of the course list produced by the network.
However, the network proposed some additional
courses. The courses proposed additionally occur
always at the end of the course list. This, in addition to the fact that the system is an advisory one,
makes these errors tolerable. In four of the eight
cases the additional courses were correct choices.
In the other four cases some of the additional
courses were wrong choices because the student
has not yet taken their prerequisite courses.

Benefits and Limitations of the

Neural Network Solution
The research reported showed the applicability of
the neural network approach to academic advising.
The rewards of a successful marriage of neural
networks and expert systems are too enticing. The
back-propagation network performed favorably
and seems interesting and viable enough to be
used to automate the process of knowledge acquisition, usually looked upon as the bottleneck
problem of expert systems applications, and help
in extracting human expertise, needed for solving
the problem of academic advising. In this vein, a
possible way to incorporate neural networks into
expert systems was described. This would smooth
the way for a real and significant fusion of neural networks and expert systems. With a network
topology of 85-100-100-85 and systematically
selected network parameters (learning rate of 0.01,
8000000 training cycles), the multiple-layered,
fully connected back-propagation network was
able to deliver a considerable performance.
Despite the promising results demonstrated in
various applications, back-propagation network
development still requires extensive experimentation, parameters selection, and human judgment.
Developing a good neural network is not a trivial
task as the technique currently stands. Some
interesting research questions about adopting
neural networks remain to be answered. For example, can we develop a systematic and simple
methodology and blueprint for neural network
modeling and analysis? Can we automate the
selection of topology and parameters? Can we let
the network learn incrementally without major
network re-training when new information is discovered (e.g., curriculum changes)? Incremental
learning would require the exploration of other
neural network models, such as fuzzy ARTMAP
(Carpenter, Grossberg, Markuzon, Reynolds, &
Rosen, 1992) and ARAM (Tan, 1995).

1129

MASACAD

System Evaluation
To evaluate the advising multi-agent system in
real academic environment, 20 computer science
students in different stages of study were involved.
Each of them was asked to use the system to get
academic advice for the coming term. Most of
the problems that occurred at the beginning were
of technical nature, mainly concerning the communication over the network. It took a while to
deal with this kind of problems and to obtain a
relatively stable system.
The advice delivered by the system in each
of the 20 cases was analyzed carefully by the
concerned student together with an academic
adviser with the aim of evaluating its suitability
and detecting possible errors. The results were
very encouraging. In none of the 20 cases was the
advice found to contain errors, that is, to contain
courses that, if chosen, will violate the university regulations such as those concerning course
prerequisites. Also, all of the results (suggested
courses) were judged by the concerned students
and academic advisers to be suitable and in most
of the cases even very appropriate to be taken by
the students. Actually, 15 of the 20 students took
during the term for which advising was needed
exactly the courses suggested by the advising
system. In one of the remaining five cases, the
student, because of her poor health, was later (at
the beginning of the term) forced to drop three
courses and to take only two of the five suggested
ones. In another case, the student didnt take any
course during the term because of leave of absence
for the whole term. In the other three cases, each
of the students exchanged a suggested course (the
same course in all three cases) with another one.
Later, it became clear that the three students were
in a clique and that one of them didnt care for
the instructor of the course, so they moved all to
the other course.

1130

Benefits and Limitations of

MASACAD
MASACAD, in addition to providing a solution
for the problem of academic advising, was also
used to demonstrate, on an example, the suitability of the multi-agent paradigm for dealing with
information customization and personalization.
Information customization and personalization
provides users with relevant content based on
their needs and interests and this is exactly what
MASACAD does. It provides the user (student)
with an interactive, personal Web experience and
customizes the information to the users desire
and needs. The preferences (profile) of the user
determine the behavior of the system. Using
the multi-agent approach combined with other
techniques, we were able to devise an elegant
solution to a problem that depends on distributed
information existing in differing formats. Conversely, as mentioned earlier, the distributed and
dynamic nature of todays information resources
makes the adoption of the multi-agent approach
necessary.
However, there are some other important characteristics of information customization systems
that were not demonstrated on this example. In
general, in an information customization system
there should be some strategy for discovering
the location of potential useful information for
example among the whole Web, among a reasonable part of the Web, among a set of databases, or
among any other heterogeneous set of resources.
Also it should be possible to automate the creation
of the mediation agent and the agent wrappers,
that is, the creation of the whole multi-agent
system. The multi-agent system to be created
depends on the resources to be invoked and on
their locations. The automation task may therefore
be very difficult: the resources can be numerous
and may have very different formats (they may be
unstructured Web pages, databases, agents) and

MASACAD

this complicates the creation of the agent wrappers (how to wrap an unknown application?), as
well as the creation of the mediation agent (which
route to take to navigate to the different agent
wrappers, how to communicate with the different
agent wrappers).

Future Directions of Work

As mentioned earlier, the MASACAD system in
its current version is available only for computer
science students. The next step for this work is
therefore to focus on extending the system to cope
with all other degrees in the university. This will
imply that for each of these degrees, data about
advising students should be collected in order to
construct the training, selection, and testing sets.
Also, for each degree, a neural network should be
developed which requires extensive experimentation, parameters selection, and performance
judgment.
However, before extending the system to cope
with other degrees, the system should be tested on a
larger scale to detect possible errors, be convinced
of its usefulness, and better assess its power. This
will be done by trying to involve all Computer
Science students (more than 300) in evaluating
the system. Before allowing a student to register
to any course, the student should get advice from
the system and discuss its appropriateness with
his human academic adviser.
Another improvement, necessary to think
about, is how can we let the neural networks
learn incrementally without major network retraining when new information that is suspected
to affect the process of academic advising is
discovered (e.g., curriculum changes)? This will
make the system more flexible and avoid the
repeated new development of neural networks
each time a change occurs.
Another problem that should be considered
is how can the system be made flexible against
changes in the location of course announcements

and student profiles (where do they reside?) and

against changes in the type of application that
enables access to this information (is it a Web
site, a data base, an intelligent agent, etc.?). To
solve this problem, one can think of automating
the multi-agent system, that is, creating the mediation agent and the agent wrappers automatically.
This will, of course, require knowing which
sites should be wrapped by agent wrappers (i.e.,
knowing the location of resources), and how this
wrapping should be done (based on the type of
the site/application).

Related Research
In recent years, the administration of academic
advising has undergone radical transformation as
technological developments have altered the processes by which information is collected, stored,
and accessed; the systems by which communication is enabled; and the structures by which transactions are conducted. Technological innovations
have created an abundance of opportunities for
new practices and enhanced services frequently
characterized as real-time, student-centered,
and any time, any place (Moneta, 1997, p. 7). The
technological environments on many campuses
are evolving rapidly and comprise numerous elements: information dissemination, transactional
interaction, communications applications, and
educational technologies. Steele and McDonald
(n.d.) contains an annotated bibliography compiled
by George Steele and Melinda McDonald for
research related technology and advising.
The problem of course assignment has been
studied in the literature from different angles.
Advising tools are usually intended to complement the student-advisor relationship. The courseplanning consultant (CPC), for example, is a
computerized Web-based prerequisite checker
developed for students in the General Engineering
undergraduate curriculum at the University of Illinois at Urbana-Champaign. It was reported that

1131

MASACAD

the program has been field-proven to have helped

prevent costly mistakes when planning future
courses (Course Planning Consultant, n.d.).
There are also few learning environments
where similar problems, such as learning objects
assignment, were addressed using the multi-agent
system approach. In Sheremetov and Arenas
(2002), for example, personalized learning trajectories are constructed by agents in the learning
space based on users profiles.
It is surely difficult to compare the performance
of MASACAD with these systems because they
do not perform exactly the same task. However,
the technical novelty of MASACAD seems to lay
in its ability to perform academic advising while
taking into account not only visible static information but also hidden expertise. MASACAD
also demonstrates the capability of exploiting
the digital infrastructure, enabled by the online
mode of teaching and learning, to extract and
infer useful knowledge. This is done in an elegant
way through a multi-agent system inspired from
human teams and organizations involved in the
problem-solving activities.

Conclusion
Academic advising is an important area in any
university, which seldom gets adequate resources.
Hence, any system that can support this operation
will be worthwhile. The idea of realizing such a
system is very interesting and highly related to
the current research and development trends. In
this paper, MASACAD, a well defined architecture for a multi-agent system for addressing the
problem of academic advising was proposed. It
uses intelligent agents and neural networks for
learning and recommendation. The prototype
implementation and preliminary testing show
that the multi-agent paradigm, combined with
ideas from machine learning, user modeling, and
Web mining, would be a good approach for the

1132

problem of information customization. It was very

interesting to see an e-learning system that is able
to learn and react according to the characteristics
of the client (in this case the student). The results
are encouraging and future work will be focused
on improving the system and studying how such
simple examples, built with insight, should lead
to the identification of key difficulties, useful
abstractions, and a general method for solving
the problem and revelation of the issues.

References
Carpenter, G. A., Grossberg, S., Markuzon, N.,
Reynolds, J. H. & Rosen, D. B. (1992). Fuzzy
ARTMAP: A neural network architecture for
incremental supervised learning of analog multidimensional maps. IEEE Transactions on Neural
Networks, 3, 698-713.
Cooley, R., Mobasher, B. & Srivastava, J. (1997,
November). Web mining: Information and pattern
discovery on the World Wide Web. In Proceedings of the 9th IEEE International Conference on
Tools with Artificial Intelligence (ICTAI97) (pp.
558-567), Newport Beach, CA.
Course Planning Consultant (CPC). (n.d.). Background and guest user information. Retrieved May
28, 2005, from http://www.ge.uiuc.edu/ugrad/advising/cpc.html
Feigenbaum, E. A. (1977). The art of artificial intelligence: Themes and case studies of knowledge
Engineering. In Proceedings of the 5th International Joint Conference on Artificial Intelligence
(pp. 1014-1029), Cambridge, MA.
Hamdi, M. S. (2005). Extracting and customizing
information using multi-agents. In A. Scime (Ed.),
Web mining: Applications and techniques (pp.
228-252). Hershey, PA: Idea Group Publishing.
Jackson, P. (1999). Introduction to expert systems
(3rd ed.). Harlow, England: Addison-Wesley.

MASACAD

Kawamura, T., Hasegawa, T., Ohsuga, A. &

Honiden, S. (2000). Bee-gent: Bonding and encapsulation enhancement agent framework for
development of distributed systems. Systems and
Computers in Japan, 31(13), 42-56.
Moneta, L. (1997). The integration of technology
with the management of student services. New
Directions for Student Services: Using Technology to Promote Student Learning, 78, 5-16.
Patterson, D. (1996). Artificial neural networks.
Singapore: Prentice Hall.
Shavlik, J. W., Mooney, R. J., & Towell, G. G.
(1991). Symbolic and neural net learning algorithms: An empirical comparison. Machine
Learning, 6, 111-143.

Sheremetov, L. & Arenas, A. G. (2002). EVA:

An interactive Web-based collaborative learning
environment. Computers & Education, 39(2),
161-182.
Steele, G. & McDonald, M. (n.d.). Technology
and advising. Retrieved May 28, 2005, from
http://www.nacada.ksu.edu/Clearinghouse/Research_Related/advtechbib.htm
Tan, A. H. (1995). Adaptive resonance associative
map. Neural Networks, 8(3), 437-446.
Watson, M. (2005). Practical artificial intelligence programming in Java. Retrieved May 28,
2005, from http://www.markwatson.com/opencontent/

This work was previously published in International Journal of Intelligent Information Technologies, Vol. 2, Issue 1, edited
by V. Sugumaran, pp. 1-20, copyright 2006 by IGI Publishing, formerly known as Idea Group Publishing (an imprint of IGI
Global).

1133

1134

Chapter 3.9

Human Factors Problems of

Wearable Computers
Chris Baber
The University of Birmingham, UK
James Knight
The University of Birmingham, UK

Abstract

INTRODUCTION

In this chapter wearable computers are considered from the perspective of human factors. The
basic argument is that wearable computers can
be considered as a form of prosthesis. In broad
terms, a prosthesis could be considered in terms
of replacement (i.e., for damaged limbs or organs),
correction (i.e., correction to normal vision or
hearing with glasses or hearing aids), or enhancement of some capability. Wearable computers offer
the potential to enhance cognitive performance
and as such could act as cognitive prosthesis,
rather than as a physical prosthesis. However,
wearable computers research is still very much
at the stage of determining how the device is to
be added to the body and what capability we are
enhancing.

There is a wide range of technologies that have

been developed to be fitted to the person. Depending on ones definition of technology, this could
range from clothing and textiles, through to spectacles, to cochlear implants. The use of these different technologies can be basically summarized
as the supplementation or augmentation of human
capability, for example the ability to regulate
core temperature (clothing), to see (spectacles)
or to hear (cochlear implant). One reason why
such supplementation might be required is that
the current capability does not fit with environmental demands, either because the environment
exceeds the limits over which the human body
can function or because the capability is impaired
or limited. From this perspective, a question for
wearable computers should be what are the cur-

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Human Factors Problems of Wearable Computers

rent human capabilities that are exceeded by the

environment and require supplementation by
wearable computers? In the majority of cases for
wearable computers, the answer to this question
hinges on communicative, perceptual, or cognitive
ability. As Clark (2006) notes, ...the use, reach,
and transformative powers of these cognitive
technologies is escalating(p. 2). but the essential
point to note is that Cognitive technologies are
best understood as deep and integral parts of the
problem-solving systems that constitute human
intelligence(p.2). Thus, such technologies could
represent a form of cognitive prosthesis in that
they are intended to support cognitive activities,
for example, having a camera (performing facerecognition) to advise the wearer on the name
of the person in front of them. The immediate
challenge is not necessarily one of technology
but of cognition. If the technology is doing the
recognition, the question is raised what is the
human left to do? To draw on a commonly cited
analogy, spectacles serve as a perceptual prosthesis, that is, to improve or correct a persons
vision. For wearable computers (and the related
field of augmented reality), the improvement
could be to reveal to the person objects that are
not present by overlaying an artificial display
onto the world. The display could simply take
the form of labels or directional arrows, or could
be more a sophisticated presentation of moving
(virtual) objects. In both cases, the augmentation
could either enhance the persons understanding of the environment or could substitute this
understanding.
In terms of communication, mobile telephones
and MP3 devices contain significant computing
power such that they can easily be considered as
being computers, albeit with limited functionality. These devices can be worn, for example MP3
players can be worn on the upper arm, attached to
belts or neck-straps, or placed in hats, and mobile
telephones can be attached to belts. Furthermore,
with the wireless (Bluetooth) headset, the user
interface of a mobile telephone can be worn on the

ear at all times. Thus, both can be always present

and both can be considered part of the person. A
definition of wearable computers ought to, at least,
allow differentiation from devices that can slip
into the users pockets (if this technology is to be
treated as a new area of research and development).
Two early definitions of wearable computers,
from Bass (1996) and Mann (1997), emphasize
that wearable computers are designed to exist
within the corporeal envelope of the user and that
this makes them part of what the user considers
himself or herself. In many respects this allows an
analogy to be drawn between wearable computers
and prosthetic devices. Having something added
to the body, whether externally, such as spectacles,
artificial limbs, hearing aids, or internally, such
as pace-makers, or cochlear implants, changes
the performance of the person and (for external
prosthesis) the appearance of the body. This now
becomes a very different concept from the mobile
telephone, MP3 player and the computer that we
traditionally encounter. This raises all manner
of interesting questions relating to physical, perceptual, and cognitive aspects of human factors,
as well as a whole host of emotional aspects of
wearing devices (for the wearer and the people
with whom they interact). At the moment, there
remains a gap between what a wearable computer
is intended to be and what mobile telephones and
MP3 players currently are. This gap can best be
considered as a form of perceptual and cognitive
prosthesis, in which the wearers ability to view
the world, retrieve pertinent information, and
respond to environmental demands are enhanced
by the technology across all aspects of everyday
life. At present mobile telephones and MP3 players are able to be tailored (by the user) and can
deal with a limited set of situations (relating to
communications or music playing) but do not fill
the specification that one might have for a wearable computer. The basic difference between a
wearable computer and these other technologies
lies in the question of how well the user can interact with both the device and the environment

1135

Human Factors Problems of Wearable Computers

simultaneously. Obviously, listening to an MP3

player or speaking on a mobile telephone can be
performed while walking through an environment. However, performing control actions on
the devices can be sufficiently demanding to draw
the users attention from the environment. The
ideal wearable computer would allow the user to
manage attention to both device and environment.
In other words, it should allow the user to manage both foreground and background interaction
(Hinckley et al., 2005).

Moores law continues to guarantee that the processors will get smaller, and work in the field of
micro-electrical-mechanical systems (MEMS)
shows how it is possible to create application
specific processors that are small enough to be
incorporated into buttons on clothing or into
jewelry (Figure 1a).
Furthermore, it is feasible to assume the
widespread development of general purpose processors, such as the mote (and similar) concept,
that combine low power with sufficient processing capability to deal with a number of different
sensors (figure 1b).
One direction for wearable computers is that
the miniaturization of technology will mean
it is possible to implant processors under the

skin (or embed them into clothing). This is not a

particularly novel idea as the medical world has
been experimenting with mechanical implants for
many years and has, over the past decade or so,
developed digital implants that can, for example,
regulate drug administration or improve hearing
ability. While the development of such technology
is exciting and likely to lead to fascinating discoveries, it lies somewhat outside the remit of this
chapter. The problem is how can these extremely
small devices support interaction with the person
wearing (or supporting) them? To a great extent,
the human simply has these devices fitted into
them. While this represents the logical extension
of the argument that the wearable compute exists
within the wearers corporeal envelope, it does
preclude the possibility of developing means of
interacting with the device. Indeed, as technology
continues to get smaller, ever more difficult the
challenge of supporting human-computer interaction becomes, for example, how can one read
displays with very small font or press buttons that
are much smaller than the human finger? This
could mean that we should redefine our concept
of interaction, for example, one could argue that
interaction should be any activity that the person
performs, that is sensed by the device and that
allows the device to make a response. A potential
problem with such a concept is that the person
might not be able to selectively control what the
device senses or to fully understand why the
device is behaving in the manner that it is. This

Figure 1a. MEMS device

Figure 1b. Intels Mote prototype1

FORM-FACTOR AND PHYSICAL

ATTACHMENT

1136

Human Factors Problems of Wearable Computers

means that one either locks the human out of the

interaction (in which case one has a sensor system
that acts on the environment and which coincidentally affects people in that environment) or one
must develop different ways in which the person
can understand and manage the behavior of the
device. This is not a trivial problem and one can
point to many developments in the ubiquitous and
pervasive computing domain in which the role of
the human in device performance is merely one
of passive recipient of device activity. From the
point of view of human-computer interaction, this
will not only lead to the irony of automation2
but also to frustration, annoyance and a perceived
lack of control. In other words, who wants to live
in a world that is basically run by the descendants
of the Microsoft Office Paperclip?

Physical Effects of Wearing

Computers
We might expect a computer to consist of a processor and storage, some form of power, a display,
and an interaction device. For wearable computers,
each of these components can be separated and
worn on different parts of the body. Thus, as the
prototypes in Figure 2 illustrate, the processor and

storage could be incorporated into a single unit

and mounted on the waist or the back, the display
could be mounted on the head (either near the eyes
or the ears, depending on the feedback provided
to the wearer), the power could be located near the
processor unit, and the interaction device could be
mounted within easy reach of the hand (if manual
control) or the mouth (if speech input). These prototypes are typical of much contemporary work on
wearable computers in terms of the relative size
and placement of components.
Wearable computers represent a load on the
person and consequently can affect the physical
activity of the person (Zingale et al., 2005). Legg
(1985) proposes that the human body can carry
loads at the following points: head, shoulder, back,
chest, trunk, upper arm, forearm, hands, thighs,
feet, a combination of these, and aided (i.e., by pulling, pushing or sharing a load). A glance through
the literature of wearable computers shows that
most of these sites have been experimented with
in various designs. Gemperle et al. (1998) offer the
term wearability to describe the use of the human
body to physically support a given product and, by
extension, the term dynamic wearability to address
the device being worn while the body is in motion.
Given this notion of wearability, there is the question

Figure 2. Wearing prototype wearable computers

1137

Human Factors Problems of Wearable Computers

of where a device might be positioned on the body,

and, once positioned, how it might affect the wearer
in terms of balance, posture and musculoskeletal
loading Given the notion of dynamic wearability,
there are the questions of how the device will be
carried or worn, how this might affect movement,
and how this might lead to either perceptions of
differences in movement patterns, physiological
strain or psychological stress on the wearer. These
changes can be assessed, using subjective self-report
techniques (Bodine & Gemperle, 2003; Knight et
al., 2002, 2006; Knight & Baber, 2005) and through
objective analysis (Nigg & Herzog, 1994).

Energy Expenditure, Muscle Activity

and Using Wearable Computers
An attached load to the body will have a direct
affect on the energy expended by the body, as the
muscles burn more energy to generate force to
counteract the weight of the load. This situation
sees a wearer of a wearable computer potentially
increasing their overall energy expenditure to
overcome inertial changes. Thus, any additional
weight on the body can create demands on the
musculoskeletal system to support and move the
object. In addition, the kinematics of posture and
movement can compound the loading effects of
a wearable computer. Knight and Baber (2004)
report the following about head-mounted loads,
such as head mounted displays: (1) different head
postures have a measurable effect on musculoskeletal loading; (2) the heavier the frontal load
(as is representative of the loading of a HMD),
the greater the muscle activity required to keep
the head in a fixed position and; (3) in neutral or
extended postures, the wearer can sustain larger
loads than if the head is in a flexed or rotated
head position Thus, it is not simply a matter of
reducing load of head-mounted equipment, but
one must determine the posture that the wearer of
such equipment is likely to adopt. An alternative
location for display technology is on the forearm.
In their study of pointing devices, Thomas et al.

1138

(1998) demonstrate that for general purpose activity, the forearm would be an appropriate place to
mount the device. This location is attractive as it
allows the wearer to easily move the display into
the field of vision. However, Knight and Baber
(2007) have questioned this location or at least
raised concerns. Recording shoulder and upper
arm muscle activity and measuring perceptions
of exertion while participants interacted with
arm mounted computers of different weights they
found that the mere act of holding the arm in an appropriate posture to interact with an arm mounted
computer was sufficient to exceed recommended
levels of muscle activity for sustained activity .
In addition it induced symptoms of fatigue after
only two minutes where the addition of weight in
the form of the mounted technology compounded
this physical effect.

Reducing Size and Separating

Components
The discussion so far has suggested ways in which
weight and placement of loads on the body can lead
to problems The implication is that most wearables
take the form factor of the brick-on-the-back (as
illustrated by Figure 2). It is possible to reduce
the size of components further by removing the
interaction and display components. Thus, the
Bluetooth headset for a mobile phone effectively
reduces the user interface (for talking on the phone)
to a unit that clips on the ear. Various MP3 players
can be worn on the upper arm while jogging (see
comment about usability). Alternatively, there has
been much interest in the use of active badges
that signal a persons location to a network (Want
et al., 1992). These devices can be made small
enough to be worn as badges or incorporated into
items of clothing, for example Schmidt et al. (1999)
report a device, mounted in the wearers tie, which
detects changes in ambient sound levels or wearer
movement. In these examples, the on-body part
of the computer system is basically a sensor and/or
transmitter and/or receiver that link the wearer

Human Factors Problems of Wearable Computers

to a network of other technologies. BodyMedia

have developed a device that places sensors on
the persons upper arm in order to record data
relating to everyday activity (in much the same
way that an MP3 player is a data display device)
and needs to be connected to a computer in order
to process the tunes on it (see Figure 3). However,
it makes sense to ask whether reducing the size
of the devices will reduce or even eliminate these
problems. Physical discomfort may arise from
pressure of the device on the body. The pinnar
is very sensitive and many users complain of
discomfort around the ear from using ear plugs
and ill-fitting head phones. HMDs place pressure around the forehead and crown. Even the
small displays produced by Microoptical (see
Figure 4) can cause discomfort around the ears
and nose; especially given that the glasses they
come with are hard plastic, not specifically fitted
for the individual wearer. Items attached around
the arm, specifically during physical activity
(i.e., for jogging), have to be attached tightly so
that they do not bounce against the arm or slip
down, as such they may result in discomfort and
result in vasoconstriction leading to sensations
of numbness and tingling in the lower arm and
hand, not to mention that the display is not easily

Figure 3. SenseWear device from BodyMedia3

viewable when the device is in position which

raises usability issues.

Perceptual Impacts of Wearing

Computers
A wearable computer can be considered a perceptual prosthesis in the ways that it can provide
additional information to, or enhance perceptual
capability of, the wearer. Information provision
can simply mean allowing the wearer access to
information that is not currently present in the
world, for example through a visual or audio
display. The underlying concept in many of these
applications is that the wearers perception of the
world can be augmented or enhanced (Feiner et
al., 1998). Before considering this argument, we
can consider the MP3 player as a device that can
present media that is added to the world (in the
form of a personal soundtrack to the persons
activity). To some extent this could be considered as form of augmenting perception. What
makes augmented reality different from simply
displaying information is that (in most systems)
the information is presented in accordance with
the context in which the person is behaving. Such
information could be in the form of an opaque
visual display or could be visual information
overlaid onto the world. In basic systems, the
information can be called up by the wearer, or
pushed from another source, for example a radio
or telephone link. In more sophisticated systems,
the information is presented on the basis of the
computers interpretation of context.
While augmented reality displays could be
beneficial, there are potential problems associated
with the merging of one source of information (the
computer display) with another (the world). Contemporary wearable computers tend to combine
monocular head-mounted displays (see Figure 4)
with some form of interaction device, for example
keyboard, pointing device or speech. NRC (1997)
point out that monocular head-mounted displays
could suffer from problems of binocular rivalry,

1139

Human Factors Problems of Wearable Computers

Figure 4. Monocular head-mounted displays can

restrict field of view

might impair performance. Apache helicopter

pilots currently wear monocular, head-mounted
displays and a review of 37 accidents concluded
that 28 of these accidents could be attributed to
wearing of the display (Rash et al., 1990).

Computer Response to Physical

Activity

that is, information presented to one eye competes for attention with information presented
to the other, which results in one information
source becoming dominant and for vision to be
directed to that source. A consequence of this
phenomenon is that the wearer of a monocular
display might find it difficult to share attention
between information presented to the display
eye and information seen through the free eye.
This problem is compounded by the field of view
of such displays.
The field of view for the commercial monocular
HMDs ranges from 16-60, which is considerably less than that of normal vision (around 170
for each eye, Marieb 1992). Narrow field of view
can degrade performance on spatial tasks such
as navigation, object manipulation, spatial awareness, and visual search tasks. Restrictions on field
of view will tend to disrupt eye-head coordination
and to affect perception of size and space (Alfano
& Michel, 1990). One implication of a restricted
field of view is that the wearer of a see-through
HMD will need to engage in a significant amount
of head movement in order to scan the environment (McKnight & McKnight, 1993). Seagull and
Gopher (1997) showed longer time-on-task in a
flight simulator when using a head down visual
display unit than with a head-mounted, monocular
display. Thus, it appears that a monocular display

1140

While it is possible that wearing technology affects user performance, it is also possible that
the physical activity of the person can affect the
computer, for example through the use of sensors
to recognize actions and use this recognition to
respond appropriately.
Given the range of sensors that can be attached
to wearable computers, there has been much interest in using data from these sensors to define and
recognize human activity. This has included work
using accelerometers (Amft et al., 2005; Junker
et al., 2004; Knight et al., 2007; Ling & Intille,
2004, Van Laerhoven & Gellersen, 2004; Westeyn
et al., 2003) or tracking of the hand (Ogris et al.,
2005). The approach is to collect data on defined
movements to train the recognition systems, and
then use these models to interpret user activity.
At a much simpler level, it is possible to define
thresholds to indicate particular postures, such as
sitting or standing, and then to use the postures
to manage information delivery (Bristow et al.,
2004). The use of data from sensors to recognize
human activity represents the merging of the
research domains of wearable computers with
that of pervasive computing, and implies the
recognition not only of the actions a person is
performing but also the objects with which they
are interacting (Philipose, et al., 2004; Schwirtz
& Baber, 2006).
An area of current interest for the wearable or
ubiquitous computing communities is the interpretation of human movement related to maintenance
and related activity. The artifacts with which the
person is interacting could be instrumented. For
example, a simple approach is to fit switches on

Human Factors Problems of Wearable Computers

components (in a self-assembly furniture pack)

and for the user to depress the switches when each
components is handled (Antifakos et al., 2002).
A less intrusive approach would be to fit radio
frequency identification person is using (Scwhirtz
et al., 2006) and to use the activation of these tags
to infer user activity. This requires that either the
components or the tools be adapted to the task.
The decreasing costs of RfiD suggest that, within
a few years, tags will be universally used in a
wide range of consumer products. Alternatively,
the activity of the user could be taken to develop
predictive models in order to infer the activity
that is being performed. For some activities it
might be sufficient to use data from very generic
sensors, such as microphones, to collect data to
define actions (Ward et al., 2006), while in other
it might be necessary to rely on more specific sensors, such as the use of accelerometers to define
movements that are characteristic of assembly
and maintenance tasks (Schwirtz & Baber, 2006;
Westyn et al., 2003). Steifmeier et al. (2006) show
how tracking the motion of the hand (e.g., using
ultrasonic tracking and inertial sensors) can be
used to define specific types of movement that
relate to maintenance tasks, such as spinning a
wheel or rotating the pedals, unscrewing a cap
or using a bicycle pump. In a related domain,
there is some interesting work on the collection
of activity data relating to nursing, for example
through a combine infra-red proximity sensing
and accelerometers (Noma et al., 2004).

USING WEARABLE COMPUTERS

Given that a wearable computer provides a means
of anytime, anywhere access to information,
most forms of cognitive activity have been mooted
as possible areas that can be supported. In this
section, we consider three areas: (i.) supporting
memory; (ii.) supporting navigation; and (iii.)
information search and retrieval.

Supporting Memory
A common example used to illustrate the benefits
of a wearable computer is what can be termed
the context-aware memory. Imagine you are
attending a conference (or any large social gathering) and having to remember someones name.
Systems using some form of badge or face-recognition have been proposed to help with such
situations; the computer would register the person
and provide you with name and some additional
details about the person, for example when you
last met, what research interests are listed on
their web-page, where they work, and so forth.
There has been little research on whether and how
these systems improve memory, and this example
points to a possible confusion between supporting processes involved in recalling information
from memory and the provision of contextuallyrelevant information. An associated question is
whether wearable computers (particularly having
a head-mounted display) positioned on the eye
can have an impact on recall. The analogy is
with the tourist watching a parade through the
lens of a video cameradoes the act of recording something weaken the ability to process and
recall information? Baber et al. (2001) use a search
task coupled with surprise recall to show that,
in comparison with not using any technology,
participants using a digital camera and wearable
computer conditions showed lower performance,
and that overall the wearable computer showed
the biggest impairment in recall. There are many
reasons why interruption at initial encoding can
limit the ability to remember something, and the
question is whether the head-mounted display
serves to interrupt encoding; either due to distraction (with a host of information appearing on the
screen), or through limitations of field of view, or
for some other reasons.

1141

Human Factors Problems of Wearable Computers

Navigation and Way-Finding

Wayfinding requires people to travel through the
world in order to reach specific locations. Thus,
there is a need to manage both the act of traveling
(for wearable computers this usually consists of
walking) and relating a view of the world to the
defined location. Sampson (1993) investigated
the use of monocular, head-mounted displays for
use when walking. Participants were presented
with either spatial or alphanumeric information
and required to traverse paths with or without
obstacles. In general, participants performed
equally well when standing or when traversing
paths without obstacle, but were significantly
worse when obstacles were present. Thus, the
need to maintain visual attention on display and
environment can be seen to impair performance
which suggests that the amount and type of information which can usefully be presented on such
displays needs to be very limited. As global positioning systems (GPS) become smaller, cheaper,
and more accurate, there has been an increase in
their application to wearable computers. Seager
and Stanton (2004) found faster performance with
a paper map than GPS, that is routes completed
faster. This was due, in part, with the increase
in time spent looking at the GPS display and the
number of updates made on the digital view (pre-

sumably because the GPS was moving a marker

along the map to show the participants location).
Participants also were less likely to orientate the
digital view in the direction of travel (possibly
because they might have assumed that the display
would orient to their direction of travel rather than
North up). Studies into the effect of perspective
views, that is aligning a map with direction of
travel, have not shown significant performance
advantage over 2D views to date (Suomela et al.,
2003), although this approach does seem beneficial
in systems that support wayfinding in moving
vehicles (Aretz, 1991). Systems that overlay routes
onto the head-mounted display (Figure 5) could
also assist in simple wayfinding tasks.
The use of visual support for navigation requires the user to divide attention between the
environment and a visual display. In terms of the
background or foreground of activity (discussed
by Hinckley et al., 2005), this essentially places
all the tasks in the users foreground. It is possible
that navigation prompts could be provided using
auditory cues. In simple terms, the heading could
be indicated by varying the parameters of simple
beeps, for example changing in pitch or in intensity as the person deviates from a path (rather like
auditory glide-slope indicators in aircraft). More
recent developments have replaced the simple
beeps with music. In this way, a more subtle

Figure 5. Displaying a route overlay on a head-mounted display

1142

Human Factors Problems of Wearable Computers

(background) form of cueing can be achieved to

useful effect. The music could be manipulated
to vary quality with deviation from a path, for
example through distortion (Strachan et al., 2007,
2005), or through modifying the panning of music
in stereo presentation (Warren et al., 2005).

Finding and Retrieving Information

Having the ability to access information as you
need it is a core concept of wearable computer
research. Early examples had the user enter queries
and view information on a monocular display. This
approach of having the user ask for information
has been superseded by having the computer push
information, on the basis of its interpretation of
context. In such examples, the benefit of wearing
the computer comes from its permanent presence
and state of readiness, allowing access either of
data stored on the computer or via the World Wide
Web. Rhodes and Starner (1997) describe the remembrance agent, which monitors the information that a user types into a computer and makes
associations between this information and data it
has stored. Obviously this creates an overhead on
the user, in terms of the need to type information
into the computer. However, it is only a relatively
small modification to replace the typed entry with
speech recognition (see Pham et al., 2005). While
the agents in either of these examples can run
on desktop computers, it is the fact that they are
continuously running on the computer worn by
the user that makes them interesting. The role of
the wearable computer in these examples is to
discretely run searches in the background and alert
the wearer to interesting links and associations
between the current topic of conversation (or typing) and information to which the computer has
access. It might also be useful for the computer
to track user activity and then to either record
patterns of activity (in order to refine its model
of context) or to offer information relevant to the
user. Thus, a very common application domain
for context-aware, wearable computer research is

the museum visitor (Sarini & Strapparava, 1998).

The idea is that when the visitor stands at a certain
location, say in front of a painting, the computer
offers information relating to the objects in that
location. The manner in which the location (and
other contextual factors) is used to select information and the manner in which the information is
presented to the user might vary across applications. The examples considered thus far present
the wearable computer as a device that is able
to push potentially relevant information to the
wearer. However, rather than simply overloading
the wearer with information, the role of the agents
or context-awareness is to manage and tailor the
information to the person.

Impact on User Performance

While there are many applications reported in the
literature, there is a surprising lack of research
into how effective these applications are in improving or otherwise changing user performance.
One domain in which wearable computers have
received both interest and support has been in
maintenance (Mizell, 2003). A study exploring
the use of head-mounted displays to support
maintenance work show that performance could
be improved, providing information was displayed
in an appropriate format (Kancler et al., 1998).
However, other studies have been more equivocal.
Tasks requiring participants to follow instructions
on a wearable computer or printed on paper have
shown both the wearable computer (Baber et al.,
1998) and the paper (Baber et al., 1999c; Ockerman et al., 1997; Siegel & Bauer, 1997) to lead to
superior performance. One explanation of these
differences lies in the design of the information
that was presented, for example the manner in
which information is presented can impact on
overall performance times (Baber et al., 1999b;
Sampson et al., 1993). One implication of these
studies is that participants using the wearable
computer tend to follow the same sequence of
tests (as defined on the visual display), whereas

1143

Human Factors Problems of Wearable Computers

the participants in the paper condition would order the tests as they saw fit (Baber et al., 1999b,
Ockerman & Pritchett, 1998).
In terms of interacting with wearable computers and the appropriate devices to use, there has
been very little work to date. While one might
assume that the optimal interaction techniques
would be ones that support hands-free interaction, such as speech recognition, studies suggest
that walking has a negative impact on speech
recognition performance (Oviatt, 2000; Price et
al., 2004). In terms of entering data, Thomas et al.
(1997) showed that a forearm mounted QWERTY
keyboard led to superior performance over a
five-button chording device or a virtual keyboard
controlled using an isometric button. However, one
might question the recommendation of a forearm
mounted device, based on consideration of musculoskeletal strain. In terms of selecting objects
on a display, Thomas et al. (1998) found that a
touchpad mounted on the forearm was preferred
by users, but that one mounted on the thigh lead
to superior performance when sitting, kneeling or
standing. Thus, the mounting of a pointing device
can have a bearing on performance (although one
might question whether pointing is an appropriate
means of performing selection tasks on a wearable computer). Zucco et al. (2006) considered the
performance of drag and drop tasks while stationary and whilst walking, using different devices.
They found that a gyroscopic mouse lead to best
performance while stationary, but that touchpad
or trackball were lead to better performance when

the user was walking (and that all devices were

superior to the Twiddler keypad).
Rather than seeing these studies as recommendations for specific interaction devices, I feel
that they illustrate that the relationship between
the activity that a person is performing and the
demands of the ongoing activity in the environment interact in ways that require careful planning in the design of wearable computers. This
brief discussion raises questions on one might
develop standards for the deployment of wearable
computers in these (and related) applications, and
also what level of performance improvement one
might expect from this technology.

SMART CLOTHING AND TEXTILES

It is worth mentioning the efforts to incorporate
at least some aspects of technology into clothing
and textiles (Berzowska, 2005). This could then
take the form of clothing that has been adapted
to incorporate some of the components (see
Figure 6).
Alternatively, the textiles can exhibit some
form of smartness. Often this refers to Textiles
that are able to sense stimuli from the environment, to react to them and adapt to them... (Van
Langehoven & Hertleer, 2004). Taking these
terms separately, we can ask how could textiles
sense, react and adapt. In terms of sensing, there
is a wide range of possible approaches, including thermal, chemical, mechanical, as well as

Figure 6. Incorporating consumer electronics into clothingthe Philips-Levi Strauss jacket4

1144

Human Factors Problems of Wearable Computers

biosensors. Buechley (2006) shows how simple

off-the-shelf sensors and actuators can be incorporated into items of clothing. In particular, this
work, following the earlier work of Post and Orth
(1997) demonstrates how fabrics can be knitted
or woven to provide some electrical conductivity,
and then fitted with components. In contrast, other
researchers report ways in which it is possible to
use to electrical properties of prepared fabrics,
such as change in resistance, inductance or capacitance (Wijesiriwardana et al., 2003, 2004), to
incorporate specific sensing capabilities, such as
stretching (Farringdon et al., 1999; Huang et al.,
2006) or impact (Lind et al., 1997). What has yet
to be fully realized from this smart textile work
is that manner in which the user would interact
with the computing aspects in order to entry
data or perceive displayed information. Thus,
much of the research is still concerned with the
development of the bus onto which sensors,
processors, displays, batteries, and so forth can
be mounted.

EMOTIONAL IMPACT OF WEARING

COMPUTERS
There are three broad categories of impact that will
be considered in this section. The first concerns
the emotional response to wearable computers by
the people wearing these devices and the people
with whom they interact. The second concerns
the response of the computer to the emotions of
the wearer, that is affective computing. The third
concerns the manner in which collaboration can
be supported within a group of wearable computer
wearers.
While head-mounted displays have been used
for many years in the domain of military aviation, they have yet to find widespread use on the
ground. Most cinema goers will be familiar with
concepts of wearable computers from science
fiction films and might feel uncomfortable with
cyborgs in their midst. This discomfort could be

particularly acute in stressful situations (as one

of the paramedics in an earlier study pointed out
to us, If youre just coming round from a heart
attack, the last thing youd expect to see if some
robot headed bloke trying to take your pulse).
As there are so few commercial applications of
wearable computers, these devices still represent
something of a novelty and there has been very
little research into how people might respond to
people wearing such devices. One collection of
anecdotal evidence can be found in Thad Starners
Ph.D. Starner completed his Ph.D. at MIT, and
with Brad Rhodes and Steve Mann developed
a variety of wearable computers that they wore
for prolonged periods of time. This meant that
they would encounter members of the public on
a regular basis and Starner recorded some of the
reactions. A common response was to assume that
the wearable computer was simply some form of
familiar technology, perhaps a very sophisticated
video camera or an advanced form of video game
console or a medical device. The implication
of this is that people might try to explain the
technology using a familiar mental model but,
as Starner points out, people have yet developed
a mental model of a wearable computer on the
basis of familiarity and exposure. While this is
partly a matter of the unfamiliar appearance of
the wearer it is also related to the manner in which
the wearer interacts with the device. For example,
Starner points out that when you ask someone the
time they will raise their wrist in order to consult
their watch, but with a head-mounted display one
might simply glance up at the screen. This means
that the physical behavioral cues might be far less
obvious. This is, of course, similar to the way in
which Bluetooth headsets allow people to speak
on their mobile telephone in a manner that makes
it looks as if they are speaking to themselves
(indeed, some users of Bluetooth headsets make
gestures that look as if they are holding a phone,
for example holding the microphone or holding
their hands near their faces). There will be a
trade-off between the familiarity of wearable

1145

Human Factors Problems of Wearable Computers

computers, their appearance, and the ability of

people to explain their use. However, there is an
additional factor at play here (which is also hinted
at by the Bluetooth headset discussion), and that
is that the wearable computer provides information to an individual in a highly individualized
mannerit is not possible for other people to
see what the wearer is looking at or guess what
they are doing. An analogy can be drawn in this
instance with the person on a train reading a
humorous book at laughing out loudthe other
people feel uncomfortable because they can not
read the what the person is reading. The removal
of the observer from the source of information can
be disconcerting and can lead to such comments
as (from Starner) we cant tell if youre talking
about us behind our backs or when you wear
your display, how can I tell if you are paying attention to me or reading your e-mail?
One implication of the physical appearance and
individualized interaction of wearable computers
is the sense that people who wear such technology
are different from normal people. As Sheridan et
al. (2000) note, people wearing computers could be
perceived (by themselves or by the people around
them) as a cyborg community that is different
from other people. With the use of communication and networking capabilities, it is possible for
a group of wearable computer users to be able
to share information and maintain contact as a
community (Wellman, 2001). However, we should
be careful to distinguish between the ability to
maintain contact with other people (which one
can do easily with mobile telephones, even to the
extent of setting up talk groups) and the ability to
share the wide range of information that wearable
computing can support. This could simply mean
the sharing of the same documents or video, but
could also allow new forms of collaboration,
sharing and exchange of information.
Context could be defined by changes in the
physiological state of the wearer (Picard, 1997).
This requires a more intimate means of recording
data from the wearer, perhaps through monitoring

1146

of pulse or heart activity. The use of context to

initiate image capture has been demonstrated
by several projects; most notably in Healey and
Picards (1998) StartleCam, in which changes
in galvanic skin response (GSR) was used to
trigger image capture. There has been surprisingly little attempt at extending this work in the
years since it was reported, although over uses of
context in image capture have explored the use
of ambient sound which is captured at the same
time as the image (e.g., Frolich & Tallyn, 1999;
Ljunblad et al., 2004). Bristow et al. (2005) used a
set of context identifiers to take still images when
context changed, and showed that these were
surprisingly consistent with photographs taken by
humans. In addition to the computer sensing the
physiological responses of the wearer (and hence
drawing some inference as to the affective state),
it is also possible to infer the state of people with
whom the wearer is interacting in order to develop
emotionally intelligent interfaces. By monitoring
the changing facial expressions of ones conversational partner, it might be possible to provide
support for people who find it difficult to judge
the emotional response, for example people with
Autism (El Kaliouby & Robinson, 2003).

DISCUSSION
Wearable computers continue to raise many significant challenges for human factors research.
These challenges involve not only cognitive
aspects of presenting information but also perceptual aspects of displaying the information against
the backdrop of the everyday environment and
physical aspects of mounting the devices on the
person. This chapter has overviewed some of
the developments in the field and offered some
consideration of how these human factors can be
considered. While the field is largely motivated
by technological advances there is a need to carefully ground the developments in the physical and
cognitive characteristics of the humans who are
intended to wear them.

Human Factors Problems of Wearable Computers

REFERENCES
Alfano, P.L., & Michel, G.F.(1990). Restricting the
field of view: Perceptual and performance effects.
Perceptual and Motor Skills, 70, 35-45.
Amft, O., Junker, H., & Troster, G. (2005). Detection of eating and drinking arm gestures using
inertial body-worn sensors, Ninth International
Symposium on Wearable Computers ISWC 2005
(pp.160-163). Los Alamitos, CA: IEEE Computer
Society.
Antifakos, S., Michahellis, F., & Schiele, B. (2002).
Proactive instructions for furniture assembly.
UbiComp 2002 4th International Conference on
Ubiquitous Computing (pp. 351-360). Berlin:
Springer.

Baber, C., Sutherland, A., Cross, J., & Woolley,

S. (2001). A wearable surveillance system: Implications for human memory and performance.
In D. Harris (Ed.), Engineering psychology and
cognitive ergonomics industrial ergonomics,
HCI, and applied cognitive psychology, 6, (pp.
285-292). Aldershot: Ashgate.
Bainbridge, L. (1987) Ironies of automation. In
J. Rasmussen, J. Duncan & J. Leplat (Eds.), New
technology and human errors (pp. 271-286). New
York: Wiley
Bass, L. (1996). Is there a wearable computer in
your future? In L.J. Bass and C. Unger (Eds.),
Human computer interaction (pp. 1-16). London:
Chapman and Hall.

Aretz, A.J. (1991). The design of electronic map

displays. Human Factors, 33(1), 85-101.

Bass, L., Siewiorek, D., Smailagic, A,. & Stivoric,

J. (1995). On site wearable computer system.
CHI95 (pp. 83-84). New York: ACM.

Baber, C. (1997). Beyond the desktop: designing

and using interaction devices. San Diego, CA:
Academic Press.

Berzowska, J. (2005). Electronic textiles: wearable

computers, reactive fashions and soft computation. Textile, 3, 2-19.

Baber, C., Haniff, D.J., Knight, J., Cooper, L., &

Mellor, B.A. (1998). Preliminary investigations
into the use of wearable computers. In R. Winder
(Ed.), People and computers XIII (pp. 313-326).
London: Springer-Verlag.

Bodine, K., & Gemperle, F. (2003). Effects of

functionality on perceived comfort of wearables,
Digest of Papers of the 7th International Symposium on Wearable Computers, (pp. 57-61). Los
Alamitos, CA: IEEE Computer Society.

Baber, C., Haniff, D.J., & Woolley, S.I. (1999a).

Contrasting paradigms for the evelopment of
wearable computers. IBM Systems Journal, 38(4),
551-565.

Buechley, L. (2006) A construction kit for electronic textiles. In Digest of Papers of the 10th
International Symposium on Wearable Computers
(pp. 83-90). Los Alamitos, CA: IEEE Computer
Society.

Baber, C., Knight, J., Haniff, D., & Cooper, L.

(1999b). Ergonomics of wearable Computers.
Mobile Networks and Applications, 4, 15-21.
Baber, C., Arvanitis, T.N., Haniff, D.J., & Buckley,
R. (1999c). A wearable computer for paramedics: studies in model-based, user-centred and
industrial design, In M.A. Sasse and C. Johnson
(Eds.), Interact99 (pp. 126-132). Amsterdam:
IOS Press.

Bristow, H., Baber, C., Cross, J., Knight, J., &

Woolley, S. (2004). Defining and evaluating context for wearable computers International Journal
of Human Computer Studies, 60, 798-819.
Clark, A. (2006). Natural born cyborgs? Edge.
Retrieved from www.edge.org/3rd_culture/clark/
clark_index.html

1147

Human Factors Problems of Wearable Computers

El Kaliouby, R., & Robinson, P. (2003). The emotional hearing aidAn assistive tool for autism,
HCI International
Farringdon, J., Moore, A.J., Tilbury, N., Church,
J., & Biemond, P.D. (1999). Wearable sensor badge
and sensor jacket for contextual awareness. Digest
of Papers of the 3rd International Symposium on
Wearable Computers (pp. 107-113). Los Alamitos,
CA: IEEE Computer Society.
Feiner, S., MacIntyre, B., Hollerer, T., & Webster,
A. (1997). A touring machine: Prototyping 3D
mobile augmented reality systems for exploring the urban environment. Digest of Papers
of the 1st International Symposium on Wearable
Computers, (pp. 74-81). Los Alamitos, CA: IEEE
Computer Society.
Frolich, D., & Tallyn, E. (1999). AudioPhotography: Practice and prospects. CHI99 Extended
Abstracts (pp. 296-297). New York: ACM.
Gemperle, F., Kasabach, C., Stivoric, J., Bauer,
M., & Martin, R. (1998). Design for Wearability.
Digest of Papers of the 2nd International Symposium on Wearable Computers (pp. 116-123). Los
Alamitos, CA: IEEE Computer Society.
Healey, J., & Picard, R.W. (1998). StartleCam: a
cybernetic wearable computer, Proceedings of
the 2nd International Symposium on Wearable
Computers (pp. 42-49). LosAlamitos, CA: IEEE
Computer Society.
Hinckley, K., Pierce, J., Horvitz, E., & Sinclair, M.
(2005). Foreground and background interaction
with sensor-enhanced mobile devices. ACM Trans.
Computer-Human Interaction, 12(1), 31-52.
Huang, C-T., Tang, C-F., & Shen, C-L. (2006).
A wearable textile for monitoring respiration,
using a yarn-based sensor. Digest of Papers of
the 10th International Symposium on Wearable
Computers, Los Alamitos, CA: IEEE Computer
Society, 141-142

1148

Junker, H., Lukowicz, P., & Troster, G. (2004).

Continuous recognition of arm activities with
body-worn inertial sensors. Eighth International
Symposium on Wearable Computers ISWC04
(pp.188-189). Los Alamitos, CA: IEEE Computer
Society.
Kancler, D.E., Quill, L.L., Revels, A.R., Webb,
R.R., & Masquelier, B.L. (1998). Reducing cannon plug connector pin selection time and errors
through enhanced data presentation methods.
Proceedings of the Human Factors and Ergonomics Society 42nd Annual Meeting (pp. 12831290). Santa Monica, CA: Human Factors and
Ergonomics Society.
Knight, J. F., Baber, C., Schwirtz, A., & Bristow,
H. W. (2002). The comfort assessment of wearable
computers. The Sixth International Symposium of
Wearable Computers (pp. 65-72). Los Alamitos,
CA: IEEE Computer Society.
Knight J.F., & Baber, C. (2004). Neck muscle
activity and perceived pain and discomfort due
to variations of head load and posture. Aviation,
Space and Environmental Medicine, 75, 123131.
Knight, J. F., & Baber, C. (2005). A tool to assess
the comfort of wearable Computers. Human Factors, 47(1), 77-91.
Knight, J. F., & Baber, C. (2007). Assessing the
physical loading of wearable computers. Applied
Ergonomics, 38, 237-247.
Knight, J.F., Williams, D.D., Arvanitis, T.N.,
Baber, C., Sotiriou, S., Anastopoulou, S. et al.
(2006). Assessing the wearability of wearable
computers. Digest of Papers of the 10th International Symposium on Wearable Computers
(pp. 75-82). Los Alamitos, CA: IEEE Computer
Society.
Knight, J., Bristow, H.W., Anastopoulou, S., Baber,
C., Schwirtz, A., &Arvanitis, T.N. (2007). Uses
of accelerometer data collected from a wearable

Human Factors Problems of Wearable Computers

system. Personal and Ubiquitous Computing

11(2), 117-132.

damentals of wearable computers and augmented

reality (pp. 447-469). Mahwah, NJ: LEA.

Legg, S. J. (1985). Comparison of different

methods of load carriage. Ergonomics, 28(1),
197-212.

Naya, F., Ohmura, R., Takayanagi, F., Noma, H.,

&
Kogure, K. (2006). Workers routine activity
recognition using body movement and location
information. Digest of Papers of the 10th International Symposium on Wearable Computers (pp.
105-108). Los Alamitos, CA: IEEE Computer
Society.

Lin, R., & Kreifeldt, J. G. (2001) Ergonomics in

wearable computer design. International Journal
of Industrial Ergonomics, 27(4), 259-269.
Lind, E.J., Jayaraman, S., Park, S., Rajamanickam,
R., Eisler, R., Burghart, G., et

al. (1997). A sensate

liner for personnel monitoring applications. Digest
of Papers of the 1st International Symposium on
Wearable Computers (pp. 98-107). Los Alamitos,
CA: IEEE Computer Society.
Ling, B., & Intille, S. S. (2004). Activity recognition from user-annotated acceleration data.
Pervasive 2004 (pp. 117). Berlin: Springer.
Ljungblad, S., Hakansson, M., Gaye, L &
Holmquist, L.E. (2004). Context photography:
Modifying the digital camera into a new creative
tool, CHI04 Short Papers, (pp. 1191-1194). New
York: ACM.
Mann, S. (1997). An historical account of
WearComp and WearCam inventions developed for applications in Personal Imaging. In
The First International Symposium on Wearable
Computers (pp. 66-73). Los Alamitos, CA: IEEE
Computer Society.
Marieb, E. N. (1992). Human anatomy and
physiology [2nd edition]. Redwood City, CA:
Benjamin/Cummings.
McKnight, A.J., & McKnight, A.S. (1993). The
effects of motor cycle helmets upon seeing and
hearing. In Proceedings of the 37th Annual conferences of the Association for the Advancement
of Automotive Medicine (pp. 87-89). Des Plaines,
Ill.
Mizell, D. (2001). Boeings wire bundle assembly
project. In W. Barfield & T. Caudell (Eds.), Fun-

Nigg, B.M., & Herzog, W. (1994). Biomechanics

of the musculoskeletal system. England: Wiley
NRC. (1997). Tactical display for soldiers. Washington, DC: National Academy.
Noma, H., Ohmura, A., Kuwhara, N., & Kogure,
K. (2004). Wearable sensors for auto-event
recording on medical nursingUser study of
ergonomic requirements. Digest of Papers of
the 8th International Symposium on Wearable
Computers (pp. 8-15). Los Alamitos, CA: IEEE
Computer Society.
Ockerman, J.J., & Pritchett, A.R. (1998). Preliminary investigation of wearable computers for task
guidance in aircraft inspection. Digest of Papers
of the 2nd International Symposium on Wearable
Computers (pp. 33-40). Los Alamitos, CA: IEEE
Computer Society.
Ogris, G., Stiefmeier, T., Junker, H., Lukowics,
P., & Troster, G. (2005). Using ultrasonic hand
tracking to augment motion analysis based recognition of manipulative gestures. ISWC 2005
(pp. 152-159). Los Alamitos, CA: IEEE Computer
Society.
Oviatt, S. (2000). Taming recognition errors with
a multimodal interface. Communications of the
ACM, 43, 45-51.
Patterson, D. (2005). Fine grained activity recognition by aggregating abstract object usage. Digest
of Papers of the 9th International Symposium on
Wearable Computers (pp. 44-51). Los Alamitos,
CA: IEEE Computer Society.

1149

Human Factors Problems of Wearable Computers

Pham, N.V., Terada, T., Tsukamato, M., & Nishio,

S. (2005). An information retrieval system for
supporting casual conversation in wearable
computing environments. Proceedings of the Fifth

International Workshop on Smart Appliances and

Wearable Computing (pp. 477 483). Washington,
DC: IEEE Computer Society.

Philipose, M., Fishkin, K. P., Perkowitz, M., Patterson, D. J., Fox, D., Kautz, H., & Hahnel, D.
(2004). Inferring activities from interactions with
objects. Pervasive Computing 2004, 5057.
Picard, R. (1997). Affective computing. Cambridge, MA: MIT Press.
Post, E.R. & Orth, M. (1997). Smart fabric, or
wearable clothing. Digest of Papers of the 1st
International Symposium on Wearable Computers
(pp. 167-168). Los Alamitos, CA: IEEE Computer
Society.
PriceK. J., Min L., Jinjuan, F., Goldman, R.,
Sears, A., & Jacko, J.A. (2004). Data entry on the
move: An examination of nomadic speech-based
text entry. User-centered interaction paradigms
for universal access in the information society
(pp.
460-471).
Berlin: Springer.
Rash, C.E., Verona, R.W., & Crowley, J.S. (1990).
Human factors and safety considerations of night
vision systems flight using thermal imaging systems. Proceedings of SPIEThe International
Society for Optical Engineering, 1290, 142-164.
Rhodes, B.J., & Starner, T. (1997). Remembrance
agent: A continuously running automated information retrieval system. The Proceedings of The
First International Conference on The Practical
Application Of Intelligent Agents and Multi Agent
Technology (PAAM 96), pp. 487-495.
Rohaly, A. M., & Karsh, R. (1999). Helmetmounted displays. In J. M. Noyes & M. Cook
(Eds.), Interface technology: The leading edge (pp.
267-280). Baldock: Research Studies Press Ltd.

1150

Sampson, J. B. (1993). Cognitive performance of

individuals using head-mounted displays while
walking. In Proceedings of the Human Factors
and Ergonomics Society 37th Annual Meeting (pp.
338-342). Santa Monica, CA: Human Factors and
Ergonomics Society.
Sarini, M., & Strapparava, C. (1998, June 20-24).
Building a user model for a museum exploration
and information-providing adaptive system. Proceedings of the Workshop on Adaptive Hypertext
and Hypermedia. Pittsburgh, PA.
Sawhney, N., & Schmandt, C. (1998). Speaking
and listening on the run: Design for wearable
audio computing. Digest of Papers of the 2nd International Symposium on Wearable Computers
(pp. 108-115). Los Alamitos, CA: IEEE Computer
Society.
Schmidt, A., Gellerson, H-W., & Beigl, M. (1999).
A wearable context-awareness component: Finally
a good reason to wear a tie. Digest of Papers of
the 3rd International Symposium on Wearable
Computers (pp. 176-177). Los Alamitos, CA:
IEEComputer Society.
Schwirtz, A., & Baber, C. (2006). Smart tools
for smart maintenace. IEE and MoD HFI-DTC
Symposium on People in Systems (pp. 145-153).
London: IEE.
Seager, W., & Stanton, D. (2004, April). Navigation in the city. Ubiconf 2004, Gresham College,
London.
Seagull, F. J., & Gopher, D. (1997). Training head
movement in visual scanning: An embedded approach to the development of piloting skills with
helmet-mounted displays. Journal of Experimental Psychology: Applied, 3, 163-180
Siegel, J., & Bauer, M. (1997). On site maintenance
using a wearable computer system, CHI97 (pp.
119-120). New York: ACM.

Human Factors Problems of Wearable Computers

Sheridan, J. G., Lafond-Favieres, V., & Newstetter, W. C. (2000). Spectators at a Geek Show: An
Ethnographic Inquiry into Wearable Computing,
Digest of Papers of the 4th International Symposium on Wearable Computers (pp. 195-196). Los
Alamitos, CA: IEEE Computer Society.
Starner, T. (1999). Wearable computing and
contextual awareness. Unpublished Ph.D. thesis,
Masachusetts Institute of Technology, USA.
Stiefmeier, T., Ogris, G., Junker, H., Lukowicz, P.
& Trster,

G. (2006). Combining motion sensors

and ultrasonic hands tracking for continuous activity recognition in a maintenance scenario. Digest
of Papers of the 10th International Symposium on

Alamitos,
Wearable Computers (pp. 97-104). Los
CA: IEEE Computer Society.
Stein, R., Ferrero, S., Hetfield, M., Quinn, A.,
& Krichever, M. (1998). Development of a commercially successful wearable data collection
system. Digest of Papers of the 2nd International
Symposium on Wearable Computers (pp. 18-24).
Los Alamitos, CA: IEEE Computer Society.
Strachan, S., Williamson, J., & Murray-Smith, R.
(2007). Show me the way to Monte Carlo: density-based trajectory navigation. In Proceedings
of ACM SIG CHI Conference, San Jose.

Thomas, B., Tyerman, S., & Grimmer, K. (1997).

Evaluation of three input mechanisms for wearable
computers. In Proceedings of the 1st International
Symposium on Wearable Computers (pp. 2-9). Los
Alamitos, CA: IEEE Computer Society.
Thomas, B., Grimmer, K., Makovec, D., Zucco,
J., & Gunther, B. (1998). Determination of placement of a body-attached mouse as a pointing input
device for wearable computers. Digest of Papers
of the 3rd International Symposium on Wearable
Computers (pp. 193-194). Los Alamitos, CA:
IEEE Computer Society.
Van Laerhoven, K., &Gellersen, H. W. (2004).
.Spine versus porcupine: A Study in distributed
wearable activity recognition. In Proceedings of
the 8th IEEE International Symposium on Wearable Computers (ISWC04), (pp. 142-149). Los
Alamitos, CA: IEEE Computer Society.
Van Langehoven, L., & Hertleer, C. (2004). Smart
clothing: a new life. International Journal of
Clothing Science and Technology, 13, 63-72.
Want, R., Hopper, A., Falcao, V., & Gibbons, J.
(1992). The active badge location system. ACM
Transactions on Information Systems, 10(1),
91-102.

Strachan, S., Eslambolchilar, P., Murray-Smith,

R., Hughes, S., & OModhrain, S. (2005). GpsTunesControlling navigation via audio feedback. Mobile HCI 2005.

Ward, J.A., Lukowicz, P., & Trster,

G. (2006).
Activity recognition of assembly tasks using using body-worn microphones and accelerometers.
IEEE Transaction on Pattern Analysis and Machine Intelligence.

Suomela, R., Roimela, K., & Lehikoinen, J.

(2003), The evolution of perspective view in
WalkMap. Personal and Ubiquitous Computing,
7, 249-262.

Warren, N., Jones, M., Jones S., & Bainbridge,

D. (2005, April 2-7). Navigation via continuously
adapted music. Extended Abstracts ACM CHI
2005, Portland, Oregon.

Tan, H. Z., & Pentland, A. (1997). Tactual displays for wearable computing. Digest of Papers
of the 1st International Symposium on Wearable
Computers (pp. 84-89). Los Alamitos, CA: IEEE
Computer Society.

Wellman, B. (2001). Physical place and cyberspace: The rise of personalized networks. International Journal of Urban and Regional Research,
25, 227-252.

1151

Human Factors Problems of Wearable Computers

Westeyn, T., Brashear, H., Atrash, A. & Starner,

T. (2003). Georgia Tech Gesture Toolkit: Supporting Experiments in Gesture Recognition,
ICMI03, ??
Wijesiriwardana, R., Dias, T. & Mukhopadhyay, S.
(2003). Resisitivefiber meshed transducers. Digest
of Papers of the 7th International Symposium on
Wearable Computers (pp. 200-205). Los Alamitos,
CA: IEEE Computer Society.
Wijesiriwardana, Mitcham, K. R., & Dias, T.
(2004). Fibre-meshed transducers based on
real-time wearable physiological information
monitoring system. Digest of Papers of the 8th
International Symposium on Wearable Computers
(pp. 40-47). Los Alamitos, CA: IEEE Computer
Society.
Zingale, C., Ahlstrom, V., & Kudrick, B. (2005).
Human factors guidance for the use of handheld,
portable and wearable computing devices. Atlantic City, NJ: Federal Aviation Administration,
reprt DOT/FAA/CT-05/15
Zucco, J. E., Thomas, B. H. & Grimmer, K.
(2006). Evaluation of four wearable computer
pointing devices for drag and drop tasks when
stationary and walking. Digest of Papers of
the 10th International Symposium on Wearable
Computers (pp. 29-36). Los Alamitos, CA: IEEE
Computer Society.

KEY TERMS
Activity Models: Predictive models of human
activity, based on sensor data
Augmentation Means: Devices that can augment human bevaiora term coined by Doug
Engelbart, and covering: Tools & Artifacts: the
technologies that we use to work on the world
which supplement, complement or extend our
physical or cognitive abilities; Praxis: the ac-

1152

cumulation and exploitation of skills relating to

purposeful behavior in both work and everyday
activity; Language: the manipulation and communication of concepts; Adaptation: the manner
in which people could (or should) adapt their
physical and cognitive activity to accommodate
the demands of technology.
Comfort: Subjective response to wearing a
wearable computer (ranging from physical loading to embarrassment)
Context-Awareness: The capability of a
device to respond appropriately to changes in a
persons activity, environment, and so forth.
Form-Factor: The overall size (and shape)
of a device
Sensors: Devices that produce digital output in
response to some change in a measured parameter,
for example dependent on environmental change
or on user activity
Wearable Computers: Devices worn on the
person that provided personalized, context-relevant information

Endnotes

http://www.intel.com/research/exploratory/
motes.htm
2

Bainbridge (1987) argued that full automation can lead to the ironic situation that, the
role of the human operator is to intervene
when something goes wrong. However, the
automation is such that the human is locked
out of the process and has little understanding as to what is happening. Consequently,
the human will not be able to intervene in an
informed and efficient manner. Ultimately,
it means that, by designing the human out
of the system, the potential for a flexible and
intelligent response to unknown situations
is lost.
1

Human Factors Problems of Wearable Computers

3
4

http://www.bodymedia.com/main.jsp
http://www.extra.research.philips.com/
pressmedia/pictures/wearelec.html

This work was previously published in Handbook of Research on User Interface Design and Evaluation for Mobile Technology, edited by J. Lumsden, pp. 158-175, copyright 2008 by Information Science Reference, formerly known as Idea Group
Reference (an imprint of IGI Global).

1153

1154

Chapter 3.10

Modelling Interactive Behaviour

with a Rational Cognitive
Architecture
David Peebles
University of Huddersfield, UK
Anna L. Cox
University College London, UK

ABSTRACT
In this chapter we discuss a number of recent studies that demonstrate the use of rational analysis
(Anderson, 1990) and cognitive modelling methods to understand complex interactive behaviour
involved in three tasks: (1) icon search, (2) graph
reading, and (3) information retrieval on the World
Wide Web (WWW). We describe the underlying
theoretical assumptions of rational analysis and the
adaptive control of thought-rational (ACT-R) cognitive architecture (Anderson & Lebiere, 1998),
a theory of cognition that incorporates rational
analysis in its mechanisms for learning and decision making. In presenting these studies we aim
to show how such methods can be combined with
eye movement data to provide detailed, highly
constrained accounts of user performance that are
grounded in psychological theory. We argue that

the theoretical and technological developments

that underpin these methods are now at a stage
that the approach can be more broadly applied to
other areas of Web use.

INTRODUCTION
With the rapid increase in Internet use over the past
decade there is a growing need for those engaged
in the design of Web technology to understand the
human factors involved in Web-based interaction.
Incorporating insights from cognitive science
about the mechanisms, strengths, and limits of
human perception and cognition can provide a
number of benefits for Web practitioners. Knowledge about the various constraints on cognition,
(e.g., limitations on working memory), patterns
of strategy selection, or the effect of design

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Modelling Interactive Behaviour with a Rational Cognitive Architecture

decisions (e.g., icon style) on visual search, can

inform the design and evaluation process and
allow practitioners to develop technologies that
are better suited to human abilities.
The application of cognitive psychology to
human-computer interaction (HCI) issues has
a long history going back to Card, Moran, and
Newells (1983) introduction of the goals, operators, methods, and selection rules (GOMS) task
analysis technique and model human processor
(MHP) account of human information processing in the early 1980s. Since then, their cognitive
engineering approach has developed into a family of methods (John & Kieras, 1994; Olson &
Olson, 1990) which are widely used to produce
quantitative models of user performance in interactive tasks.
Another, more recent approach to modelling
human performance in interactive tasks has
emerged in the last decade from theoretical and
technological advances in research into cognitive
architectures. Cognitive architectures are theories
of the fundamental structures and processes that
underlie all human cognition, of which there are
several currently in existence including EPIC
(executive process / interactive control; Kieras
& Meyer, 1997), Soar (Laird, Newell, & Rosenbloom, 1987; Newell, 1990), and ACT-R (Anderson & Lebiere, 1998; Anderson et al., 2004). An
important feature of these architectures is that
they are all implemented as computer programming systems so that cognitive models may be
specified, executed, and their outputs (e.g., error
rates and response latencies) compared to human
performance data.
Originally ACT-R and Soar were theories
of central cognition only and did not explicitly
specify mechanisms for perception or motor control. EPIC however, was unique in that from its
inception it incorporated processors for cognition,
perception, and motor control. Recent adaptations
to ACT-R (Byrne & Anderson, 1998) and Soar
(Chong & Laird, 1997) have now ensured that
both architectures incorporate perceptual motor

components that allow models to include visual

attention processes and manual interactions with a
keyboard and mouse. This is an important development for the study of HCI as cognitive models
can now be embodied (Kieras & Meyer, 1997) in
the sense that the architectures are now able to
simulate perceptual-motor contact with computer
interfaces and devices and so capture the complex
interactions between the task environment, cognition, and perceptual-motor behaviour.
Modelling interactive behaviour with an
embodied cognitive architecture has a number
of advantages over the traditional cognitive engineering approach exemplified by GOMS and its
relatives. Perhaps the most important of these is
that computational models can actually execute
the task, allowing a direct test of the sufficiency
of the hypothesised processes. Second, although
most cognitive architectures contain built-in
timing parameters taken from the psychological
literature, unlike cognitive engineering models,
they do not require prior estimated times for all
subcomponents of a task. In addition, some architectures such as ACT-R and Soar contain
learning mechanisms which allow them to model
various effects of practice on performance. This
allows cognitive architectures to be used to model
novel tasks, novice users, or tasks involving components without prior time estimates.
One of the promises of embodied cognitive
architectures is that, once they are equipped with
sufficient knowledge, they will begin to provide
a priori predictions of user performance and
eventually evolve into artificial users that can be
employed to evaluate novel tasks and environments (Ritter, Baxter, Jones, & Young, 2000;
Young, Green, & Simon, 1989). In this chapter we
will describe one of these architectures, ACT-R,
and show how it has been used to provide detailed and sophisticated process models of human
performance in interactive tasks with complex
interfaces. ACT-R is an appropriate choice for
this discussion because, in contrast to other
cognitive architectures, ACT-R also embodies

1155

Modelling Interactive Behaviour with a Rational Cognitive Architecture

the rational theory of cognition (Anderson, 1990)

which analyses cognitive phenomena in terms of
how they are adapted to the statistical structure of
the environment. Rational analysis and ACT-Rs
mechanisms have been used recently to provide
novel insights into Web-based interactions. The
chapter proceeds as follows: First we describe the
basic assumptions and mechanisms of rational
analysis and the ACT-R cognitive architecture. We
then show how these have been used to develop
a model of information foraging on the Web and
discuss the model in relation to a rational analysis
model of the task and the data from eye-tracking
studies of interactive search. In the final sections
of this chapter we briefly outline ACT-R models
of two interactive tasks; graph reading (Peebles
& Cheng, 2003) and icon search (Fleetwood &
Byrne, in press). Although neither of these studies
involves a specifically Web-based task, they both
describe user interaction with items commonly
found on Web pages. They are also illustrative
of a methodology that combines task analysis,
eye tracking, and formal modelling to provide a
detailed account of the cognitive, perceptual, and
motor processes involved in the performance of
the task. These studies are also useful because in
both cases the model is validated by comparing
the simulated eye movements with those recorded
from human subjects. Both studies, therefore,
are clear demonstrations of a novel approach to
understanding interactive behaviour that can be
applied to Web-based tasks.

define the cognitive mechanisms required by the

agent to solve the task, rational analysis suggests
that we should consider the structure of the task
itself, the environment in which it is encountered,
together with some minimal assumptions about
the computational limitations of the system. From
these initial statements the analysis proceeds by
the specification of an optimal solution to the
problem and the comparison of human behavioural
data to see how close an approximation it is to
the optimal solution.
By identifying the best way to complete the
task (the optimal strategy) we can often infer what
the cognitive mechanisms of a rational agent must
be as although humans do not always complete
tasks in the most optimal way their behaviour is
usually similar to the optimal strategy. That is,
humans usually behave in such a way that they
appear to be trying to complete their tasks in the
most efficient manner, that is, they try to maximise their returns while minimising the cost of
achieving their goals.
Rational analysis has been applied to several
aspects of human cognition (see e.g., Oaksford
& Chater, 1998), from the original analyses of
memory, categorisation, causal inference, and
decision making conducted by Anderson (1990),
to more recent analyses of exploratory choice (Cox
& Young 2004; Young, 1998) and the updating of
memory during tasks in dynamic environments
(Neth, Sims, Veksler, & Gray, 2004).

RATIONAL ANALYSIS

THE ACT-R COGNITIVE

ARCHITECTURE

Rational analysis (Anderson, 1990) is a method

for understanding the task an agent attempts to
complete. It assumes that humans have evolved
cognitive mechanisms that are useful for completing tasks that we encounter in our environment,
and that these mechanisms work in an efficient
way to complete these tasks. Therefore, rather
than concerning ourselves with firstly trying to

ACT-R is a theory of human cognition developed

over a period of 30 years by John Anderson and
his colleagues (Anderson & Lebiere, 1998; Anderson et al., 2004) that incorporates the theory
of rational analysis. It is a principal effort in the
attempt to develop a unified theory of cognition
(Newell, 1990). As a cognitive architecture,
ACT-R attempts to specify the basic cognitive

1156

Modelling Interactive Behaviour with a Rational Cognitive Architecture

structures and processes that underlie all human

cognition.
Figure 1 illustrates the components of the
architecture relevant to our discussion. ACT-R
consists of a set of independent modules that
acquire information from the environment,
process information, and execute motor actions
in the furtherance of particular goals. There are
four modules that comprise the central cognitive
components of ACT-R. Two of these are memory
stores for two types of knowledge: a declarative
memory module that stores factual knowledge
about the domain, and a procedural memory
module that stores the systems knowledge about
how tasks are performed. The former consists
of a network of knowledge chunks whereas the
latter is a set of productions, rules of the form
IF <condition> THEN <action>: the condition
specifying the state of the system that must exist
for the rule to apply and the action specifying the
actions to be taken should this occur. The other two
cognitive modules represent information related
to the execution of tasks. The first is a control
state module that keeps track of the intentions
of the system during problem solving, and the
second is a problem state module that maintains
the current state of the task.
In addition to these cognitive modules there
are four perceptual-motor modules for speech,
audition, visual, and motor processing (only the
latter two are shown in Figure 1). The speech and
audition modules are the least well-developed

and, at present, simply provide ACT-R with the

capacity to simulate basic audio perception and
vocal output for the purpose of modelling typical
psychology experiments. The visual and motor
modules are more well-developed and provide
ACT-R with the ability to simulate visual attention shifts to objects on a computer display and
manual interactions with a computer keyboard
and mouse.
Each of ACT-Rs modules has an associated
buffer that can hold only one chunk of information
from its module at a time, and the contents of all
of the buffers constitute the state of an ACT-R
model at any one time. Cognition proceeds via
a pattern matching process that attempts to find
productions with conditions that match the current contents of the buffers. There then follows a
process to select the best production from those
that match the conditions, after which the most
appropriate production fires and the actions
(visual or manual movements, requests for the
retrieval of a knowledge chunk from declarative
memory, or modifications to buffers) are performed. Then the matching process continues on
the updated contents of the buffers so that tasks
are performed through a succession of production
rule firings. As an example, two production rules
(written in English rather than in ACT-R code)
that instantiate part of a search task may look
something like Box 1.
The processing in ACT-Rs modules is serial
but the modules run in parallel with each other so

Box 1.
IF

the goal is to find the meaning of eudaimonia (control state)

AND

there is nothing in declarative memory about eudaimonia (declarative)

THEN

set the goal to search the WWW for eudaimonia (control state)

IF

the goal is to search the WWW for eudaimonia (control state)

AND

the Web browser is open (problem state)

THEN

look for the menu labelled Bookmarks (visual)

AND

update the problem state to looking for Google (problem state)

1157

Modelling Interactive Behaviour with a Rational Cognitive Architecture

Figure 1. The modular structure of ACT-R 6.0

that the system can move visual attention while

also moving the mouse and attempting to retrieve
knowledge from declarative memory. ACT-R
processes also have associated latency parameters
taken from the psychology literature. For example,
it typically takes 50 ms for a production to fire
and the time taken to move the mouse cursor to
an object on the computer screen is calculated
using Fitts Law (Fitts, 1954).
ACT-R implements rational analysis in two
ways. The first is its mechanism for retrieving
knowledge chunks from declarative memory
which is based on the notion of activation. Each
chunk in declarative memory has a level of
activation which determines its probability and
latency of retrieval, and the level of activation for
a chunk reflects the recency and frequency of its
use. This enables us to understand how rehearsal
of items in a short-term memory task can boost
the activation levels of these chunks and consequently increase the chances of recall/retrieval
from declarative memory. The level of activation
of a chunk falls gradually over time, and without
retrieval or activation spreading from chunks in
the current goal, it may fall below a threshold
level which then results in retrieval failure. This
enables ACT-R models to forget knowledge

1158

without having to explicitly delete chunks from

the declarative memory store.
The second way that ACT-R implements rational analysis is in its mechanism for choosing
between alternative production rules. According
to rational analysis, people choose between a
number of options to maximise their expected
utility. Each option (i.e., production rule) has an
expected probability of achieving the goal and an
expected cost. It is assumed that when carrying
out computer-based tasks people interact with
the task environment and choose actions that
will optimise their efficiency (i.e., maximise the
probability of achieving the goal while minimising the cost, usually measured in units of time).
At each decision step in the cycle, therefore, all
possible production rules that match against the
current goal are proposed in a choice set, and the
one with the highest level of efficiency is chosen
and executed.
ACT-R has been used to model a wide range
of cognitive phenomena (Anderson & Lebiere,
1998), and in recent years, with the inclusion of the
perceptual-motor modules, it has been applied to
a number of complex interactive tasks in the area
of HCI and human factors research, for example,
menu selection (Byrne, 2001), cell phone menu

Modelling Interactive Behaviour with a Rational Cognitive Architecture

interaction (St. Amant, Horton, & Ritter, 2004),

and driving (Salvucci & Macuga, 2002). Although
individually these models do not yet offer us a
virtual user which can be sat in front of a Web
browser and asked to complete any goal, together
they provide us with insights into how and why
users behave in particular ways, for example,
when searching for information on the Web. In
this chapter we will concentrate on three particular
areas of work that are relevant to understanding
Web behaviour: icon search, graph reading, and
information foraging on the WWW.

MODELLING INTERACTIVE
BEHAVIOUR
In the following section, we will summarise a
number of recent studies which employ rational
analysis, cognitive modelling, eye tracking, or a
combination of all three, to understand human
performance in Web-based or HCI tasks. We
first discuss recent efforts to model information
foraging and interactive search on the WWW.
These studies show how ACT-R and rational
analysis can be successfully applied to explain
different aspects of peoples behaviour when conducting interactive search tasks. This can include
both high-level behaviours such as backtracking
through Web-pages and low-level behaviours
such as patterns of visual attention obtained from
eye-tracking studies. We then describe two studies which combine experimental data collection,
eye movement recording, and cognitive modelling
methods using ACT-R to provide detailed accounts
of the cognitive, perceptual, and motor processes
involved in the tasks. These studies were chosen
because both develop a detailed process model
which not only captures the human response
time data from the experiment, but also provides
a close match to the patterns of visual attention
revealed by the eye movement study. This level of
detail in modelling is still relatively uncommon
and the strong constraints added by seeking to

match model and human eye movement scan paths

during the course of the task provide a further
validation of the models.

Information Foraging on the World

Wide Web
Information foraging theory (IFT; Pirolli &
Card, 1999; Pirolli, 2005) describes an account
of information gathering behaviour based on the
ecological behaviours of animals when foraging
for food. The account can be applied to situations
in which people are searching for information
in a number of different situations such as in
a library or on the WWW. The theory rests on
rational analysis in that it proposes that human
behaviour is directed by the objective to maximise
gain and minimise effort, and that this process
is sensitive to changes in the environment. In
contrast to animal studies, where the assumption
is that animals seek to reduce the ratio of calorie
intake to energy expenditure, the assumption in
IFT is that people attempt to reduce the ratio of
information gained to time spent.
The way in which the environment is structured
determines the costs of search for information.
For example, the structure of a Web site will determine how many pages the user has to navigate
through in order to satisfy his/her goal. When
searching for information on the WWW, many
people make use of search engines. After entering
some key words the user is presented with a list of
search results which are usually ordered in terms
of their relevance to the key words. Each of the
results returned can be considered to be a patch
of information. The user has to choose to either
investigate one of the patches or to redefine their
search criteria. Conducting another search using
different key words will result in a change in the
environment. This process is known as enriching
the environment as it is hoped that the result is
that the cost of obtaining the required information
will be reduced compared to the perceived cost of
obtaining it in the previous environment. Deci-

1159

Modelling Interactive Behaviour with a Rational Cognitive Architecture

sions about whether or not to pursue a particular

information patch or to continue enriching the
environment are based on a number of factors
such as the perceived value of the information
returned, the perceived costs of acquiring that
information, interface constraints, and previous
knowledge.
The decision to forage within a particular patch
of information is based on an ongoing assessment
of information scent. Information scent is the
perception of the value of the distal information
based on the proximal information available, that
is, it is an estimate of the relevance of the information contained on a yet unseen page based on the
cues from the icon or wording of the link on the
page currently viewed. The theory predicts that
as more time is allocated to within-patch foraging, the rate of information return increases but
only up to an optimal point, after which the rate
starts to decrease. Therefore, after a particular
amount of within-patch foraging (searching within
a Web site) it becomes more profitable to move to
the next patch (select another Web site from the
list of search results) even though there are still
pages within the previous patch that have not yet
been visited.

SNIF-ACT
Scent-based Navigation and Information Foraging in the ACT architecture (SNIF-ACT) (Pirolli
& Fu, 2003) is a model of human behaviour in
an interactive search task. The model makes use
of ACT-Rs spreading activation mechanism so
that the information scent of the currently viewed
Web page activates chunks in declarative memory
as does the spreading activation from the goal.
Where these two sources of activation coincide
there are higher levels of activation and this indicates a high degree of relevance between the
goal and the page being attended to. This activation is what ultimately drives the behaviour of
the model. The model includes the use of search

1160

engines to provide a set of search results and the

processing of the page that is returned. The links
on the page are attended to and eventually one of
the links is selected.
The behaviour of the model is compared to
user behaviour and successfully demonstrates
that people tend to select the highest scent item
in a list. SNIF-ACT does this by assessing the
information scent of all the links on a page and
then choosing the highest one. The model is also
able to explain the point at which a user abandons
a particular Web site and returns to the search
results in order to select another item from the list
or selects a link that takes them to another Web
site. If the mean information scent of the currently
viewed page is lower than the mean information
scent of a page on another site the model selects
that action that takes them to the other site.

Eye-Tracking Experiments in
Interactive Search
When presented with a list of search results or
items on a menu within a Web site (i.e., a patch
of information), the user has to choose between
selecting an item which will move him/her to
another patch and doing some assessment on
either the currently attended item or some other
item in the list (i.e., consume the information
presented within the current patch). As has been
mentioned previously, IFT proposes that the user
will make use of the information scent of the items
to guide their behaviour. If the information scent
of a particular item in the list is higher than the
rest (i.e., that item appears to be relevant to the
task and the user believes that clicking it will
lead them to better information) then the item
will be selected.
Eye-tracking experiments have been used to
investigate what people attend to when conducting
interactive search tasks (Brumby & Howes, 2004;
Silva & Cox, 2005). Participants were given an
information goal and a list of items and asked to
select the label that they thought would lead to the

Modelling Interactive Behaviour with a Rational Cognitive Architecture

information they required. Brumby and Howes

demonstrated that people often examine only a
subset of the list before selecting the target item,
and that this behaviour is affected by the relevance
of the other items in the list. When the other items
in the list are more relevant to the goal (i.e., they
have high levels of information scent), people tend
to look at more items in the list and also tend to
look at individual items on more occasions than
when the items are irrelevant. When there are
a number of items with high scent (i.e., two or
more items look like they would lead to relevant
information) people need to consider more items
than when only one item looks sensible.
However, one limitation of this work is that the
analysis of eye-tracking data is rarely sensitive
enough to determine whether a lack of fixation of
the eyes on an item really means that people have
not assessed the relevance of the item. In order
to address this, Silva and Cox (2005) additionally employed a recognition task in their study
in order to assess the level of processing of each
item in the list.
Figure 2 represents a simplified scan path of
a participant completing one of these tasks. The
items are represented on the y axis with time along
the x axis. The highlighted item is the target item
and was selected by the participant. The figure
demonstrates how the user starts at the top of the
list and scans down the list fixating items in the
list. Some of the items (3 & 6) are skipped over.
The results from Silva and Coxs (2005) recognition task suggest that in such cases the lack of
fixations of particular items in the menu can be
explained by parafoveal processing. However,
parafoveal processing can only explain lack of
fixations on up to two items below the last fixation (i.e., items 8 & 9) and cannot explain why
the user does not attend to other items in the list
(i.e., items 10 to 16).
SNIF-ACT would be able to produce a trace
that would match the behaviour of users in these
studies in terms of which items from the menus
the user selected. However, the model does not

account for the fact that some of the items in the

menus were not assessed by the users as it assumes
that users have knowledge about information
scent of all the items in the list and then selects
the item with the highest level of scent. Consequently, SNIF-ACT is unable to provide us with
any explanation for why users should choose to
select an item when they have not even read the
entire list presented to them.
Cox and Young (2004) propose an alternative
model to that of SNIF-ACT that is able to capture
this fine-grained level of detail of user behaviour.
Their model is a rational analysis of an interactive
search task that provides a rational explanation
of why the user would select an item without first
assessing all the items in the list.
In interactive search, the agent has the goal of
selecting the item that will lead to goal completion. However, as the menu presented is novel,
the first thing that the model has to do is to gain
some information about the menu. The model

Figure 2. A simplified scan path of a participant

performing an interactive search task
-1
-2
-3
-4
-5
-6
-7
-8
-9
- 10
- 11
- 12
- 13
- 14
- 15
- 16
- 17

1161

Modelling Interactive Behaviour with a Rational Cognitive Architecture

therefore includes two types of exploratory acts

(EAs) (these are the different types of things the
model can do): assess information SCENT and
ANTICIPATE the result of selecting this item.
The SCENT EA should be thought of as being
an amalgamation of perceiving the label, reading
the label (at a lexical level), and considering the
semantic similarity between the label and the
current task. The ANTICIPATE EA should be
thought of as some additional cognitive effort
that considers whether the label is likely to lead
to the goal. For example, given the goal of finding
an armchair for your living room on a furniture
shop Web site, imagine the model considering
the first item in the menu home. The SCENT
EA would return a moderately high rating as the
label has a moderately high level of information
scent given the goal (home and armchair).
The ANTICIPATE EA models the agents consideration of whether the label home is likely to
lead to the home page of the site, or to a list of
home furnishings. Each of these EA types has
a cost associated with it with the ANTICIPATE
EA type being more expensive in mental effort
than the first type. There is also a fixed cost of
moving attention from one item in the menu to
the next.
Before assessing any items, the model knows
the number of items in the menu and considers
each of these items to be equally (ir)relevant to
completing the task. The scent ratings of the items
in the menu are used as the basis for determining
the new relevance (R) value of an item following
an assessment. On each page, the set of relevancies Ri are mapped into a set of probabilities Pi
by the transformation Pi = odds(Ri)/odds(Rj),
where odds(R) is defined in the standard way as
odds(R) = R/(1R). Note that Pi = 1, reflecting
the fact that exactly one option on the page leads
to the goal.
When the model is run on a set of menus it
demonstrates how different patterns of information scent result in different behaviours. As
Brumby and Howes (2004) demonstrated, the

1162

levels of information scent of both the goal item

and the distractors affect behaviour. However, it
is also interesting to note that the model predicts
that just the change in position of the goal item
relevant to the distractors results in different
patterns of behaviour: Sometimes the model
predicts that users will scan to the bottom of the
menu before selecting the target item, and other
times they will select the item immediately after
assessing the item leaving other items in the menu
unassessed. To explain how this occurs we will
compare the behaviour of the model when the high
scent item is in position two (as an example of
occurring early in the menu) and in position 12 (as
an example of occurring late in the menu) in more
detail. In both examples, initially, all 16 menu
items are rated equally and all have an R value
of 0.06. The relevance values are translated into
efficiencies (E) which are then used to determine
which of the EAs is most likely to lead to the goal
and therefore which EA is executed in each cycle.
In the first cycle, the EA that proposes assessing
the scent of the first item in the menu is rated as
having the highest E value due to it having the
lowest cost. Consequently, the model assesses the
first item which gets rated as very low scent. As
a result, the new R value of this item is set at 0.
On the next cycle, the EA that proposes SCENT
assessment on the second item in the list is the
most efficient (due to the lower cost) so this item
gets assessed. This behaviour continues until the
model assesses the high scent item.
In menus where the high scent item occurs early
on in the menu, the second item in the menu gets
an R value of 0.5097 which raises the probability
that this item will lead to the goal to 0.6220. On
the following cycle the R value of the high scent
item leads to an E value of 0.008 while the second
best item (an item yet to be assessed) has an R
value of 0.06 which results in an E value of 0.006.
Although the E values of the two EAs are very
similar, one is larger than the other, and this is
what determines which EA is chosen.

Modelling Interactive Behaviour with a Rational Cognitive Architecture

In our example of a menu where the high scent

item occurs later on in the menu, the relevance of
each of the low scent items that have already been
assessed falls to 0. When the model assesses the
twelfth item its R value is 0.5097, which raises
the probability that this item will lead to the goal
to 0.6220. On the following cycle the R value of
the high scent item only has an E value of 0.005
while the item with the best efficiency (an item
yet to be assessed) has an R value of 0.05 which
results in an E value of 0.006. The result is that the
model continues to assess each item in the menu
until it reaches the bottom because the efficiency
of conducting a SCENT assessment of a new
item is greater than the efficiency of conducting
the ANTICIPATE assessment on the high scent
item in position 12. This has the effect of slowly
increasing the probability of the item in position
12 leading to the goal.
The detail of the model explains that the reason
the behaviour is different for the two types of
menus is because the detail of the mathematics of
the rational analysis. Comparisons of the traces of
the model with the empirical data suggest that the
model provides a good explanation of the cognitive
processes involved in this task. This suggests that
participants make an assessment of the relevance
of a label to the current goal and then, together
with the estimated relevance of previous items,
choose to either (1) select that item as the one that
will lead to the goal, (2) conduct some further
assessment of the current item, or (3) move on to
another item and assess that. Which of these EAs
is chosen is driven by the pattern of information
scent that has been experienced so far.
The model provides us with an explanation
of how and why the position of the goal and the
quality of the distractor items affect the behaviour
of the participants on the task. Regardless of the
pattern of scent of the menu, the model predicts
that the agent will tend to stop exploring the menu
as soon as it comes across a menu item that has
high information scent (self-terminates) if this is
encountered early in the menu. On menus where

there is one high scent item among a set of low

scent items and the high scent item occurs later
in the menu, the agent continues to assess the
other items in the menu before conducting further assessment of the high scent item and finally
selecting it. The model enables us to explain why
we see these different patterns of behaviour on
menus which have such similar patterns of information scent. This is due to the effect of the
interdependence of the probability that each of
the items will lead to the goal. The actual point
on the menu at which the model swaps from one
behaviour to the other is sensitive to a number
of factors such as the length of the menu and the
costs of the EAs. It would appear therefore that it
is in the nature of interactive search that there are
close calls which suggest that people can rationally
do either behaviour and that a number of factors
have an effect on the behaviour of participants
exploring real menus.
Together the two models described previously
provide us with a good understanding of how
people perform search tasks on the WWW. SNIFACT and the rational model explain different aspects of the interaction: SNIF-ACT demonstrates
the higher level, page by page, link following
behaviour seen in such tasks, whereas the rational model explains the lower level interactions
with just one page. Given information about the
information scent of the items on a new Web site
both models are able to make predictions about
user behaviour on the site.

Modelling Graph Reading

Peebles and Cheng (2003) conducted an experiment, eye movement study and cognitive modelling analysis to investigate the cognitive, perceptual, and motor processes involved in a common
graph-reading task using two different types of
Cartesian graph. The purpose of the study was
to determine how graph users ability to retrieve
information can be affected by presenting the same
information in slightly different types of the same

1163

Modelling Interactive Behaviour with a Rational Cognitive Architecture

class of diagram. The two types of graph, shown

minimum sequence of saccades and fixations rein Figure 3, represent amounts of UK oil and gas
quired to reach the location of the given variables
production over two decades. The only difference
value and then from there to the location of the
between the two graph types is in which variables
corresponding value of the target variable (Lohse,
are represented on the axes and which are plotted.
1993; Peebles & Cheng, 2001, 2002; Peebles,
In the Function graphs, the argument variable
Cheng, & Shadbolt, 1999). Experiment partici(AV: time in years) is represented on the x-axis
pants (some of whom had their eye movements
and the quantity variables (QV: oil and gas) on
recorded) completed 120 trials, each participant
the y-axis whereas in the Parametric graphs, the
using only one graph type. The 120 questions
quantity variables are represented on the x and y
were coded into three classes (QVQV, QVAV,
axes and time is plotted on the curve.
and AVQV) according to which variables value
In the experiment, participants were presented
was given and which was required (QV denotes a
with the value of a given variable and required
quantity variable, oil or gas, and AV denotes the
to use the graph to find the corresponding value of
argument variable, time). On each trial, a question
a target variable, for example, when the value
(e.g., GAS = 6, OIL = ?) was presented above
of oil is 2, what is the value of gas? This type of
the graph and participants were required to read
task has typically been analysed
terms ofBehaviour
the
theaquestion,
find the answer
using301
the graph on
Modeling in
Interactive
with
Rational Cognitive
Architecture
Figure 3. Function
and parametric
graphs graphs
used inused
Peebles
and Cheng
(2003)
depicting
values of oil
Figure 3. Function
and parametric
in Peebles
and Cheng
(2003)
depicting
values
of
oil
and
gas
production
for
each
year
and gas production for each year

Notes: The graphs on the left (labelled 1) show years 1970 to 1979 while those on the right (labelled 2) show years
1980 to 1989. Dashed lines indicate the optimal scan path required to answer the question, when the value of
oil is 3, what is the value of gas?

1164

corresponding value of the target variable (Lohse, 1993; Peebles & Cheng, 2001, 2002;

Modelling Interactive Behaviour with a Rational Cognitive Architecture

the screen and then enter their answer by clicking

on a button labelled Answer in the top right corner
of the window which revealed a circle of buttons
containing the digits 0 to 9. RTs were recorded
from the onset of a question to the mouse click
on the Answer button.
The RT data from the experiment, displayed
in Figure 4, showed that the graph used and the
type of question asked both had a significant effect
on the time it took for participants to retrieve the
answer. This was all the more surprising because,
for two of the three question types, participants
were faster using the less familiar parametric
graphs by nearly a second.
The results of the eye movement study were
also surprising. It was found that in 63% of trials
(irrespective of the graph used or question type
being attempted), after having read the question
at the start of a trial, participants redirected their
visual attention to elements of the question at least
once during the process of problem solving with
the graph. This was not predicted by the simple
minimal fixation sequence account outlined
previously but two possible explanations may
be provided: (1) participants initially encode the
three question elements but are unable to retain
all of them in working memory and retrieve
them by the time they are required to do so, or
(2) to reduce the probability of retrieval failure,

participants break the problem into two sections,

the first allowing them to reach the given location and the second to then proceed to the target
location corresponding to the solution.
Peebles and Cheng (2003) constructed two
ACT-R models of the experiment (one for each
graph type) that were able to interact with an exact
replica of the experiment software. The models
consisted of a set of productions to carry out the
six basic subgoals in the task; (1) read the question;
(2) identify the start location determined by the
given variable; (3) identify the given location on
the graph representing the given value of given
variable; (4) from the given location, identify the
target location representing the required variable;
(5) identify the target value at the target location;
and (6) enter the answer. Many of the productions
were shared by the two models, the main difference between them being the control structure that
sequences the execution of the productions. Figure
4 shows that the mean RTs from the parametric
and function graph models are a good fit to the
observed data (R 2 = .868, RMSE = 0.123, and R 2
= .664, RMSE = 0.199 respectively). Perhaps more
importantly however, were the insights into the
observed eye movement data that came from the
modelling process itself. When ACT-R focuses
attention on an object on the screen, representations of the object and its location are created in

Figure 4. Mean response times for experimental participants and ACT-R models for each question type
(Peebles & Cheng, 2003)

1165

Modelling Interactive Behaviour with a Rational Cognitive Architecture

the systems visual buffers which can be accessed

by productions. Eventually these representations
go into declarative memory with initial activation
values and, as long as these values are above a
certain threshold, they can be retrieved by the
cognitive system and replaced in a buffer. However, ACT-R includes a mechanism by which
the activation of representations in declarative
memory decreases over time which allows it to
simulate processes involved in forgetting. These
mechanisms played a crucial role in the ACT-R
models ability to capture the eye movement data
observed in the experiment. At the start of each
trial, the models read the three question elements
and during the problem solving these elements are
placed in declarative memory. As a consequence,
at least one question element must be retrieved
from memory at each stage of the problem in
order to continue. However, as soon as a question
element is placed in declarative memory its activation starts to decay and, as a consequence, the
probability that it cannot be retrieved increases.
Typically, if a retrieval failure occurs, an ACT-R
model will halt as it does not have the appropriate information to solve the problem. During the
process of model development it was found that

on a significant proportion of trials the model was

not able to retrieve question elements at the later
stages of the trial because their activation had
fallen below the retrieval threshold. As a consequence new productions had to be added to allow
the model to redirect attention to the question in
order to re-encode the element and then return
to solving the problem. This was precisely the
behaviour observed in the eye movement study.
This is illustrated in Figure 5 which compares
screen shots of the model scan path and eye
movements recorded from one participant for
the same question using the 1980s parametric
graph. The numbered circles on the model screen
shot indicate the sequence of fixations produced
by the model. The pattern of fixations in both
screenshots is remarkably similar.

Modelling Icon Search

Fleetwood and Byrnes study of icon search
(2002, in press) is a another demonstration of
how an ACT-R cognitive model can provide a
detailed account of the cognitive and perceptual
processes involved in a common HCI task that
closely matches peoples response times (RTs) and

Figure 5. Screen shots showing an experimental participants eye movement data (left) and the ACT-R
models visual attention scan path (right) for the QVQV question oil = 6, gas = ? using the 1980s
parametric graph

Note: In the model screen shot, numbered circles on the scan path indicate the location and sequence of fixations.

1166

Modelling Interactive Behaviour with a Rational Cognitive Architecture

patterns of eye movements. Fleetwood and Byrnes

more of these were shared by several other icons
model differs from that of Peebles
and Cheng
in the group. From the visual search literature,
304 Peebles & Cox
(2003) in that it incorporates eye movements and
it can be predicted that search time will increase
movement of attention (EMMA) (Salvucci, 2001),
as icon distinctiveness decreases. An additional
Modelling
Iconfactor
Search
a computational model of the relationshipFleetwood
between and Byrnes
in Fleetwood and Byrnes (2006) study also
study of icon search (2002, in press) is a another demonstraeye movements and visual attention.
affectcan
search
time
(at leastaccount
for certain
tionEMMA
of how can
an ACT-R known
cognitivetomodel
provide
a detailed
of the cognitive and
be easily integrated into the ACT-Rperceptual
architecture,
stimuli)
is
the
number
of
distractors
in
the
display,
processes involved in a common HCI task that closely matches
peoples response
times (RTs)
and patterns
of eye
movements.
Fleetwoodwith
and Byrnes
modelof
differs from that
allowing models to make more detailed
predicwith
search
time increasing
the number
of Peebles
and Cheng distractors
in that it incorporates
eyeset.
movements
and movement of attention
tions of actual eye movements, rather
than simple
in the search
In their experiment,
(EMMA) (Salvucci, 2001),
a computational
model
of the relationship
between
shifts of visual attention.
Fleetwood
and Byrne
had search
sets of 6, 12,
18 eye movements and visual attention. EMMA can be easily integrated into the ACT-R architecture,
One of the main aims of Fleetwood
and Byrnes
and 24 icons.
allowing models to make more detailed predictions of actual eye movements, rather than
research is to investigate the notionsimple
of iconshifts
qualIn the experiment, participants were required
of visual attention.
ity (defined in terms of an icons distinctiveness
to
find,
as rapidlyand
as Byrnes
possible,
different
One of the main aims of Fleetwood
study
was to quality
investigate the notion
of icon
in terms
an icons
and visual
complexity) and
and visual complexity) and to examine
thequality
effect (defined
target
iconsof in
searchdistinctiveness
sets of differing
sizes.
examine
the effect that
quality icon
may and
havefile
on identification
that differences in quality may havetoon
identificaOn differences
each trial, in
a target
name were performance.
Theyclasses
created of
three classes
of iconfollowed
(examples1500
of which
are shown
in Figure 6). Good
tion performance. They created three
presented
ms later
by a button
quality icons were designed to be easily distinguishable from others based on the primitive
icon (examples of which are shown
in Figure 6).
labelled Ready for the participant to click when
features of colour and shape. All icons in this set were a combination of one colour (from
Good quality icons were designed
beone
easily
he/she
six)toand
shape (from
two).felt ready to continue. When this button
distinguishable from others based on theIn
primitive
was
clicked,
target
icon was
by theonly by a relacontrast, poor quality
iconsthe
were
designed
to bereplaced
distinguishable
tivelyin
careful
inspection
but to set
be relatively
indistinguishable
in a large
set. These
features of colour and shape. All icons
this set
search
and the participant
had simply
to distractor
look
icons were
of target
the same
basic
(a combination
of black,
were a combination of one colour poor
(fromquality
six) and
forallthe
icon
andshape
clickand
on colour
it as quickly
as
white,
and
shades
of
grey).
An
intermediate
class
of
fair
quality
icons
was
also
designed
one shape (from two).
possible; when an icon was clicked upon, the
with shapes more distinctive than the poor quality icons but more complex than the good
In contrast, poor quality iconsquality
were designed
next trial started. Participants completed a total
icons, and with the same range of greyscale colours as the poor quality icons. The
to be distinguishable only by a relatively
careful
of 144 trials,
allalevels
of the
search structure
set
main effect
of the manipulation
wasinvolving
to produce
different
similarity
for each
inspection but to be relatively indistinguishable
and
icon
quality
variables,
and
on
each
trial
the
class of icons. Good quality icons could be identified as a single combination of features,
example,
yellow triangle.
In contrast,
qualitybetween
icons were
defined
in a large distractor set. These poorforquality
icons
participants
RT (thefair
duration
clicks
on by more than
(typically
foricon
example:
grey rectangle;
were all of the same basic shapeone
andcombination
colour (a of features
the Ready
buttonthree,
and an
in the search
set) was black square;
black diagonal-right),
some of which
were shared
other icons.
In the poor quality
combination of black, white, and shades
of grey).
recorded.
The results
of thewith
experiment
(shown
group, icons were defined by an average of four feature combinations and many more of
An intermediate class of fair quality
icons was
in Figure 7) revealed that, as predicted, both icon
these were shared by several other icons in the group. From the visual search literature, it
also designed with shapes more distinctive
than that search
quality
andwill
search
set size
haddistinctiveness
a significant effect
can be predicted
time
increase
as icon
decreases. An addithe poor quality icons but more complex
than
the
on
search
time.
tional factor in Fleetwood and Byrnes study also known to affect search time (at least for
stimuli)
good quality icons, and with the certain
same range
ofis the number of distractors in the display, with search time increasing with
the number
of distractors in the search set. In their experiment, Fleetwood and Byrne had
greyscale colours as the poor quality
icons. The
search
sets of 6, 12, 18 and 24 icons.
main effect of the manipulation was
to produce
Figure 6. Examples of icons of good, fair, and
a different similarity structure for each class of
poor quality used in the experiment of Fleetwood
icons. Good quality icons could be
identified
as
Figure
6. Examples
of icons
of good,
and poor quality used in the experiment of Fleetand Byrne
(infair,
press)
and Byrne (in press)
a single combination of features,wood
for example,
yellow triangle. In contrast, fair quality icons
were defined by more than one combination
of features (typically three, for example: grey
Fair - No Border
Poor - Square
Good - Circle
rectangle; black square; black diagonal-right),
some of which were shared with other icons. In
the poor quality group, icons were defined by an
Good - No Border Poor - Circle
Fair - Square
average of four feature combinations and many

Copyright 2007, Idea Group Inc. Copying or distributing in print or electronic forms without written permission of
Idea Group Inc. is prohibited.

1167

Modelling Interactive Behaviour with a Rational Cognitive Architecture

To provide an explanation of their data, Fleetwood and Byrne (2006) produced an ACT-R
model of the task that was able to interact with
the same experiment software as the participants.
As described previously, each experiment trial is
comprised of two stages, the first where the target
icon and its file name are encoded and the second
in which it is sought. The model has a set of seven
productions to carry out the first stage: (1) locate
the target icon and (2) encode an attribute pair (e.g.,
grey rectangle), (3) look below the icon and (4)
encode the associated file name, and finally (5)
locate and (6) click on the Ready button. In the
second stage, the model locates and attends to an
icon with the previously encoded target feature
and then shifts visual attention to the file name
below it. If the file name matches the target file
name, visual attention is returned to the icon and
the mouse clicks on it. If the file name is not the
target, however, the model continues the search
by locating another icon at random with the same
target features. This sequence of events requires
four productions and takes 285 ms to complete.
Figure 7 reveals a close correspondence
between the mean RTs produced by the model
and those of the experiment participants (R 2 =
.98, RMSE = 126ms) and shows that an ACT-R

model based on the similarity structure of the

search set and the strategy of identifying a single
combination of features and random search can
provide a reasonable account of the data. However,
Byrne, Anderson, Douglass, and Matessa (1999)
had shown in an earlier study of visual search in
a menu selection task that alternative strategies
can produce similar aggregate RTs, necessitating
the incorporation of eye movement data to add
further constraints on the proposed theory. As a
result, Fleetwood and Byrne (2006) carried out
an eye movement study to test their model further and found two major discrepancies between
the observed eye movements and the patterns of
visual attention produced by their model. First,
they found that, although the model successfully
reproduced the patterns of visual attention across
the icon quality and set size conditions, for all
conditions the number of saccades per trial produced by the model was significantly greater than
those recorded in the experiment. Second, when
analysing the eye movement data, Fleetwood and
Byrne found that patterns of icon search were
not random as their model predicted, but were
systematic, in the sense that participants sought
to minimise the distance between successive
fixations, typically looking at target icons closest

Figure 7. Response time by set size and icon quality for Fleetwood and Byrnes (in press) revised model
and the experiment data
4000
Data-P oor

3500

R es pons e Time (millis econds )

Data-F air
3000

Data-G ood
Model- P oor

2500

Model- Fair
Model- G ood

2000
1500
1000
500
0

1168

12

S et S ize

18

24

Modelling Interactive Behaviour with a Rational Cognitive Architecture

to their current fixation point. This produced a

search pattern that revealed a systematic scanning
of areas of the display.
Both of the discrepancies between the model
and human data are explained by Salvuccis (2001)
EMMA model. It is been demonstrated previously
that the relationship between eye movements and
visual attention is not direct, and that people often
do not move their eyes to their focus of attention
(e.g., Henderson, 1992; Rayner, 1995). EMMA
attempts to capture this relationship by providing
an account of if and when eye movements occur,
and if they do occur, the location of their landing
relative to their targets. Integrating EMMA into
ACT-R allows models to simulate actual eye movements rather than just visual attention shifts and
provides a more realistic output to be compared
with human eye movement data. In addition,
EMMA predicts that efficient search strategies
minimise average saccade distance, resulting in
search patterns in which objects nearest to the
current fixation point are examined soonest.
Fleetwood and Byrne (2006) modified their
models search strategy according to the EMMA
account and incorporated EMMAs eye movement computations into their model, resulting

in a greatly improved fit (shown in Figure 8) to

the human eye movement data (R 2 = .99, RMSE
= 0.58).

Conclusion
In this chapter we have presented a number
of recent examples of research that we believe
clearly demonstrate the value of rational analysis
and cognitive modelling in the study of complex
interactive behaviour. Such tasks typically involve
the complex interaction of three elements: (1) the
perceptual and cognitive abilities of the user; (2)
the visual and statistical properties of the task
environment; and (3) the specific requirements
of the task being carried out. The use of rational
analysis and an embodied cognitive architecture
such as ACT-R allows all three of these elements
to be brought together in an integrated theoretical
account of user behaviour. Rational analysis provides a set of assumptions and methods that allow
researchers to understand user behaviour in terms
of the statistical structure of the task environment
and the users goal of optimising (i.e., reducing the
cost/benefit ratio of) the interaction. Developing

Figure 8. Mean number of shifts of visual attention per trial made by Fleetwood and Byrnes (in press)
revised model relative to the mean number of gazes per trial made by participants
Avera ge G azes / S hifts of V is ua l A ttention per T ria l

Data-P oor
Data-Fa ir

Data-G ood
6

Model-P oor
Model-F air

Model-G ood

4
3
2
1
0

12

S et S ize

18

24

1169

Modelling Interactive Behaviour with a Rational Cognitive Architecture

cognitive models of interactive behaviour in a cognitive architecture such as ACT-R allows researchers to specify precisely the cognitive factors (e.g.,
domain knowledge, problem-solving strategies,
and working memory capacity) involved. In addition, the recent incorporation of perceptual-motor
modules to cognitive architectures allows them
to make predictions about users eye movements
during the entire performance of the task, which
can be compared to observed eye movement data
a highly stringent test of the sufficiency and
efficacy of a model. The use of these methods has
increased rapidly over the last 5 years, as has the
range of task interfaces being studied. Although
we are still a long way from achieving the goal of
an artificial user that can be applied off the shelf
to novel tasks and environments, the models of
interactive behaviour described here demonstrate
a level of sophistication and rigour still relatively
rare in HCI research. As these examples illustrate,
developing more detailed accounts of interactive
behaviour can provide genuine insights into the
complex interplay of factors that affect the use
of computer and Web technologies, which may
inform the design of systems more adapted to
their users.

Note
All correspondence to: David Peebles, Department
of Behavioural Sciences, University of Huddersfield, Queensgate, Huddersfield, HD1 3DH, UK;
[email protected].

References
Anderson, J. R. (1990). The adaptive character
of thought. Hillsdale, NJ: Lawrence Erlbaum
Associates.
Anderson, J. R., Bothell, D., Byrne, M. D.,
Douglass, S., Lebiere, C., & Qin, Y. (2004). An

1170

integrated theory of the mind. Psychological

Review, 111(4), 1036-1060.
Anderson, J. R., & Lebiere, C. (1998). The atomic
components of thought. Mahwah, NJ: Lawrence
Erlbaum Associates.
Brumby, D. P., & Howes, A. (2004, July 30-August
1). Good enough but Ill just check: Web-page
search as attentional refocusing. In Proceedings
of the 6th International Conference on Cognitive
Modelling, Pittsburgh, PA.
Byrne, M. D. (2001). ACT-R/PM and menu
selection: Applying a cognitive architecture to
HCI. International Journal of Human-Computer
Studies, 55, 41-84.
Byrne, M. D., & Anderson, J. R. (1998). Perception
and action. In J. R. Anderson & C. Lebiere (Eds.),
The atomic components of thought, (pp. 167-200).
Mahwah, NJ: Lawrence Erlbaum Associates.
Byrne, M. D., Anderson, J. R., Douglass, S., &
Matessa, M. (1999). Eye tracking the visual search
of click-down menus. In Proceedings of the ACM
CHI99 Conference on Human Factors in Computing Systems (pp. 402-409). New York: ACM.
Card, S. K., Moran, T. P., & Newell, A. (1983).
The psychology of human-computer interaction.
Hillsdale, NJ: Erlbaum.
Chong, R. S., & Laird, J. E. (1997).
Identifying
dual-task executive process knowledge using
EPIC-Soar. In Proceedings of the 19th Annual
Conference of the Cognitive Science Society
(pp. 107-112). Mahwah, NJ: Lawrence Erlbaum
Associates.
Cox, A. L., & Young, R. M. (2004). A rational
model of the effect of information scent on the
exploration of menus. In Proceedings of 6th International Conference on Cognitive Modeling:
Integrating Models, (pp. 82-86). Mahwah, NJ:
Lawrence Erlbaum Associates.

Modelling Interactive Behaviour with a Rational Cognitive Architecture

Fitts, P. M. (1954). The information capacity of

the human motor systems in controlling the amplitude of movement. Journal of Experimental
Psychology, 47, 381-391.
Fleetwood, M. D., & Byrne, M. D. (2002). Modeling icon search in ACT-R/PM. Cognitive Systems
Research, 3, 25-33.
Fleetwood, M. D., & Byrne, M. D. (2006). Modeling the visual search of displays: A revised
ACT-R/PM model of icon search based on eye
tracking data. Human Computer Interaction,
21, 153-197.
Henderson, J. M. (1992). Visual attention and eye
movement control during reading and picture
viewing. In K. Rayner (Ed.), Eye movements and
visual cognition: Scene perception and reading,
(pp. 261-283). New York: Springer-Verlag.
John, B. E., & Kieras, D. E. (1994). The GOMS
family of analysis techniques: Tools for design
and evaluation (Tech. Rep. No. CMUHCII94106).
Pittsburgh, PA: Carnegie Mellon University, Human-Computer Interaction Institute.
Kieras, D. E., & Meyer, D. E. (1997). An

overview of the EPIC architecture for cognition and

performance with application to human-computer
interaction. Human-Computer Interaction, 12,
391-438.
Laird, J. E., Newell, A., & Rosenbloom, P. S.
(1987). SOAR: An architecture for general intelligence. Artificial Intelligence, 33, 1-64.
Lohse, G. L. (1993). A cognitive model for understanding graphical perception. Human-Computer
Interaction, 8, 353-388.
Neth, H., Sims, C. R., Veksler, V. D., & Gray, W.
D. (2004). You cant play straight TRACS and
win: Memory updates in a dynamic task environment. In K. D. Forbus, D. Gentner, & T. Regier
(Eds.), Proceedings of the 26th Annual Meeting
of the Cognitive Science Society (pp. 1017-1022).
Hillsdale, NJ: Lawrence Erlbaum Associates.

Newell, A. (1990) Unified theories of cognition.

Cambridge, MA: Harvard University Press.
Oaksford, M., & Chater, N. (Eds.). (1998). Rational models of cognition. UK: Oxford University
Press.
Olson, J. S., & Olson, G. M. (1990). The growth of
cognitive modeling in human-computer interaction since GOMS. Human Computer Interaction,
5, 221-265.
Peebles, D., & Cheng, P. C.-H. (2001). Graphbased reasoning: From task analysis to cognitive
explanation. In Proceedings of the 23rd Annual
Conference of the Cognitive Science Society
(pp. 762-767). Mahwah, NJ: Lawrence Erlbaum
Associates.
Peebles, D., & Cheng, P. C.-H. (2002). Extending
task analytic models of graph-based reasoning:
A cognitive model of problem solving with Cartesian graphs in ACT-R/PM. Cognitive Systems
Research, 3, 77-86.
Peebles, D., & Cheng, P. C.-H. (2003). Modeling
the effect of task and graphical representation on
response latency in a graph reading task. Human
Factors, 45, 28-46.
Peebles, D., Cheng, P. C.-H., & Shadbolt, N. (1999).
Multiple processes in graph-based reasoning. In
Proceedings of the 21st Annual Conference of the
Cognitive Science Society (pp. 531-536). Mahwah,
NJ: Lawrence Erlbaum Associates.
Pirolli, P. (2005) Rational analyses of information foraging on the Web. Cognitive Science, 29,
343-374.
Pirolli, P. L., & Card, S. K. (1999). Information foraging. Psychological Review, 106(4), 643-675.
Pirolli, P., & Fu, W.-T. (2003, June 22-26). SNIFACT: A model of information foraging on the
World Wide Web. Paper presented at the Ninth
International Conference on User Modeling,
Johnstown, PA.

1171

Modelling Interactive Behaviour with a Rational Cognitive Architecture

Rayner, K. (1995). Eye movements and cognitive

processes in reading, visual search, and scene
perception. In J. M. Findlay, R. Walker, & R.
W. Kentridge (Eds.), Eye movement research:
Mechanisms, processes, and applications, (pp.
3-22)
. New York: Elsevier.
Ritter, F. E., Baxter, G. D., Jones, G., & Young,
R. M. (2000). Supporting cognitive models as
users. ACM Transactions on Computer-Human
Interaction, 7, 141-173.
Salvucci, D. D. (2001). An integrated model of
eye movements and visual encoding. Cognitive
Systems Research, 1(4), 201-220.
Salvucci, D. D., & Macuga, K. L. (2002). Predicting the effects of cellular-phone dialling on
driver performance. Cognitive Systems Research,
3, 95-102.

Silva, M., & Cox, A. L. (2005, August 31-September 2) Eye-movement behaviour in interactive
menu search: Evidence for rational analysis.
Paper presented at the BPS Cognitive Section
Conference 2005, University of Leeds.
St. Amant, R., Horton, T. E., & Ritter F. E. (2004).
Model-based evaluation of cell phone menu interaction. In Proceedings of the CHI04 Conference
on Human Factors in Computer Systems (pp.
343-350). New York: ACM.
Young, R. M. (1998). Rational analysis of exploratory choice. In M. Oaksford & N. Chater (Eds.),
Rational models of cognition (pp. 469-500). UK:
Oxford University Press.
Young, R. M., Green, T. R. G., & Simon, T. (1989).
Programmable user models for predictive evaluation of interface designs. In Proceedings of CHI
89: Human Factors in Computing Systems (pp.
15-19). AMC Press.

This work was previously published in Human Computer Interaction Research in Web Design and Evaluation, edited by P.
Zaphiris, pp. 290-309, copyright 2007 by Information Science Publishing (an imprint of IGI Global).

1172

1173

Chapter 3.11

Device Localization in
Ubiquitous Computing
Environments
Rui Huang
University of Texas at Arlington, USA
Gergely V. Zruba
University of Texas at Arlington, USA
Sajal Das
University of Texas at Arlington, USA

Abstract
In this chapter, we will study the localization
problem in ubiquitous computing environments.
In general, localization refers to the problem of
obtaining (semi-) accurate physical location of the
devices in a dynamic environment in which only
a small subset of the devices know their exact
location. Using localization techniques, other
devices can indirectly derive their own location
by means of some measurement data such as
distance and angle to their neighbors. Localization is now regarded as an enabling technology
for ubiquitous computing environments because
it can substantially increase the performance of
other fundamental tasks such as routing, energy

conservation, and network security. Localization

is also a difficult problem because it is computationally intractable. Furthermore, it has to be
implemented in a highly dynamic and distributed
environment in which measurement data is often
subject to noise. In this chapter, we will give an
overview of localization in terms of its common
applications, its hardware capacities, its algorithms, and its computational complexity.

Introduction
In a ubiquitous computing environment, devices
are often connected to one another on the fly to
form an infrastructure-less network that is fre-

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Device Localization in Ubiquitous Computing Environments

quently referred to as a mobile ad hoc network

(MANET). Since MANET serves as an abstract
model that can be seen as a superset of diverse
sub-areas such as sensor networks, mesh networks
or an enabler for pervasive computing, it has attracted significant research interests in the past
several years. A major advantage of MANETs over
regular wired or wireless networks is their infrastructure-less nature, as MANETs can potentially
be deployed more rapidly and less expensively
than infrastructure-based networks. However,
the lack of an underlying explicit infrastructure
also becomes a major disadvantage in adapting
MANETs to a wider array of applications, since
existing network algorithms and protocols are
often not plug-in solutions for such dynamic
networks. New algorithms need to be designed
for fundamental network tasks such as addressing, topology discovery, and routing.
Location discovery is emerging as one of the
more important tasks as it has been observed that
(semi-) accurate location information can greatly
improve the performance of other MANET tasks
such as routing, energy conservation, or maintaining network security. For instance, algorithms
such as location aided routing (LAR)(Ko, 2000),
GRID(Liao, 2001), and GOAFR+(Kuhn, 2003)
rely on location information to provide more
stable routes during unicast route discovery.
The availability of location information is also
required for geocast (multicast based on geographic information(Jiang, 2002)) algorithms
such as location-based multicast (LBM)(Ko,
1999), GeoGRID(Liao, 2000) and position-based
multicast (PBM)(Mauve, 2003). To minimize
power consumption, the geographical adaptive
fidelity(GAF) algorithm(Xu, 2001) uses location
information to effectively modify the network
density by turning off certain nodes at particular
instances. Furthermore, in (Hu, 2003), the authors have shown that wormhole attacks can be
effectively prevented when location information
is available. As more algorithms are being proposed to exploit location information of devices in

1174

ubiquitous networks, it is clear that obtaining such

information efficiently and accurately becomes
of great importance.
A direct way of obtaining location information is to install global positioning system (GPS)
receivers on devices. However, this is currently
impractical as GPS receivers are still relatively
expensive, power-hungry, and require clear line
of sight (i.e., making indoor usage impossible)
to several earth-bound satellites. In ubiquitous
environments (e.g., sensor networks), devices
are imagined as small as possible and operating
on a very restricted power source, thus, it may
not be feasible to install GPS receivers onto all
sensor nodes. Localization in MANET refers to
the problem of finding the locations of those nonGPS enabled nodes based on limited information
such as the locations of some known beacons
(also referred to as anchors) and measurements
such as ranges or angles among the neighbors.
In this chapter, we will study why localization is
important and how it can be accomplished within
the ubiquitous computing environment.

Background
The localization problem is hard for a number
of reasons:
1

Geometric limitations: To pinpoint its exact

location in 2-D, a node needs to know the
locations and its distances to at least three
beacons. Alternatively, nodes could calculate their own location based on a distance
and an (absolute) angle measurement from
one beacon. However, even if obtaining
such measurements were possible and the
measurements were exact, guaranteeing that
the right number of beacons surround each
node is often impossible since MANETs may
be randomly deployed and that in general
only a small percentage of nodes are indeed
beacons. Thus, localization algorithms often

Device Localization in Ubiquitous Computing Environments

need to take advantage of multi-hop information, that is, estimating node locations
based on other nodes location estimates.
Availability of measurements: For localization algorithms that require distance or angle
measurements, certain sensory devices will
need to be available to provide such readings.
However, it is likely that not all nodes have
the same sensory capacity. In other words,
there is a need for the localization algorithm
to work in a heterogeneous environment,
in which devices with different location
sensory capacities coexist.
Measurement error and error propagation: Even when measurement devices are
available, there is a general consensus that
those measurements are prone to errors. For
instance, a distance measurement can be
derived based on a received signal strength
indication (RSSI) reading, in which a receiving device measures the strength of the
signal from a sending device and obtains the
distance via an estimated signal propagation
model. RSSI reading is prone to multi-path
fading and far field scattering. The error
can be especially high when there are a
significant number of obstacles in between
the sender and the receiver. Since most
localization algorithms require measurements from nodes several hops away, the
measurement error is likely to aggregate
along the path and eventually completely
throw off the location estimate.

Despite the difficulties listed above, there is

an increasing amount research effort spent on
the localization problem. The amount of effort
is justified because localization is considered an
enabling technology that needs to be resolved
with the best possible outcome upon which other
location-dependent technologies for MANETs
can be successfully employed. Researchers have
been working on problem in both hardware (i.e.,
improving measurement accuracy of devices)
and software (i.e., improving the localization

algorithm design). This chapter covers the latest

advances in this field, including the following
topics:
1.

2.

3.

4.

5.

Applications of localization: In this section,

we will establish the need for better localization techniques by surveying a number
of proposed algorithms for MANETs that
rely on localization.
Measurement types for localization: In this
section, we will cover the latest advances
in hardware design that enables localization
on the smaller devices commonly seen in
the ubiquitous computing environment,
including the devices that measure distance
ranging, angle of arrival (AoA), and interferometric ranging.
Survey of localization algorithms: We
will survey some of the most popular localization algorithms, including those using connectivity information, ranging, and
angle information. We will study the pros
and cons of each algorithm, and suggest
their appropriate applications in ubiquitous
computing.
Localization theory: We will cover the
theoretic basis of localization techniques. We
will study the necessary and sufficient conditions for a network to be localized based on
the latest results from graph theory. We will
show that the localization problem in general
is NP-Complete. We will also introduce the
reader to the Cramer Rao Bound (CRB) that
is often used to analyze the hardness of different localization scenarios.
Future directions: We will look into a
number of promising future directions for
the localization techniques.

Applications of Localization
There have been numerous algorithms proposed
for MANETs that rely on localization data; in
this section, we provide a brief survey of these
1175

Device Localization in Ubiquitous Computing Environments

algorithms. We will consider algorithms in four

categories based on their functionalities: unicast
routing; multicast routing; energy consideration;
and network security.

Unicast Routing
Routing refers to the task of finding the correct
route from a sending device (source) to a receiving device (destination). Routing is an especially
challenging task for MANETs because their
frequent topology change implies the underlying
instability of any established routes. As such,
routes are needed to be frequently rediscovered,
reestablished, and repaired. In general, routing
(i.e., route discovery and repair) involves flooding the routing control packets throughout the
network. Flooding can often be expensive in
terms of delay and bandwidth usage it incurs,
both of which can greatly affect the network
performance. Thus, there is a strong incentive to
design efficient routing algorithms that minimize
the overhead caused by any unnecessary packet
flooding. Unicast routing based on location information, often called geometric routing or location
based routing, has shown to be one of the viable
solutions to this problem.
Location-aided routing (LAR) (Ko, 2000)
protocol is the first MANET routing algorithm
proposed that uses location data. In LAR, every
node is assumed to know its own location, and each
individual location is then periodically broadcast
throughout the network. Thus, at any time t, every
node knows the locations of any other nodes at
some previous time <t. Based on this location
information and an estimated velocity, a node
can derive an estimated location range, called
expected zone, of a target node at the current
time. Instead of flooding the entire network, the
routing request packets can be directed to search
for the target node only at this expected zone.
Global flooding is performed only after the location based routing request has failed. Limiting
route discovery to a smaller expected zone with

1176

LAR reduces the number of routing requests

compared to the standard flooding scheme.
GRID (Liao, 2001) protocol uses location
information as a way to form geographical clusters within the network. Based on node locations
and their residency within a pre-determined grid
system, nodes within the same grid block are
grouped into a cluster. A cluster head or gateway in (Liao, 2001) is then selected for each
grid block. The cluster head is responsible for
servicing the routing packets. Furthermore, the
cluster head can monitor the status of existing
routes and reroute packets as deemed necessary.
Since the cluster formation effectively simplifies
the network topology, the routing overhead is
reduced. A critical requirement of forming such
geographical-based clusters is the availability of
node location information.
In (Kuhn, 2003), the authors provided a theoretical bound to the geometric routing problem
and proposed an algorithm called GOAFR+.
Assuming that node locations are known using
some localization technique, GOAFR+ first tries
to greedily route the packet by forwarding it to
the neighbor located closest to the destination.
However, such greedy selection does not guarantee message delivery since the intermediate
node closest to the destination might not have a
route to it. In such cases, GOAFR+ explores the
boundaries of the faces of a planarized network
graph by employing the local right hand rule (i.e.,
always turn right) to escape the local minimum.
This method of escaping local minima is also
called parameter routing, which is used in a
number of other location based routing protocols
as well.
In terms of performance, simulations performed by (Ko, 2000) and (Liao, 2001) have
shown up to 50% of reduction in routing packets
when using geographic routing compared to
standard flooding. Since the overhead of flooding
is proportional to network density, it has been
observed that the amount of performance gain
becomes more significant when network density

Device Localization in Ubiquitous Computing Environments

is increased. Furthermore, although the routing

performance is impacted by the localization
error, such impact is observed to be minimal.
This indicates that in the case of routing, highly
precise location data is not required. After all,
location data is used by routing algorithms to
give a direction that guides the routing packets;
imprecise location data can still be used as long
as the general direction is valid.

Multicast Routing
Similar to unicast routing, multicast routing can
also benefit from location data. Multicast routing
using geographic information is often referred to
in the literature as geocast routing. The LocationBased Multicast (LBM) algorithm (Ko, 1999) is a
multicast extension to the unicast Location-Aided
Routing (LAR). Like LAR, which forwards the
routing requests according to the location of the
destination node, LBM forwards the requests according to the direction of the geocast region that
contains all the multicast destinations. GeoGRID
(Liao, 2000) is the multicast extension to GRID
(Liao, 2001). Like in GRID, location information
is used by GeoGRID to identify the grid block
where nodes reside. Multicast is done through the
gateway node selected at each grid block. Based
on the location of the source node and the geocast
region, LBM and GeoGRID define a forwarding region that contains the intermediate nodes
responsible for forwarding packets. The size
and shape of the forwarding region have a direct
impact on the overall performance; shapes such
as rectangles and cones have been proposed in
(Ko, 1999).
While the standard shapes such as rectangles
and cones work well in most cases, there are
situations where viable routes exist only outside
the forwarding region. For instance, a network
can be partitioned into two sub-networks connected only through a narrow linkage due to
some obstacles (e.g., two islands connected by a
bridge). When the source and the destination are

in separate partitions, a geometrically defined

forwarding region is unlikely to cover the linkage. To prevent routing failure in such a case, a
routing zone based on Voronoi diagrams was
proposed in (Stojmenovic, 2006), which partitions
the network graph based on the proximity of the
nodes. Again, the proximity information relies
on localization information.
The Position-Based Multicast (PBM) protocol
proposed in (Mauve, 2003) attempts to optimize
the multicast tree it generates by minimizing the
overall path length and the overall bandwidth
usage; two often contradictory objectives. To
minimize the overall path length, PMB takes a
greedy approach using location information. At
each intermediate node, packets are forwarded to
a set of neighbors based on their overall distances
to the multicast destinations. In particular, a set
of the neighbors with the minimum overall distance to every destination is selected as the next
set of forwarding nodes. To take account of the
bandwidth usage, the greedy selection also tries
to minimize the size of the forwarding set. PBM
also uses parameter routing to deal with local
minima. Both greedy routing and parameter
routing employed by PBM rely on the location
information.

Power Management
MANET is often used as the model for sensor
networks, an emerging technology for pervasive
computing. One of the major challenges of sensor
networks is power management. Since sensors are
commonly small in size and are battery powered,
conserving energy would prolong their service
time and, thus, the lifespan of the entire network.
The Geographical Adaptive Fidelity (GAF) algorithm (Xu, 2001) is a network topology management algorithm with reduced energy consumption
as its primary objective. The idea behind GAF is
that there are often a large number of nodes that
are redundant during packet routing in MANETs.
If the redundant nodes can be identified, they

1177

Device Localization in Ubiquitous Computing Environments

can then turn off their radio to save energy. For

GAF, the identification of redundant nodes is
accomplished by analyzing the relative location
information among the neighboring nodes. More
specifically, GAF divides the network into virtual
grids such that all nodes in grid block A are the
neighbors of all nodes in grid block B. This way,
all nodes within the same virtual grid block can be
considered equivalent. To conserve energy during
packet routing, GAF only turns on the radio for one
of the nodes in each grid block. The active node
is periodically round-robinned to achieve loadbalancing. Analysis and simulations performed
in (Xu, 2001) show that GAF can reduce overall
energy consumption by 40% to 60%.

Security
In (Hu, 2003) the authors proposed a technique
called packet leashes to defend against wormhole attacks in MANETs. A wormhole attack
is a type of security breach where an adversary
intercepts incoming packets and tunnels them to
another part of the network via a single long-range
directional wireless link or through a direct wired
link. From there, the adversary can retransmit
the packets to the network. Note that this type of
capture-and-retransmit attack can be immune
to common packet encryption methods, since the
adversary does not need to read the packet content.
Wormhole attacks can severely disrupt ad hoc
routing protocols such as Ad hoc On-Demand Distance Vector Routing (AODV) or Dynamic Source
Routing (DSR), and cause a denial of service to
the network. The core of packet leashes is based
on two assumptions: i) all nodes know their own
locations; and ii) all nodes are synchronized. To
enable packet leashes, the sender node encloses
its location and transmission time-stamp within
the packet. At the receiver node, the packet leash
is validated against the receivers own location
and clock. In particular, the sender location information gives the distance from the original
sender to the receiver, and the time-stamp gives

1178

the transmission duration of the packet. Based

on the transmission duration and signal propagation model, factored in some error tolerance, the
receiver can validate the estimated distance the
packet has traveled against the true distance to
determine whether the packet is indeed coming
from the original sender or an imposer at some
other location. Thus, the location information
and time-stamp provide a virtual leash to limit
the effective range of the packet so that it cannot
be exploited by wormhole attackers.
From the previous discussion on the locationdependent algorithms that encompass a wide range
of problem domains, it is clear that providing location information (i.e., localization) to MANETs
is becoming an increasingly important task. In
fact, localization is now widely regarded as an
enabling technology for MANETs that needs
to be addressed before other location-dependent
techniques can be realized in the real world
(Patwari, 2003).

Measurement Types for Localization

In this section, we study a number of measurement
types provided by onboard hardware devices that
enable localization in MANETs for ubiquitous
computing environments. A Global Positioning
System (GPS) (Parkinson, 1996) receiver can
provide the absolute location. However, its cost,
size, and power requirement prevent it from being
installed at every network node. How do nodes
not equipped with GPS obtain their location
information then? They have to rely on sensory
measurements provided by alternative hardware
devices. There are five general types of measurements: i) connectivity only; ii) RSSI (radio signal
strength indicator) ranging; iii) ToA (time of arrival) ranging; iv) AoA (angle of arrival), and v)
interferometric ranging. We will describe each
of their capacity, usage and mathematical models
when applied to the localization problem.

Device Localization in Ubiquitous Computing Environments

Connectivity Only
Measurement
At a minimum, a node can detect connectivity
to its immediate neighbors, that is, its one-hop
neighborhood. The connectivity only measurement is a binary reading between two nodes of
either true or false indicating whether they
are neighbors. Based on this connectivity information, one can derive the general proximity of the
nodes as a way to localize the network.

RSSI Ranging Measurement

A node can be localized using multilateration
(Niculescu, 2001) if the distances (i.e., the ranges)
to three or more known locations are obtained.
The distances can be obtained, for example, by
measuring RSSI (radio signal strength indicator)
or ToA (time of arrival). In RSSI, the receiver
measures the received signal strength and compares it with the transmitted signal strength. The
difference (in dB) is then applied to the inverse
of the signal propagation model to provide a
distance estimate. Sensors that measure RSSI
are widely available to mobile devices. Indeed,
most off-the-shelf technologies implicitly provide
such information (e.g., most WiFi, Bluetooth,
and IEEE802.15.4 chipsets do). The drawback
of RSSI-based measurements is that they can be
very inaccurate because an exact model of the
propagation environment is often unavailable.
Experiments in (Savvides, 2001) have shown that
when no obstacle exists between the sender and
the receiver, RSSI can provide a distance estimate
of accuracy within a few meters. However, in a
less than ideal environment, the result is often
unpredictable. Furthermore, low cost RSSI receivers are often variable in their transmission power
due to the lack of calibration.
In the outdoor environment with a minimum
of obstacles, signal propagation decay is proporn
tional to d p, where d is the distance the signal
has traveled, and p is an environment-dependent

path loss exponent. However, in the actual environment where obstacles exist, multipath signals
and shadowing become two major sources of
noise that impact the actual RSSI. In general,
those noises are commonly modeled as a random
process during localization. Let Pi, j be the RSSI
(in dB) obtained at the receiver node j from the
sender node i. Pi, j is commonly modeled as a
Normal distribution (Patwari, 2003)
2
Pi , j = N ( Pi , j , dB
)

(1)

where Pi, j is the mean power in dB and 2

dB
is the variance caused by noise factor such as
shadowing. Pi, j is further defined as the power
reduction from a reference location:
Pi , j = P0 10n p log10 (di , j / d 0 )

(2)

where P0 is the power at a reference location at

the distance d0(commonly d0 = 1m). p is an environment-dependent path loss exponent that is
assumed to be known from prior measurements
(theoretically p = 2). di,j is the Euclidean distance
between nodes i and j.

ToA Ranging Measurement

Although ToA is used for radio signals in GPS,
it is mostly used in the context of acoustic or ultrasonic signals in inexpensive ToA tracking (as
propagation speeds are five orders of magnitude
less). For instance, the Medusa node in (Savvides,
2001) is an implementation of ToA ranging using
ultrasonic signals. ToA measures the time signals
travel from the sender to the receiver. The distance
between nodes is obtained by multiplying this time
with the signal propagation speed. In spite of the
additive noise and multipath, in general distance
measures based on ToA are more accurate than
RSSI-based measures. However, special acoustic
transceivers have to be employed on each node
and synchronization among the nodes needs to
be established. Clock synchronization algorithms

1179

Device Localization in Ubiquitous Computing Environments

designed for sensor networks that are accurate to

the order of 10s have been reported (Sivrikaya,
2004). As mentioned earlier, ToA may also be used
together with radio signals, but current technology is not mature enough to provide satisfactory
precision over smaller distances inexpensively.
Let i be the sender node and j be the receiver
node, ToA measurement Ti,j is often modeled as
a Normal distribution (Patwari, 2003):
Ti, j

N( di, j / c, 2 )

where di,j is the Euclidean distance between i and

j, c is the signal propagation speed, and is the
variance caused by noise.

AoA Measurement
A node can be localized if the angles between
it and two beacons are known. Thus, the angle
information (i.e., bearing, or angle of arrival
(AoA)) can be used to localize the network. Currently, there is no off-the-self device that offers
AoA sensing capability. However, a number of
prototype devices are available. For instance,
Cricket Compass(Priyantha, 2001) is a small form
device that uses ultrasonic measurements and
fixed beacons to obtain acoustic signal orientations. In(Niculescu, 2004) a rotating directional
antenna is attached to an 801.11b base station; by
measuring the maximum received signal strength,
a median error of can be obtained from the sensor. The challenge here is to design AoA sensing
devices with small form factor and low energy
consumption. In(Chintalapudi, 2004), the authors
outline a solution with a ring of charge-coupled
devices (CCDs) to measure AoA with relatively
low energy consumption.
In general, AoA is also modeled as a Normal
distribution. Let the true angle between the sender
i and j be i, j , the AoA measurement between i
and j is therefore
Ai, j

1180

N( Ai, j , 2 )

where 2 is the angle variance. Theoretical results

A
for acoustic-based AoA estimation show standard
deviation A is between 2 to 6 , depending on
range (Patwari, 2005). RSSI-based AoA method
with A on the order of 3 has been reported in
(Ash, 2004).
o

Interferometric Ranging
Measurement
Interferometric ranging is a widely used technique in both radio and optical astronomy to
determine the precise angular position of celestial
bodies as well as objects on the ground (Kus,
2006). Interferometric ranging exploits the
property that the relative phase offset between
two receivers determines their distances to two
simultaneous senders. Due to the recent advancement in hardware, it is now possible to implement
interferometric ranging sensors in much smaller
form factor to be used for localization (Marti,
2005). By synchronizing the transmission at
the two senders, each of which sends a signal
at a slightly different frequency, the receivers
can derive the relative phase offset of the two
signals by comparing the RSSI readings. The
distance difference (also called the q-range) can
then be calculated from the relative phase offset
with high accuracy. A q-range obtained from
interferometric ranging from two senders A and
B, and two receivers C and D is the distance difference dABCD=dADdBD+dBCdAC+e where e
is the measurement error (Figure 1).
A major advantage of interferometric ranging
is that the measurement could be extremely accurate compared to noise-prone RSSI readings.
In a recent experiment (Marti, 2005), in which
16 nodes are deployed in a 4x4 grid over a 18x18
meter flat grassy area with no obstruction, the
maximum q-range error was shown to be around
0.1 meters while the medium error was less than
0.04 meters. However, interferometric ranging is
more difficult to implement due to the following
reasons.

Device Localization in Ubiquitous Computing Environments

Figure 1. The interferometric ranging measurement of the q-range dABCD=dADdBD+dBCdAC+e.

Here, node A and B are the senders, and node C and D are the receivers.

1.

2.

The measurement can be impacted by various sources of noise such as frequency drift,
ground multipath error, and time synchronization error (Marti, 2005). Frequencies
of the transmissions need to be precisely
calibrated, as any carrier frequency drift
and phase noise would directly impact
the observed phase offset. Precise time
synchronization is needed at the senders of
a q-range. Thus, there will be overhead to
maintain clock synchronization.
A significantly larger number of measurements are required for localization than using
direct ranging techniques. While there are
also a large number of measurements available (O(n4)) even for a small network, only a
small subset of them are independent of each
other. The rest merely provide redundant
information. It has been shown in (Kus,
2006) that the number of independent measurement using interferometric measurements is O(n2), which is significantly higher
than with RSSI and AoA ranging (O(n)).
Considering the localization problem in
relative coordinates, for a network of n nodes
there are 2n-3 unknowns in two dimensions
and 3n-6 unknowns in three dimensions.
This is because the relative coordinates are

3.

invariant under translation, rotation, and

reflection. Thus, in two dimensions, we have
2n-3 degrees of freedom, where translation,
rotation, and reflection each reduce one degree of freedom. Thus, the smallest network
that can be localized using interferometric
measurements is a fully-connected network
with a population of n=6, where there are
9 independent measurements available to
cover 9 unknowns. The large number of
q-ranges available/required indicates a scalability issue for larger networks.
Since each measurement involves four
nodes, more collaboration is required between nodes. Due to the requirement of
synchronized transmission, the senders have
to collaborate in scheduling their transmission. Also, the receivers have to collaborate
to derive the relative phase offset. Such collaboration requires sophisticated protocols
to be implemented in order to reduce the
communication overhead.

Those difficulties rooted in the physical characteristics of interferometric ranging devices

affect the algorithmic design of the localization
algorithm. As we will see in the following section,
the localization algorithms based on interfero-

1181

Device Localization in Ubiquitous Computing Environments

metric ranging measurements tend to be more

difficult to design.
Table 1 summarizes the five measurement
types described in this section.

Localization Algorithms
The previous section introduced the primary types
of measurement that can be used for localization. However, obtaining measurements such as
distance ranging and angle of arrival is only the
first step of localization. To calculate the actual
node location, we will have to mathematically
incorporate those measurement readings to derive
localization algorithms. While there are various
ways of classifying localization algorithms, we
feel it is more logical to classify them according to
the measurement assumptions as follows: i) connectivity-only; ii) range-based; iii) angle-based;
iv) interferometric ranging based; v) hybrid, and
vi) mobility-based.

Connectivity-Based Algorithms
A number of localization methods rely on connectivity information only. These types of methods
are also referred to as range-free methods in
the literature. For instance, the Centroid method
(Bulusu, 2000) estimates the location of an unknown node as the average of its neighboring
beacon locations. Clearly, in order for the location
estimate to be reasonably accurate, a large number
of beacons need to be heard. Thus, to provide
sufficient localization coverage, the Centroid
method requires more powerful beacons with a
large transmission range.
The APIT (Approximated Point-In-Triangulation) method (He, 2003) estimates the node location by isolating the area using various triangles
formed by beacons. For each triangle formed by
three beacons, the node is either in or out of the
triangle. For instance in Figure 2(a), if it can be
determined the node G is inside ABC and DEF, Gs
location can be isolated to the shaded overlapping
area of the two triangles. To determine whether
a node is inside or outside the triangle, APIT

Table 1. Measurement types for localization

Nodes
Connectivity

RSSI

Accuracy

Cost

Measured Value

N/A

Low

Proximity

Low

Distance di, j between node i and

j, derived from
power Pi, j

Ti , j = N (di , j / c, T2 )

Ai , j = N ( Ai , j , 2A )

Low

ToA

High

High

Distance di, j between node i and

j, derived from
time of arrival Ti, j

AoA

Low/
Medium

N/A

Angle Ai, j
between node i
and j,

Math Model
1 if two nodes are connected; 0
otherwise
2
Pi , j = N ( Pi , j , dB
)

Pi , j = P0 10n p log10 (di , j / d 0 )

q-range
Interfero
metric
Ranging

1182

Very High

Medium/
High

q-range (distance
difference between four nodes)

dABCD=dADdBD+dBCdAC+e
between node A, B, C and D. Noise
to each q-range is modeled using a
Normal distribution

Device Localization in Ubiquitous Computing Environments

compares the RSSI readings from the beacons at

the node with those at its neighbors. Intuitively,
smaller RSSI reading means a shorter distance
(i.e., closer to the beacon) and vice versa. If there
does not exist a neighbor that is further from (or
closer to) all beacons simultaneously, then the
node is inside triangle with high probability. For
instance in Figure 2(b), a neighbor of D, E, can
be measured to be further away from the beacon
A, B and C because it has smaller RSSI readings
comparing to D. Thus, D is considered as to be
outside ABC. Conversely, if D is inside ABC (Figure 2(c)), then it is likely that its neighbors will
be closer to (or further away from) some (but not
all) of the triangle points. Clearly, this test does
not guarantee correctness every time. However,
since there are a large number of triangles available for the test ((O(n3) for n beacons), error can
be effectively controlled. Indeed, simulations
performed in (He, 2003) indicated that APIT
gives more accurate localization than the Centroid
method when the beacon density is higher. Note
that although APIT makes use of RSSI, it is only
used to derive the relative proximity, but not the
absolute distance. Thus, we classify APIT as a
connectivity-based algorithm.
Both the Centroid and APIT methods try to
localize the node directly from the beacons 1
hop away. Thus, to provide better localization
coverage, they require either a large number of
beacons or an extended beacon transmission
range. The DV-Hop method (Niculescu, 2001)

relaxes such requirement by providing a way

to localize from the beacons several hops away.
In DV-Hop, each beacon floods its location to
the entire network much like the distance vector (DV) routing protocol (Perkins, 1997). The
algorithm contains two phases. In the first phase,
a distance-per-hop estimate is obtained using DV.
In the second phase, each node calculates its location estimate using the beacon locations and the
distance-per-hop estimate. Each node maintains
a DV table of the beacon locations it has heard
along with the shortest hop count to them. A
node will only forward the location broadcast if
it has a shorter hop count than the current one in
its table. In addition, when a beacon has heard
the broadcast originated from another beacon,
it can derive the distance-per-hop information
based on the physical distance between the two
beacons and the hop count accumulated along
the path. The distance-per-hop information is
then broadcast to other nodes. To localize, a node
extracts the hop counts to the beacons from its
DV table and converts them into distances using
the average distance-per-hop information it has
received. The node can then estimate its location
using multilateration based on the distances to the
beacons. For instance in Figure 3, the node D can
triangulate based on the location broadcast from
the beacons A, B, and C stored in its DV table.
The distance-per-hop is calculated as the average
of the distances per hop among all the beacons.
Compared to Centroid and APIT, DV-Hop re-

Figure 2. APIT. (a) localization using overlapping triangles; (b) node outside a triangle; (c) node inside
a triangle

(a)

(b)

(c)

1183

Device Localization in Ubiquitous Computing Environments

quires a much lower number of beacons. It does,

however, have greater communication overhead
since it requires multiple message flooding.
The above connectivity-based localization
methods assume the nodes are stationary. The
MCL (Monte Carlo localization) method (Hu,
2004) takes a novel approach by making use
of node mobility. As a node moves, it becomes
connected or disconnected to other nodes. Based
on the connectivity observation, a unit-disk connectivity model, and a simple random movement
model of the node, MCL updates the probability
distribution of the possible node location. Simulation in (Hu, 2004) has reported as many as three
times of localization accuracy when compared
to the Centroid method.
In general, connectivity-based localization
algorithms such as Centroid, APIT and DV-Hop
tend to be simple to implement, and they depend
less on special hardware. However, due to the
lack of more precise measurement, the location
estimates they provide tend to be less accurate. A
large number of beacons need to be deployed in
order to improve their accuracy. However, sparse
networks by nature contain less connectivity
information, and, thus, they are more difficult
to localize accurately using connectivity-based
localization methods.

Figure 3. DV-Hop

1184

RSSI and ToA Range-Based

Algorithms
Many algorithms use the RSSI and ToA measurement to derive the distance to the senders. The
DV-Distance method (Niculescu, 2001) behaves
much like the connectivity-based DV-Hop method. But instead of incrementing the hop count,
DV-Distance increments the distance between
hop to hop as beacons broadcast their locations.
Since the distance at each hop can be quite different, DV-Distance can obtain a more accurate
range to the beacons compared to DV-Hop, which
only considers the average case. However, its
performance becomes dependent on the ranging
measurement accuracy.
The Euclidean method (Niculescu, 2001)
tries to derive the distance to a beacon that is
several hops away by measuring RSSI or ToA
to its neighbors. The distance is obtained by observing some simple Euclidean constraints. For
instance in Figure 4, the node D is several hops
away from the beacon A. To derive its distance to
A, D obtains the distance using RSSI or ToA to
two neighbors B and C, where the distance AB,
AC and BC are known. The distance AD is the
second diagonal of the quadrilateral ABDC. Depending on whether ABDC is convex or concave,

Device Localization in Ubiquitous Computing Environments

Figure 4. Multihop Distance Derivation in Euclidean

two solutions of AD exist. This ambiguity can be

resolved by examining multiple quadrilaterals
like ABDC. Once the distances to at least three
beacons have been obtained, both DV-Distance
and Euclidean method estimate the node location
using multilateration.
The Collaborative Multilateration method
(Savvides, 2001) is also based on multilateration
from ranging. However, it allows nodes being
triangulated from non-beacon nodes. Initially,
all non-beacon nodes are assumed to be at some
random locations. As a node receives its neighbors estimated locations, it tries to triangulate
its new location with the least mean square error.
The newly estimated location is then sent back
to the neighbors for their own multilateration.
The process is iterated multiple times, and the
idea is that eventually the location information
from the beacons will propagate to remote nodes
via collaborative multilateration. However, it is
foreseeable that the nodes further away from the
beacons would be slow to converge. The HopTERRAIN method (Savarese, 2002) makes an
improvement in this regard by using the DV-Hop
method to derive an initial coarse location. It then
runs the collaborative multilateration to further
refine the localization results from the distance
and location information from the neighbors. The
n-Hop Multilateration method proposed in (Savvides, 2003) uses a bounding box model instead

of DV-Hop to provide initial location estimates.

For instance in Figure 5, while node D is two
hops away from the beacon B and one hop away
from the beacon A, it is still bounded by distance
constraints. The bound on the x coordinates is
[xAa,xB+b+c], where a=|AD|, b=|BC| and
c=|CD|. Using this kind of geometric bounding
through multiple hops, an initial location of the
node can be derived.
The iterative multilateration provides a way to
deal with the difficult question of how to effectively
apply the beacon information several hops away.
However, since it treats location estimates from
non-beacons the same as beacons, the beacon
information can be quickly watered down by
the inaccuracy of non-beacons. The probabilistic
localization method in (Huang, 2005) explicitly
considers the location uncertainty of non-beacons
by the means of probability distributions. In particular, each node location is not represented by
a singular value, but a probability distribution in
terms of particles. Initially, all non-beacons have
a uniformly distributed particle distribution. To
localize, nodes exchange their particle distributions among the neighbors and run Monte-Carlo
filtering based on the RSSI or ToA measurement data to update the particles. Eventually,
the particles will be refined to the true location
of where the node resides. The particle filtering
method allows collaborative localization as shown

1185

Device Localization in Ubiquitous Computing Environments

Figure 5. Bounded Region in n-Hop Multilateration

in Figure 6. Here, nodes 2, 3, and 4 are beacon

nodes, while nodes 0 and 1 are non-beacons. Of
the beacons, node 0 can receive signals only from
nodes 1 and 4, and node 1 can receive signals from
only nodes 0, 2, and 3. From the signal strength
readings, non-beacons estimate their distances to
their neighbors. The probability distribution of the
estimated location is represented by the particles
(dots) in the graph. In sub-figure (a), where node 1
is removed, node 0 can only receive signals from
node 4; thus, as the particle distribution indicates,
the probability distribution where node 0 is most
likely located concentrates on a circle around node
4. In sub-figure (b), where node 0 is removed,
node 1 can receive signals from nodes 2 and 3;
thus the most likely locations for node 1 center
around two areas where transmission circles
around node 2 and 3 intersect. Intuitively, in order
to localize itself, a node needs to receive location
information from a minimum of three beacons
either directly or indirectly. In both case (a) and
case (b), the exact location of the nodes 0 and 1
cannot be deduced because they do not receive
location information from all three beacons. In
(c) and (d), where all nodes are available, nodes
0 and 1 are able to communicate to each other
and exchange their particle distributions. Thus,
their probability densities will represent their

1186

actual locations much closer even though neither

node receives location information from all three
beacons directly.
This section introduced range-based localization methods. Compared to range-free methods,
range-based methods give more accurate location
estimates when ranging data is reliable. However,
depending on the deployment environment, ranging techniques based on RSSI tend to be errorprone and strong filtering is required. The ranging
error could ultimately throw off the localization
accuracy if it is allowed to propagate through the
network unbounded. Furthermore, different methods generally exploit the trade-off between the
estimation accuracy and the estimation coverage.
For instance, given the same network scenario, the
Euclidean method is capable of generating more
accurate location estimates of a smaller subset of
nodes, whereas the DV-Hop method has better
coverage but worse accuracy. Regardless of the
tradeoff, a common characteristic shared by many
range-based localization algorithms is that they
require a relatively high network density in order
to achieve better results. Based on the extensive
simulation of DV-Distance, Euclidean and multilateration methods performed in (Chintalapudi,
2004), it can be concluded that those range-based
localization algorithms require an average degree

Device Localization in Ubiquitous Computing Environments

Figure 6. Collaborative Localization Using Particle Filters: (a) Particle distribution of node 0 when node
1 is not present; (b) Particle distribution of node 1 when node 0 is not present, (c) Particle distribution
of node 0 when node 1 is present; (d) Particle distribution of node 1 when node 0 is present

(a)

(b)

(c)

(d)

of 11-12 nodes within the ranging neighborhood in

order to achieve 90% localization coverage with
5% accuracy (Chintalapudi, 2004).

AoA-Based Algorithms
Even though the future of AoA sensing devices is
still unclear, some works have been published on

localization using angle information. Simulation

studies in(Chintalapudi, 2004) also show that
when AoA of the signals is used in addition to the
distance measurement, the localization accuracy
and coverage can be drastically improved. This
should not come as a surprising conclusion, as
nodes need to communicate with only one neighbor to perform localization if they can obtain
1187

Device Localization in Ubiquitous Computing Environments

both AoA and distance measurements. The work

in(Chintalapudi, 2004) also presents three variations of a weighted mean square error algorithm
that localizes the nodes, each of which is designed
to work with one of the three measurement types:
i) distance-only measure; ii) distance plus a more
accurate AoA measure (up to of precision); and iii)
distance plus a less accurate AoA measure (up to
of precision). The less accurate AoA measurement
method is sometimes referred to as sectoring.
Simulations in(Chintalapudi, 2004) show that
the localization accuracy and coverage can be
greatly improved even with such coarse sectoring
measurement as well.
In order to localize with only AoA measurement, the AoA triangulation method proposed in
(Niculescu, 2003) can be used. The triangulation
takes several AoA measurements from beacons
and estimates the node location with least square
error. To propagate the AoA measurement for
more than one hop, the AoA triangulation method
uses a method called orientation forwarding that
is similar to the Euclidean method for distance
ranging. For instance in Figure 7, let AoA measurement be the bearing against South. For node D to
derive its bearing to the beacon A (i.e., D), it can
contact two neighbors, B and C, with known AoA
measurements from the beacon A (i.e., B and C are
known). Furthermore, B, C and D can measure
the AoA of each other to give the readings of B,

B, C, C, D and D. From there, all angles in ABC

and BCD can be determined. The bearing from

A to D can be derived as DD, where D is known,
and CDA can be determined from ABC and
BCD. Using the orientation forwarding method,
the bearing to beacons can be propagated through
mutlitops, which can then be used to triangulate
from remote beacons. However, much like the
case of distance propagation, measurement error
becomes aggregated at each hop. Simulations in
(Niculescu, 2003) have reported a near linear
error increase to the hop count.
In summary, due to the limited availability of
AoA sensing devices, relatively few algorithms
have been proposed for AoA. However, it is
conceivable that some localization algorithms
originally proposed for RSSI or ToA ranging can
be adapted to AoA. For instance, the probabilistic
algorithm in (Huang, 2005) can be updated to accept AoA measurements by simply providing an
alternative measurement model for AoA during
particle filtering.

Interferometric-Ranging
Based Algorithms
Due to the fact that interferometric sensing devices for localization are relatively new, there
have been only a limited number of localization

Figure 7. Multihop Distance Derivation in AoA Triangulation

1188

Device Localization in Ubiquitous Computing Environments

algorithms proposed for this type of measurement.

When compared to RSSI/ToA ranging and AoA,
all of which involve two nodes for each measurement, interferometric ranging involves four
nodes for each measurement and, thus, makes it
more difficult to propagate location information
through multihops. To eliminate the multihop
propagation issue, a simple genetic optimization
approach was taken in (Marti, 2005), which
propagates all interferometric readings within
the network to a centralized location and runs
a genetic algorithm to find the node locations
that match the readings. Such an approach is of
more theoretical (and prototyping) interest than
practical use, since any centralized method is not
scalable to large networks. A Pair-wise Distance
method was proposed in (Patwari, 2006) that
uses both interferometric and RSSI ranging. The
method uses the interferometric ranging to derive
pair-wise distances among the nodes. The node
locations can then be optimized using the least
square error method from the pair-wise distances.
The algorithm then repetitively applies the RSSI
ranging measurements to fine-tune the location
estimates. Compared to the genetic algorithm,
the Pair-wise Distance method is able to converge much faster. However, it is currently still
a centralized algorithm, which presents the same
scalability issue as the genetic algorithm.
Both of the above algorithms try to optimize
for a global solution given an entire set of interferometric measurements. Intuitively, finding
a global solution to the localization problem is
often difficult because of the large search space
and the large number of constraints given by the
interferometric measurements. Thus, it is desirable to find solutions in some subspaces first and
then incrementally build up to the global solution. For instance, an iterative approach has been
proposed in (Huang, 2007) that localizes from a
small set of seeding beacons. At each round, a set
of nodes that can hear from the seeding beacons
are localized. As additional nodes are localized
at each round, they act as pseudo-beacons that

allow other nodes to be localized at subsequent

rounds. The iterative method is fully distributed.
However, error propagation can be an issue since
any localization error at pseudo-beacons would
adversely affect the localization result at subsequent rounds. Simulation results in (Huang, 2007)
have shown a linear increase of localization error
at each round.

Hybrid Algorithms
A combination of the above techniques can be
employed to form hybrid localization methods. For
instance, a hybrid method is proposed in (Ahmed,
2005) that uses both DV-Distance (Niculescu,
2001) and Multi-Dimensional Scaling (MDS)
(Shang, 2003). The algorithm contains three
phases. In the first phase, a small subset of nodes
is selected as reference nodes. In the subsequent
phase, the reference nodes are then localized in
relative coordinates using MDS. The final phase
uses DV-Distance to localize the rest of the nodes
in absolute coordinates. The rational behind
such hybrid algorithms is to exploit the tradeoff
between different localization algorithms. For
example, MDS gives good localization accuracy,
but as the network size is increased, MDS can be
costly. Meanwhile, DV-Distance is less costly, but
it only works well when beacon ratio is high. With
the hybrid algorithm, the cost is minimized by
only running MDS on the reference nodes, and
then the reference nodes are used as beacons for
DV-Distance.

Localization Using Mobility

While most previous methods assume stationary
beacon locations, an alternative method is to localize devices using a mobile beacon. In this method,
a mobile beacon travels through the deployment
area while broadcasting its location along the
way. Devices localize themselves by monitoring
information coming from the beacon. A straightforward technique using the above method is de-

1189

Device Localization in Ubiquitous Computing Environments

scribed in (Ssu, 2005), where devices are required

to receive at least three communications with the
same RSSI reading from the beacon. Given that
the same RSSI reading implies similar distances to
the beacon locations, the physical device location
can be derived using simple geometric functions.
This method is computationally simple, making it
suitable for resource-limited sensors. However, it
requires the beacon to directly pass by the ranging
area of the device. In addition, in most cases, the
beacon has to pass by the device twice because
the sampling positions of the beacon when the
three RSSI readings are taken should not be on
the same line. This method also assumes that
errors are insignificant in the RSSI to distance
translation.
Instead of computing the location directly, a
probabilistic approach may be taken; here device
location is viewed as a probability distribution over
the deployment area. In (Sichitiu, 2004), devices
measure a series of RSSI readings from the mobile
beacons and localize themselves by a sequential
update process to the probability distributions of
their locations. Each device starts with a uniform
distribution covering the entire deployment area.
As the beacon passes through, the distribution is
updated to fit the received RSSI readings (using a
signal propagation model). The method is further
improved in (Peng, 2005) by adding the negative
information (i.e., the information that the beacon
is out of range), as well as RSSI readings from the
neighbors. These probabilistic methods provide
with much improved location estimates, but have
the drawback of being complex. For a deployment
grid of n by n units, the time and space complexity
is O(n2). As the devices such as sensors at present
time have very limited resources, it is difficult to
directly implement these methods for large sensor
deployment scenarios. Indeed, the experimental
results shown in (Sichitiu, 2004) are performed
on pocket PCs, which are much more powerful
than cheap devices like sensors.
A similar method of localizing the networks
using a mobile beacon is presented in (Galstyan,

1190

2004). Instead of the actual probability distribution, the possible device locations are represented
with a bounding box. As the beacon passes by, the
area contained by the bounding box is progressively reduced as positive and negative information
is processed. The bounding box method drastically
simplifies the probability computation, making
it possible to implement this method on sensor
devices. However, such large simplification has its
side-effects in that it sacrifices the preciseness of
the distribution for its simplicity as the box cannot precisely describe multiple possible locations.
There is also the problem of noise from ranging
devices. This method may work well when ranging
error is minimal; however, when noise is present
(which is inevitable when using RSSI ranging),
there might be situations where no bounding box
exists to satisfy all readings.
Table 2 lists all the localization algorithms
described in this section. In summary, different
measurement types and their unique properties
to a large degree dictate the design of localization
algorithms. For instance, connectivity-based measurements can only provide coarse localization
without a higher beacon ratio or nodal degrees.
Range and AoA-based measurements can provide
much finer localization results, but they are more
prone to measurement error. A quantitative comparison between the more well-known algorithms
such as DV-Hop, Euclidean and multilateralization can be obtained from (Langendoen, 2003),
in which the comparison is done in the context
of specific constraints of sensor networks, such
as error tolerance and energy efficiency. Their
results indicate that there is no single algorithm
that performs best and that there is room for
further improvement.

Theoretical Results
While there have been many localization algorithms proposed for various scenarios, only
recently have researchers started to address the

Device Localization in Ubiquitous Computing Environments

Table 2. List of localization algorithms

Connectivity-based

Centroid (Bulusu, 2000), APIT (He, 2003), DV-Hop (Niculescu, 2001), MCL (Hu,
2004)

Range-based

DV-Distance (Niculescu, 2001), Euclidean (Niculescu, 2001), Collaborative Multilateration (Savvides, 2001), Hop-TERRAIN (Savarese, 2002), n-Hop Multilateration (Savvides, 2003), Probabilistic Localization (Huang, 2005)

Angle-based

Weighted Mean Square Error (Chintalapudi, 2004), AoA Triangulation (Niculescu, 2003)

Interferometric Ranging

Genetic Algorithm (Marti, 2005), Pair-wise Distance (Patwari, 2006), Iterative

(Huang, 2007)

Hybrid
Mobility-based

DV-Distance + Multi-Dimensional Scaling (MDS) (Ahmed, 2005)

Geometric Localization using Three RSSI Readings (Ssu, 2005), Sequential
Update to Probability Distribution (Sichitiu, 2004).

theoretical aspects of the localization problem. In

this section, we briefly cover the latest theoretical
results with regard to the localization problem.
Since ubiquitous computing environment is often
modeled as a graph, it is not surprising that much
of the theoretical work is based on graph theory.
With regard to localization, we are particularly
interested in the following three theoretic problems: i) localizability; ii) complexity of localization; and iii) localization error bounds.
First of all, we would like to know that given
a network scenario (i.e., the nodes and their relative measurements such as ranging and angling)
whether it is theoretically possible to uniquely
localize the network. Such knowledge of localizablity is important to us because if we can easily
identify the scenario that is impossible to localize
uniquely, then it would be pointless to run any
localization algorithm on it. Instead, we would
have to request additional nodes or measurement data to be available (by possibly deploying
more nodes or beacons) so that the localizablity
requirement is satisfied. The following theorem
gives the necessary and sufficient condition for
distance-constrained network localizability in
two dimensions.
Theorem 4.1 The network is localizable in two
dimensions if and only if the network graph is
redundantly rigid and triconnected (Hendrickson,
1992; Berg, 2003).

The above theorem makes use of some graph

theory concepts. In graph theory, rigidity (or firstorder rigidity) in general refers to the situation in
a graph where there are no continuous motions
of the vertices satisfying distance constraints on
edges. A graph is redundantly rigid (or secondorder rigid) if the induced graph remains rigid
after removing any single edge.
Theorem 4.1 holds for two dimensions only.
The sufficient condition for higher dimension
is currently unknown. To test the localizability,
there exists a polynomial time algorithm (O(n2)
where n is the number of nodes) that tests for
the first-order rigidity; see (Hendrickson, 2002)
for one implementation. However, it is a known
NP-Complete problem to test for the second-order
rigidity of a graph (Saxe, 1979). A related but even
more difficult problem is node localizability, which
asks if a particular node (instead of the entire
network) is localizable. No sufficient condition of
node localizability is currently known even in the
two dimensional case and, thus, no deterministic
algorithm currently exists.
A second problem asks for the theoretic
complexity of localization itself. In particular,
we would like to know that given a network scenario that satisfies localizability whether there
exists a deterministic polynomial time algorithm
that would localize the network. This problem
deals with the NP-Completeness of localization.
Unfortunately, the hardness of graph realization
1191

Device Localization in Ubiquitous Computing Environments

has been shown as NP-Complete under the measurement of distance (Eren, 2004), angle (Bruck,
2005), connectivity (Breu, 1998; Kuhn, 2004), and
interferometric ranging (Huang, 2007).
The above theoretical results indicate the general intractability of the localization problem even
in the ideal case where measurements (such as
edge distances) are 100% accurate. Unfortunately,
measurements in the real world are a far-cry from
being accurate, and any optimization method
has to deal with not only different measurement
types, but also noise. The localization inaccuracy
attributed to the measurement types and noise
can be statistically qualified using Cramer-Rao
Bounds (CRB) (Patwari, 2005). The CRB is a
lower bound on the covariance of any unbiased
location estimator that uses measurements such
as RSSI, ToA, or AoA. Thus, the CRB indicates a
lower bound of the estimation accuracy of a given
network scenario regardless of the localization
algorithm. In other words, with CRB we have a
way to tell the best any localization algorithm can
do given a particular network, measurement type,
and measurement noise scenario. CRB formulas
of individual measurement types such as RSSI,
ToA, and AoA under most common noise models
(mostly Gaussian) are currently known.
The CRB of the localization error for a sample
network is shown in Figure 8 as rings of radius
being the standard deviation of the minimum
localization error that can be possibly attained at
the node. Here, the nodes represented by squares
are beacons while circles represent nodes to be
localized using RSSI ranging. The edges indicate
the communication links available to measure
RSSI readings. We assume the measurement
model to be RSSI with the path loss exponent p
and the standard deviation of the noise dB. A ring
with smaller radius (i.e., a smaller CRB) signals
that more accurate localization result can be
theoretically obtained. Conversely, a larger ring
indicates a larger localization variance and, thus,
a less accurate result. In the figure, two types of
nodes do not have rings. First, all beacons have

1192

a CRB of 0. There are also regular nodes that

have infinite CRB indicating that those nodes
are theoretically impossible to localize. The latter
case can be seen at nodes 38, 48, 49, and 78 in the
top left corner. At a minimum, three beacons are
needed to localize a connected network. However,
those nodes in the top left corner are isolated to
a different partition. Since they are connected to
only one beacon (node 91), those nodes clearly
cannot be localized. Other than those cases, the
CRB rings at the main network partition clearly
show the level of localization difficulty under
various scenarios. In general, we observe that
nodes closer to the beacons tend to have a smaller
CRB than the ones that are several hops away.
Even smaller CRB can be obtained when a node
is closer to more than one beacon. All of the above
observations are consistent with our common
intuition about localization difficulty.
It is important to note that CRB is essentially
a theoretic bound that depends on the measurement model. In the real world, its usefulness is
limited by how accurate the measurement model
reflects the reality. Nevertheless, CRB can be a
useful tool in comparing various localization
algorithms. It can be used to validate how close
a particular algorithm can come to this theoretic
lower bound and to see if there is any room for
improvement in the algorithm design.

Conclusion
In this chapter, we studied the localization
problem in ubiquitous computing environments.
Localization in general refers to the problem of
identifying the physical location of devices using
a limited amount of available measurement data.
The most common measurement types include
device connectivity (i.e., whether two devices are
neighbors), ranging using RSSI and ToA, angle of
arrival (AoA), and interferometric ranging. Given
a small number of nodes with accurate geometric
location (e.g., using GPS receivers), localization

Device Localization in Ubiquitous Computing Environments

Figure 8. The CRB of the sample network is depicted as rings of the radius i. There are two exceptions:
1) beacons, depicted as squares, have 0 CRB, and 2) some regular nodes have infinite CRB (such as
node 38, 48, 49 and 78 at the top left corner) indicating that they cannot be localized

algorithms try to derive the location of those devices that are not GPS-enabled. The motivation
of localization can be justified by the large number
of algorithms proposed for ubiquitous computing
that rely on (semi-)accurate location information
and the fact that current technology prevents GPS
from being installed on all network devices due
to power constraints and form factors. It has been
shown that localization in general, regardless of the
measurement types used, is an NP-Hard problem.
Thus, current effort in solving it relies on some
sort of stochastic optimization. Meanwhile, as
with other network-related problems in ubiquitous
computing environments, the ideal solution calls

for a distributed but efficient implementation,

which leads to additional challenges.
Like other aspects of ubiquitous computing,
the localization problem is relatively new. The
problem is also conceptually straight-forward to
characterize, and many results from other disciplines such as graph theory, optimization theory,
online algorithms can be readily applied to this
problem. Thus, researchers from other disciplines
can provide valuable insight that could lead to
better solutions to the problem. It is our hope that
this brief introduction will provide the readers
motivation and inspiration to perform research
in this exciting field.

1193

Device Localization in Ubiquitous Computing Environments

Future Directions
Device localization within ubiquitous computing
environment has been an active research field in
the past several years. Much work has been done in
the area of hardware/sensor design (in particular,
reducing the form factor and power consumption of sensory devices), algorithmic design and
theoretical analysis. However, like many areas
of ubiquitous computing, localization is still a
relatively new front with much of the work yet to
be done. In this section, we will briefly discuss
a few directions which we feel could produce
fruitful results in the near future. We hope our
discussion will encourage the readers to actively
participate and contribute their own ideas to this
exciting and important field.

Implementation and Testing

Environment
When reviewing the previous works on localization, one cannot help but notice a disturbing trend;
a majority of works on localization have been
based on either theoretical models or numerical
simulations, while works based on the result of actual hardware implementation have been relatively
few. It is not difficult to project that the primary
reason for such trends is the hardware cost. To
perform meaningful experiments for localization,
especially for those collaborative localization
methods such as DV-Distance and Euclidean, one
would normally need a large number (100+) of
devices. Although simple mobile ad hoc network
devices (e.g., sensor motes) are becoming cheaper
by the day, it is still quite costly to implement
algorithms on physical devices on such a large
scale. In addition, the sensing capacities of the
current devices are usually limited to RSSI. Currently, there is no cheap hardware that implements
AoA, ToA, or interferometric ranging and, thus,
most works using those measurement types are
all based on simulations. In a sense, the advances
in algorithmic work on the localization problem

1194

are currently outpacing the advances in hardware.

Future work needs to be done to significantly
improve the hardware design to fill this gap.
Another issue related to the testing environment is that there is no common localization test
bed. While large scale network simulators such as
NS2 have modules for simulating mobile ad hoc
networks, these modules do not contain localization. Smaller simulators for sensor networks such
as SENSE (Chen, 2004) usually lack features on
localization as well. Another simulator SENS
(Sundresh, 2004) does explicitly implement localization, but it uses a rather primitive Centroid
method and lacks more sophisticated methods.
Since NS2 is the most widely-used network
simulator, it would be very helpful for researchers to implement an interface to NS2 that allows
a plug-in for future localization algorithms.
This would give a common test bed for different
localization schemes. More importantly, it would
also allow those location-depended algorithms
(such as location-aided routing methods) to be
implemented and compared based on the result
of localization. The localization module should
also implement the Cramer Rao Bounds (CRB)
so that the theoretic error bound can be calculated
for different localization scenarios.

Interferometric Ranging
Since interferometric ranging is a relatively new
type of measurement available to the localization
problem, there are still many open problems in
this area. Of the localization algorithms proposed
for interferometric ranging, all but the iterative
algorithm proposed in (Huang, 2007) is centralized. There is a definite need to design distributed
localization algorithms for interferometric ranging so that it can be implemented with reasonable
efficiency and scalability. To reduce the number
of beacons, the distributed algorithms should
make use of multi-hop location information,
which unfortunately is much more difficult for
interferometric ranging because each measure-

Device Localization in Ubiquitous Computing Environments

ment involves four nodes. For instance, there is

a scheduling issue, both at a high level and low
level, of when a device should be scheduled to
send or receive interferometric readings. At a
high level, since not all q-ranges are independent,
it is more desirable to schedule the senders and
receivers in order to generate more independent
q-ranges. At the low level, after the senders and
receivers are selected, they have to be scheduled
to coordinate the signal transmission because
the transmission needs to be synchronized. The
design of the scheduling algorithm can have a substantial impact on the overall performance of the
localization algorithm in terms of the localization
accuracy and communication overhead.
Furthermore, a simulation study in (Huang,
2007) has shown that the multi-hop error propagation has a big impact on interferometric ranging,
which increases almost linearly as the localization
results are propagated at each hop. Therefore, the
control of the error propagation is another research
issue. There is also a need for an algorithmic
independent theoretic error bound (like CRB)
for interferometric ranging. The bound would be
more difficult to derive than those for distance
ranging and angling because more than two
nodes are involved in each measurement. Thus,
the inter-dependence between the error and the
relative locations of senders and receivers becomes
more challenging to characterize mathematically.
However, the payoff of obtaining such bound
is that it would allow us to ultimately compare
interferometric ranging with other measurement
types and identify the scenarios that are preferable
for each measurement type.

Collaborative Localization of
Multiple Measurement Types
Previous localization algorithms often assume that
the entire network has to be localized using the
same type of measurement (such as connectivityonly, RSSI, ToA, AoA, or interferometric ranging).
However, to be true to the spirit of ubiquitous

computing, it is foreseeable that future networks

will consist of devices of vastly different capacities
in terms of i) different transmission coverage, ii)
power requirement, and iii) measurement sensors.
Thus, during localization it is often desirable to
explicitly consider various devices capacities in
such heterogeneous networks. For instance, in
terms of power requirements, the localization
algorithm should exploit the devices with more
power capacity and try to minimize the calculation
performed on less powerful devices. Furthermore,
different measurement types have different error
characteristics. It would be interesting to investigate how to collaborate multiple measurement
types during localization, and in particular how the
collaboration would impact the localization error.
Intuitively, incorporating multiple measurement
types in the same localization scenario should
improve performance since such collaboration
can potentially cover the drawbacks of individual
measurement types for each other. Unfortunately,
such collaboration could mean an increased
complexity of the localization algorithms since
heterogeneous networks would invalidate some
assumptions often made by simple localization
algorithms (such as uniform transmission range).
While it is worthwhile to consider collaborative
localization algorithms, it is equally imperative to
keep the localization overhead under control.

References
Ahmed, A. A., Shi, H., & Shang, Y. (2005).
SHARP: A new approach to relative localization
in wireless sensor networks. In Proceedings of
the 25th IEEE International Conference on Distributed Computing Systems Workshops (ICDCS
05) (pp. 892-898).
Ash, J. N., & Potter, L. C. (2004). Sensor network
localization via received signal strength measurements with directional antennas. In Proceedings of
the 2004 Allerton Conference on Communication,
Control, and Computing (pp. 1861-1870).
1195

Device Localization in Ubiquitous Computing Environments

Berg, A., & Jordan, T. (2003). A proof of Connellys

conjecture on 3-connected generic cycles. Journal
of Combinatorial Theory Series B, 88(1), 77-97.

Hendrickson, B. (1992). Conditions for unique

graph realizations. SIAM Journal on Computing,
21(1), 65-84.

Breu, H., & Kirkpatrick, D. G. (1998). Unit disk

graph recognition is NP-hard. Computational Geometry. Theory and Applications, 9(1-2), 3-24.

Hu, L., & Evans, D. (2004). Localization for

mobile sensor networks. In Proceedings of the
ACM (MOBICOM04) (pp. 45-57).

Bruck, J., Gao, J., & Jiang A. (2005). Localization and routing in sensor networks by local
angle information. In Proceeding of the 6th ACM
International Symposium on Mobile Ad Hoc Networking and Computing (pp. 181-192).

Hu, Y.-C., Perrig, A., & Johnson, D. (2003). Packet

leashes: a defense against wormhole attacks in
wireless ad hoc networks. In Proceedings of the
INFOCOM03 (pp. 1976-1986).

Bulusu, N., Heidemann, J., & Estrin, D. (2000).

GPS-less low cost outdoor localization for very
small devices. IEEE Personal Communications
Magazine, 7(5), 28-34.
Chen, G., Branch, J., Pflug, M. J., Zhu, L., &
Szymanski, B. (2004). SENSE: A sensor network
simulator. Advances in Pervasive Computing and
Networking (pp. 249-267). Springer.
Chintalapudi, K., Govindan, R., Sukhatme, G.,
& Dhariwal, A. (2004). Ad-hoc localization using ranging and sectoring. In Proceedings of the
IEEE (INFOCOM 04) (pp. 2662-2672).
Eren, T., Goldenberg, D., Whiteley, W., Yang, Y. R.,
Morse, A. S., Anderson, B. D. O., & Belhumeur,
P. N. (2004). Rigidity, computation, and randomization of network localization. In Proceedings of
the IEEE (INFOCOM04) (pp. 2673-2684).
Galstyan, A., Krishnamachari, B., Lerman, K.,
& Pattem, S. (2004). Distributed online localization in sensor networks using a moving target. In
Proceedings of the IPSN 2004, (pp. 61-70).
He, T., Huang, C., Blum, B. M., Stankovic, J. A.,
& Abdelzaher, T. F. (2003). Range-free localization schemes in large scale sensor networks. In
Proceedings of the ACM (MOBICOM03) (pp.
81-95).

1196

Huang, R., Zruba, G. V., and Huber, M. (2007).

Complexity and error propagation of localization
using interferometric ranging. In Proceedings of
the IEEE ICC.
Huang, R., & Zruba, G. V. (2005). Location
tracking in mobile ad hoc networks using particle
filters. In Proceedings of the ADHOC-NOW05
(pp. 85-98).
Jiang, X., & Camp, T. (2002). Review of geocasting protocols for a mobile ad hoc network. In
Proceedings of the Grace Hopper Celebration
(GHC).
Ko, Y., & Vaidya, N. H. (2000). Location-aided
routing (LAR) in mobile ad hoc networks. Wireless Networks, 6(4), 307-321.
Ko, Y. & Vaidya, N. H. (1999). Geocasting in
mobile ad hoc networks: location-based multicast algorithms. In Proceedings of the IEEE
(WMCSA99) (pp. 101).
Kuhn, F., Moscibroda, T., & Wattenhofer, R.
(2004). Unit disk graph approximation. In Proceedings of the Workshop on Discrete Algorithms
and Methods for Mobile Computing and Communications (DIAL-M) (pp. 17-23).
Kuhn, F., Wattenhofer, R., Zhang, Y., & Zollinger,
A. (2003). Geometric ad-hoc routing: of theory and
practice. In Proceedings of the ACM (PODC03)
(pp. 63-72).

Device Localization in Ubiquitous Computing Environments

Kus, B., Marti, M., Balogh, G., V:olgyesi, P.,

Sallai, J., Ndas, A., Ldeczi, A., & Meertens, L.
(2006). Node density independent localization. In
Proceedings of the 5th International Conference
on Information Processing in Sensor Networks
(IPSN 2006) (pp. 441-448).
Langendoen, K., & Reijers, N. (2003). Distributed localization in wireless sensor networks: a
quantitative comparison. Computer Networks,
43(4), 499-518.
Liao, W.-H., Tseng, Y.-C., & Sheu, J.-P. (2001).
GRID: a fully location-aware routing protocol
for mobile ad hoc networks. Telecommunication
Systems, 18(1), 37-60.
Liao, W.-H., Tseng, Y.-C., Lo, K.-L., & Sheu,
J.-P. (2000). Geogrid: a geocasting protocol for
mobile ad hoc networks based on grid. Journal
of Internet Technology, 1(2), 23-32.

Patwari, N., & Hero, A. O. (2006). Indirect radio

interferometric localization via pairwise distances. In Proceedings of the 3rd IEEE Workshop
on Embedded Networked Sensors (EmNets 2006)
(pp. 26-30).
Patwari, N., Hero III, A. O., Perkins, M., Correal,
N. S., & ODea, R. J. (2003). Relative location
estimation in wireless sensor networks. IEEE
Transactions on Signal Processing, 51(8), 21372148.
Patwari, N., Hero, A., Ash, J., Moses, R., Kyperountas, S., & Correal, N. (2005). Locating the
nodes: cooperative geolocation of wireless sensors. IEEE Signal Processing Magazine, 22(4),
54-69.
Parkinson, B. et al. (1996). Global positioning
system: theory and application. Volume I, Progress
in Astronautics and Aeronautics, (Vol. 163).

Marti, M., Kus, B., Balogh, G., V:olgyesi, P.,

Ndas, A., Molnr, K., Dra, S., & Ldeczi,
A. (2005). Radio interferometric geolocation.
In Proceedings of the ACM 3rd Conference on
Embedded Networked Sensor Systems (SenSys)
(pp. 1-12).

Peng, R., & Sichitiu, M. L. (2005). Robust, probabilistic, constraint-based localization for wireless
sensor networks. In Proceeding of the 2nd Annual
IEEE Communications Society Conference on
Sensor and Ad Hoc Communications and Networks (SECON 2005) (pp. 541-550).

Mauve, M., Fuler, H., Widmer, J., & Lang, T.

(2003). Position-based multicast routing for
mobile ad-hoc networks (Technical Report
TR-03-004). Department of Computer Science,
University of Mannheim.

Perkins, C. E. (1997). Ad-hoc on-demand distance

vector routing. In MILCOM 97 panel on Ad Hoc
Networks.

Niculescu, D., & Nath, B. (2001). Ad hoc positioning system (APS). In Proceedings of the IEEE
(GLOBECOM01) (pp. 2926-2931).
Niculescu, D., & Nath, B. (2003). Ad hoc positioning system (APS) using AoA. In Proc. of IEEE
INFOCOM03 (pp. 1734-1743).
Niculescu, D., & Nath, B. (2004). VOR base stations for indoor 802.11 Positioning. In Proceeding
of the 10th Annual International Conference on
Mobile Computing and Networking (pp. 58-69).

Priyantha, N., Miu, A., Balakrishnan, H., & Teller,

S. (2001). The Cricket compass for context-aware
mobile applications. In Proceedings of the 6th
ACM, (MOBICOM 01) (pp. 1-14).
Savvides, A., Han, C.-C., & Srivastava, M. (2001).
Dynamic finegrained localization in ad-hoc networks of sensors. In Proceedings of the 7th ACM
International Conerence. on Mobile Computing
and Networking (MOBICOM) (pp. 166-79).
Savvides, A., Park, H., & Srivastava, M. B.
(2003). The n-hop multilateration primitive for
node localization problems. Mobile Networks
and Applications, 8, 443-451.

1197

Device Localization in Ubiquitous Computing Environments

Savarese, C., Rabay, J., & Langendoen, K. (2002).

Robust positioning algorithms for distributed adhoc wireless sensor networks. In Proceeding of
the USENIX Technical Annual Conference (pp.
317-327).
Saxe, J. (1979). Embeddability of weighted graphs
in k-space is strongly NP-hard. In Proceedings of
the 17th Allerton Conference in Communications,
Control and Computing (pp. 480-489).
Shang, Y., Ruml, W., Zhang, Y., & Fromherz,
M. (2003). Localization from mere connectivity.
In Proceedings of the ACM (MobiHoc03) (pp.
201-212).
Sichitiu, M. L., & Ramadurai, V. (2004). Localization of wireless sensor networks with a mobile
beacon. In Proceedings of the 1st IEEE Conference
on Mobile Ad-hoc and Sensor Systems (MASS
2004) (pp. 174-183).
Sivrikaya, F., & Yener, B. (2004). Time synchronization in sensor networks: a survey. IEEE
Network, 18(4), 45-50.
Ssu, K.-F., Ou, C.-H., & Jiau, H.C. (2005). Localization with mobile anchor points in wireless
sensor networks. In Proceedings of the IEEE
Vehicular Technology Conference (VTC 2005)
(pp. 1187-1197).
Stojmenovic, I., Ruhil, A. P., & Lobiyal, D. K.
(2006). Voronoi diagram and convex hull based
geocasting and routing in wireless networks.
Wireless Communications and Mobile Computing, 6(2), 247-258.
Sundresh, S., Kim, W., & Agha, G. (2004). SENS:
A sensor, environment and network simulator.
In Proceedings of the 37th Annual Simulation
Symposium (ANSS37).
Xu, Y., Heidemann, J., & Estrin, D. (2001).
Geography-informed energy conservation for
adhoc routing. In Proceedings of the ACM/IEEE
(MOBICOM01) (pp. 70-84).

1198

Additional Reading
Bulusu, N., Heidemann, J., & Estrin, D. (2000).
GPS-less low cost outdoor localization for very
small devices. IEEE Personal Communications
Magazine, 7(5), 28-34.
Eren, T., Goldenberg, D., Whiteley, W., Yang, Y. R.,
Morse, A. S., Anderson, B. D. O., & Belhumeur,
P. N. (2004). Rigidity, computation, and randomization of network localization. In Proceedings of
the IEEE (INFOCOM04) (pp. 2673-2684).
Ko, Y., & Vaidya, N. H. (2000). Location-aided
routing (LAR) in mobile ad hoc networks. Wireless Networks, 6(4), 307-321.
Langendoen, K., & Reijers, N. (2003). Distributed localization in wireless sensor networks: a
quantitative comparison. Computer Networks,
43(4), 499-518.
Liao, W.-H., Tseng, Y.-C., & Sheu, J.-P. (2001).
GRID: a fully location-aware routing protocol
for mobile ad hoc networks. Telecommunication
Systems, 18(1), 37-60.
Patwari, N., Hero III, A. O., Perkins, M., Correal,
N. S., & ODea, R. J. (2003). Relative location
estimation in wireless sensor networks. IEEE
Transactions on Signal Processing, 51(8), 21372148.
Patwari, N., Hero, A., Ash, J., Moses, R., Kyperountas, S., & Correal, N. (2005). Locating the
nodes: cooperative geolocation of wireless sensors. IEEE Signal Processing Magazine, 22(4),
54-69.
Savvides, A., Park, H., & Srivastava, M. B.
(2003). The n-hop multilateration primitive for
node localization problems. Mobile Networks
and Applications, 8, 443-451.
Stojmenovic, I., Ruhil, A. P., & Lobiyal, D. K.
(2006). Voronoi diagram and convex hull based
geocasting and routing in wireless networks.

Device Localization in Ubiquitous Computing Environments

Wireless Communications and Mobile Computing, 6(2), 247-258.

This work was previously published in Advances in Ubiquitous Computing: Future Paradigms and Directions, edited by S.
Mostefaoui, Z. Maamar, and G. Giaglis, pp. 83-116, copyright 2008 by IGI Publishing, formerly known as Idea Group Publishing (an imprint of IGI Global).

1199

1200

Chapter 3.12

Socio-Cultural Interpretations
to the Diffusion and Use of
Broadband Services in a Korean
Digital Society
Dal Yong Jin
Simon Fraser University, Canada

ABSTRACT

INTRODUCTION

This chapter attempts to ascertain the causes of

the rapid growth of broadband services in the
context of the broader socio-cultural elements. It
recognizes technology as a socio-cultural product
which has historically been constituted by certain
forms of knowledge and social practice, so this
chapter explores cultural elements contributing
to the diffusion of broadband services in the
context of the cultural environment in Korea.
Further, this chapter discusses the significant role
of the people, as users, in the process of the rapid
diffusion and growth of broadband services. In
particular, it emphasizes the way in which the
1997 economic crisis, as one of the most significant socio-cultural turning points in modern
Korean history, has influenced the deployment
of broadband services as high-speed Internet
connections have developed since 1997.

The widespread availability of broadband services

(high-speed Internet) throughout the world significantly influences people in their life. Broadband
services have made it easier to download digital
music and movies, and the wide penetration of
broadband services has enabled customers to
engage in online stock transactions and online
games. Many schools, from elementary to universities, in various countries are also connected
to broadband services and utilize information
technology (IT) for education, while Internet
broadcasters have rapidly become popular.
A wide range of developed countries, such
as the United States and the United Kingdom to
developing countries in the Third World, have initiated modern telecommunications networks that
support broadband access, providing high-speed
access and always-on connections to both homes

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Socio-Cultural Interpretations to the Diffusion and Use of Broadband Services in a Korean Digital Society

and offices, to develop the social and technical

landscape of cyberspace (Han, 2003; Ishii, 2003;
Lee & Lee, 2003; Jin, 2007). Korea is among the
leading performers in broadband (Lee, Oh, &
Shim, 2005). As of January 2006, about 77% of
Korean households were connected to broadband
services, which is one of the highest throughout
the world (MIC, 2006).
Consequently, Korea has become the worlds
best laboratory for broadband servicesand a
place to look to for answers on how the Internet business may evolve (Taylor, 2006). Much
scholarly analysis and discourse (Yun, Lee, &
Lim, 2002; Lee, OKeefe, & Yun, 2003; Reynolds & Sacks, 2003; Choudrie & Lee, 2004; Lau,
Kim, & Atkin, 2005; Lee et al., 2005; Jin, 2005)
has focused on the roles of the government and
competition among telecommunications companies in developing broadband services in Korea.
They emphasize several factors contributing to
the rapid growth of broadband services caused
by the deregulation and competition policies in
the telecommunications sector, such as a variety
of promotion policies to boost Internet use and
the strategies of broadband Internet providers.
Previous studies, however, have not paid much
attention to socio-cultural factors, which would
be one of the most important contributing factors
given that people are the major users of broadband
services. Although a few papers examine cultural
factors that contribute to broadband services,
their discussions are neither comprehensive, nor
informative.1
Unlike these preceding studies, this chapter
attempts to ascertain the causes of the rapid
growth of broadband services in the context of
the broader socio-cultural elements. It recognizes
technology as a socio-cultural product which has
historically been constituted by certain forms of
knowledge and social practice, so this chapter
explores cultural elements contributing to the
diffusion of broadband services in the context
of the cultural environment in Korea. Further,
it discusses the significant role of the people, as

users, in the process of the rapid diffusion and

growth of broadband services. In particular, it
emphasizes the way in which the 1997 economic
crisis, as one of the most significant socio-cultural
turning points in modern Korean history, has influenced the deployment of broadband services, as
high-speed Internet connections have developed
since 1997.

Technology as Cultural
Forms
It is generally recognized that technologies are
primarily neutral because they operate essentially under the same norm of efficiency in all
situations. Many users of technology argue that
technology is essentially amoral and an entity
devoid of values (Rescher, 1969; Mesthene, 1970).
This instrumental theory, the dominant view of
modern governments and the policy sciences on
which they depend, argues that if people use
technology for destruction or pollution, as in the
case of nuclear weapons and chemical pollution,
it should not be blamed on technology, but on its
misuse by politicians, the military, big business
and others (Pacey, 1983, p. 2).
For many scholars, however, technology is
not simply a means to an end, but has become an
environment and a way of life; this is its substantive impact (Borgmann, 1984). This substantive
theory of technology holds that technology is
not neutral, but has a substantive value bias.
Substantive theory, best known through the writings of Jacques Ellul, Arnold Pacey, and Martin
Heidegger, claims that technology constitutes
a new type of cultural system that restructures
the entire social world as an object of control.
Substantive theory explicates cultural aspects
of technology, such as values, ideas, and the
creative activity of technology (Feenberg, 1991).
This type of cultural system is characterized by
an expansive dynamic which ultimately mediates
every pre-technological enclave and shapes the
whole of social life.

1201

Socio-Cultural Interpretations to the Diffusion and Use of Broadband Services in a Korean Digital Society

Among these, Ellul (1964) pointed out that we

have to examine technology through its sociological aspects: We should consider the effect of
technique on social relationships, political structures, and economic phenomena. Technology is
not aloof from the social realm; on the contrary,
it is an integral part of the social (Webster, 2002).
Pacey (1983, pp. 4-6) also discussed how those
who write about the social relations and social
control of technology should be made aware of
the cultural aspects as well as the organizational
and technological aspects.
In agreement with Pacey, Christians (1989,
pp. 124-125) observes: Technology is the distinct cultural activity in which human beings
form and transform natural reality for practical
ends with the aid of tools and procedures. He
clearly argues that cultures are humans distinctive and immediate environment built from the
material order by men and womens creative effort. Furthermore, Williams (1992, pp. 127-129)
stated: How technology develops is not only a
matter of some autonomous process directed by
remote engineers, but is a matter of social and
cultural processes. Technologies are developed
and used within a particular social, as well as
economic and political context (Franklin, 1999).
As these theoreticians emphasized, technology
does not develop independently, but is part of a
particular social-economic and cultural process.
This chapter aims to contribute to this ongoing
debate of substantive theoretical discourse with
a case study of the rapidly growing broadband
services in Korea.

Historical Background Information

on the Growth in Broadband
Services
The rapid deployment of broadband services
in Korea began in 1995 when the government
enacted the Framework Act on Information,
which established the first Master Plan, which
set up a comprehensive strategy for the Korean

1202

Information Infrastructure (KII) (MIC, 2004).

The goal of the KII was to construct an advanced
nationwide information infrastructure consisting
of communications networks, Internet services,
application software, computers, and information products and services (Jin, 2007). The KII
project aimed at building high-speed networks by
providing over 80% of households with more than
20Mbps (megabits per second) broadband access
by 2005 through market competition and private
sector investment (Lee et al., 2003).2
In the post-1997 economic crisis, efforts bringing about the development of broadband services
have accelerated. The economic crisis severely affected the Korean economy due to unprecedented
rates of unemployment, corporate bankruptcies,
and the demise of the stock market.3 Koreas
economic condition demanded that Korea change
its industrial structure from traditional heavy
and chemical industries to a more IT-oriented
structuretelecommunications and computers.
More importantly, Korea identified information
technology as its path to economic recovery, following the 1997 financial crisis (Kim, 2006a). The
Korean government has exerted great effort on
deploying a high-capacity backbone and pursued
a policy of high-speed telecom infrastructure as
a foundation for the benefits of IT changes for
sustainable economic growth (Lee et al., 2003).
Under these circumstances, both government
planning and stable funding had played significant roles in the rollout of broadband services.
The Korean government invested a total of $11
billion into broadband services between 1998
and 2002, while the U.S. government planned to
invest only $2 billion in the form of tax breaks as
of May 2003 (Belson & Richtel, 2003, p. C1). The
Korean government has also initiated a variety
of promotion policies expediting the growth of
broadband services, such as creating InternetFriendly Classrooms4 and Ten Million People
Internet Education. Clearly, Koreas rapid and
expansive broadband growth can be traced, in
large measure, to government initiatives (Lau et
al., 2005, p. 351).

Socio-Cultural Interpretations to the Diffusion and Use of Broadband Services in a Korean Digital Society

Competition among telecom companies was

also dedicated to the deployment of broadband
access throughout Korea. Broadband services in
Korea were first introduced in 1998, and a few telecom companies, including Korea Telecom (KT),
the largest telecom company in Korea, Thrunet,
and Hanaro, had severely competed with each
other in the market. In May 1999, Thrunet, the
introducer of broadband services in Korea, held
63% of the market share, followed by Hanaro (35%)
and KT (2%). The market has undergone substantive change in recent years. For example, Hanaro
Telecom, controlled by U.S. investors American
International Group Inc. and Newbridge, absorbed
Thrunet to be in a better economic position to
compete with the dominant market player KT in
late 2005. LG Powercomm, a subsidy company
of the LG group, entered the market in September
2005, and cable system operators have expanded
their share in the high-speed Internet market (Kim,
2006b). As a result, as of April 2006, KT as the
largest player consisted of 49.9% of the market
share, followed by Hanaro (28.5%), cable system
operators (10.6%), Powercomm (4.4%), and others
(Park, 2006, p. 16).
The intense competition among telecom
companies led to price (monthly fee) reductions
and subsequently a rapid increase in demand for
broadband Internet services. In other words, competition among telecom companies for increasing
their respective market share has resulted in a
drop in the price of broadband subscriptions, and
it has certainly contributed to the rapid spread of
broadband in Korea. Fierce competition resulted
in prices decreasing from $40 per month in April
1999 to $30 in February 2003, and again to less than
$20 in May 2006 (ITU, 2003a; Taylor, 2006).5
Meanwhile, as a new market trend, Korea has
been developing the next generation in high-speed
Internet services. The country acknowledges the
existence of market saturation, so it has enhanced
the capacity of existing broadband services, including its speed and convergence with various
types of media. Korea has begun to upgrade its

high-speed Internet from less than 5 Mbps to 100

Mbps since 2005 in order to enjoy high-quality
Internet telephone service, movies, and online
education (Bang, 2005). Further, Korea expects
the rapid penetration of 100 Mpbs broadband
services will expedite the growth of image-related contents, and thereafter the second wave
of high-speed Internet services in 10 years.6 As
such, the Korean government along with telecom
companies played a pivotal role in the growth of
broadband services by providing infrastructure
and government initiatives.

Cultural Characteristics
Contributing to the Deployment of
Broadband Services
Although the Korean government and telecom
firms have collaboratively initiated and expedited
the deployment of high-speed Internet services,
the swift growth of broadband services would
have been impossible without people readily accepting new technology more than other nations.
The diffusion of services and the widespread
use of high-speed Internet can be attributable
to certain distinct characteristics of the Korean
people, since the diffusion of broadband services
took place with rapid acceptance by most Koreans (Jin, 2007). Regardless of the fact that many
countries around the world have initiated and
supported the growth of broadband services, the
result varies markedly due to the acceptance of
new technology by people, the users, as well as
government initiatives.
Therefore, it is crucial to explore the significance of assessing the culture in which technology
is created and the context in which it is widely
accepted. As ITU acknowledges (2003b), Internet
user demand indeed contributed most decisively to
the rapid explosion of broadband in many places,
particularly in Korea. Again Ellul (1964) emphasized that the development of technology is not
an isolated fact in society but is related to every
factor in the life of modern humanity. In particu-

1203

Socio-Cultural Interpretations to the Diffusion and Use of Broadband Services in a Korean Digital Society

lar, cultural forms rooted in Korean history after

the 1997 economic crisis should be emphasized
as significant elements which prompted the rapid
deployment of broadband, because members of
the public are the major customers of broadband
services. There are several significant cultural
characteristics (i.e., the mentality of the people)
rooted in Korean society and its historical context
that are important contributing factors to the development of broadband, such as rapid acceptance
of new services and technologies, enthusiasm for
'edutainment,' keeping up with their neighbors
(me-too culture), as well as developing dual
personalities of two competing forces, social
solidarity and individualism.
The primary characteristic among most Koreans is demanding quick change, which eventually
expedited the growth in high-speed Internet. Only
a few decades ago Koreans were characterized
as calm and patient, as one distinguished Confucian characteristic, but in just a few years, the
demand for quick change became one of the most
distinctive Korean characteristics (Han, 2003, p.
19). Since the 1960s when the country began its
high-speed transition from an impoverished state
to one of Asias major economies, Koreans cannot
wait for long periods of time, nor are they very
patient. Koreans have not been known for taking
things slowly. Instead, Korean society is well
known for its impulsivenessthat is, desiring
quick communication, quick contact, and quick
results (Jin, 2007). Most Koreans would certainly
agree that Korea is nothing if not dynamic. In this
regard, Jong Suk Lee (2000), director of the Korea
Culture and Policy Institute, states: Nine out of
ten Koreans rush into the road as soon as the light
changes to green and they cross the street on or
even before the green light is about to turn red. In
contrast, Chinese would usually wait until all the
traffic is cleared and safe even at the clear green
light. And if the light is about to turn red, they
just stop and wait for the next green light.
Indeed, balli, balliKorean for hurry,
hurryis heard incessantly on the crowded

1204

streets of Seoul, reflecting the frenetic pace of

Korean life in a global era (Jin, 2007).
Another survey conducted in the 1980s is still
compelling to explain the same trend. According to the survey (Chinese Cultural Connection,
1987), regarding Confucian dynamism which
measures long-term vs. short-term orientation,
Japan showed one of the highest scores, compared
to the U.S., the UK, and Australia. The high Japanese score in the survey reflected the patience or
long-term orientation of Japan compared with
other developing countries (McFadyen, Hoskins,
& Finn, 1998). However, Korea has lost patience
and has become addicted to short-term results,
and this changing Confucian dynamism in Korea
has contributed to the rapid penetration of highspeed Internet.
This relentless drive, which has led Korea to
chalk up a number of significant achievements,
has driven the rapid spread of broadband Internet
connections. Most Koreans have hurried to have
broadband Internet connections, and it has been
easier to develop these services compared to other
countries, including Japan. What most Koreans
wanted was the same thing: quick communication,
quick games, and quick contact. Everything needs
to go balli, balli, and this unique characteristic
of Koreans has expedited the rapid penetration of
high-speed Internet (Shim, 2006). In other words,
for most Koreans, everything needs to be done
right now. For example, in Korea, instant message-swapping is comparatively popular mainly
because not answering a message from a friend
with all due speed is considered a faux-pas in
Korean society (Taylor, 2006).
This quick pace has particularly accelerated
with the 1997 economic crisis because Koreans
began to realize that quick change and adoption
of new technology and information was needed
to survive in the globalization era. Korea is full
of early adopters who are willing to buy newly
released digital devices for consumer testing. In
a survey by Korean ad agencies in 2004, 43% of
respondents said that they consider themselves

Socio-Cultural Interpretations to the Diffusion and Use of Broadband Services in a Korean Digital Society

early adopters in the market (Kwon, 2006). According to a survey by the U.S. market research
firm Parks Association in November 2005, Korea
ranked second on the list of 13 countries in the
adoption of consumer technologies (Parks Associates, 2005).7 Although balli balli culture has
several negative aspects, such as bribery scandals
and the collapse of buildings and bridges, in the
midst of achieving social and economic successes
in the Korean society, rapid adaptability to change
is the key to the swift deployment of broadband
services.
An excessive enthusiasm for edutainmentthe
combination of education and entertainmenthas
also greatly contributed to the unique growth in
broadband services. Above all, Korea is one of
the most developed countries in terms of education. Its overall school enrollment rate (primary,
secondary, and tertiary) of 90% is the highest
throughout the world. Koreas high rate of literacy
and school enrollment are essential as prerequisites for the widespread adoption of ICTs, and
these factors have helped contribute to the growing
impact of ICT in Korean society (ITU, 2003a).
Over-enthusiasm for education in Korea is not
new, but after the 1997 economic crisis, Internet
skills were considered one of the most important
survival tools for many Koreans. They devote
their attention to the Internet due in part to the
Internet being a necessary technology for their
jobs, education, and entertainment. In particular,
for most parents broadband is a necessary tool
for childrens education (Jin, 2007). Relatively
simple initiatives such as encouraging school
teachers to post their homework assignments on
their own personal Web site and requiring students
to submit their assignments by e-mail could create a feeling among parents that the Internet is a
necessity for their childrens education (Choudrie
& Lee, 2004).
The majority of Koreans have become fast
adopters of high-tech, information-intensive products, and they are highly receptive to marketing
stories that offer education after the 1997 economic

crisis. Under the economic crisis, parents started

to realize the potential threats and opportunities of
globalization. It became a must for their children
to acquire both English and Internet skills in order
to survive in the era of globalization (Aizu, 2002).
Many parents who have children younger than
early teens, for example, have focused on cyberbased English education since the late 1990s. They
have acknowledged that their children are able to
enjoy online reading, music, game, and animation
written in English through the Internet due to
high speed and their being 24/7 support services,
which is a good alternative to learning English
in a classroom. In the Korean context, excessive
need for education has also developed so-called
edutainment, as a form of studying English while
enjoying playing in cyberspace (An, 2001).
Edutainment is not only for children. Many
mid-careers have also turned their eyes on the
Internet in the midst of a changing working environment. In Korea, the majority of companies
have begun to adopt a Western-style five-day
workweek system since the late 1990s and early 21st
century. The spreading five-day workweek system
since then has boosted information technology,
such as DVDs, mobile games, and edutainment.
With more spare time available, people began to
look for a new type of education containing more
entertainment factors, instead of the existing
cramming type of learning (Cho, 2002). Therefore,
having high-speed access to the Internet at home
became an advantage for everyone from children
to housewives to career persons, or a disadvantage
if they did not have it (Jin, 2007). Broadband is
an indispensable component for edutainment,
which is one of the major contributing factors in
broadband services.
Meanwhile, the peculiar Korean me-too
culture, which means all companies and people
end up with the same systems and taste, has substantially contributed to the spread of broadband
access. Consumers in various societies are susceptible to social pressure to keep up with their
neighbors, (e.g., keeping with the Joness in the

1205

Socio-Cultural Interpretations to the Diffusion and Use of Broadband Services in a Korean Digital Society

U.S.); however, Koreans eagerness to keep up with

their friends and neighbors is comparatively very
high (Jin, 2007). As discussed, Koreans tend to
be early adopters of technology; however more
they significantly are fast followers, as Stephen
Ward, consultant manager of the telecommunications group at Deloitte observed: Koreans are
always conscious of the need not to be left behind
by others, and the young have a great desire to
conform with the gadget-carrying norm of their
peers (Ward, 2004). The majority of Koreans
have not admitted any difference in several sectors. Businessmen should get the same salary if
they begin the job in the same company at the
same time, and parents cannot accept the fact that
their children are left behind others. They must
be in the best position in terms of the quality and
quantity of their houses, education, and income.
Korean people buy new refrigerators when their
neighbors buy them, and they buy pianos when
their neighbors buy them, although they have
refrigerators and pianos good enough to use.
Likewise, many Koreans are eager to have
high-speed Internet because they want to keep up
with their neighbors. They feel that if they have
the service, they could be part of the group; if not,
they could not be part of the group. Technology
and fashion fads tend to spread like wildfire in
Korea. If one individual adopts an innovation,
then others feel a social pressure to follow (Bajaj,
2002; Lau et al., 2005). The me-too culture, which
is despicable in some senses,8 works in broadband
services. Many old and new apartments are installing broadband Internet connections because the
construction companies and apartment complexes
do not want to be behind other apartment buildings (Cho Han, 2006). People do not want to be
left behind in broadband services because they
do not like a feeling of uneasiness as in many
other areas. The disdainful me-too culture has
ironically contributed to the rapid deployment
in broadband services.
Finally, a growing double personality between
social solidarity and individualism, in particular in

1206

the younger generation, has played a crucial role

in the rapid deployment of high-speed Internet. As
well known, Korea was a country which was proud
of its tradition of social solidarity (Crotty & Lee,
2002). People do not want to be left out. Due to
this, the New York Times reported in 2002, Korea
is a group-oriented society, where socializing in
bunches is the preferred form of interaction, and
Western-style individualism is frowned upon,
while reporting booming online games (French,
2002, p. 8). To take one documented case, after
conducting a case study of online games in Korea,
Chee (2006) also emphasized the importance
of social solidarity of teens and twenties in the
process of growing online game communities.
Indeed, it is very plausible because the flourish
of the online communication matched the culture
and emotion of the Korean people and their need
to communicate with one another through boards,
which are forums to discuss and exchange information, and clubs, which brought together people
with similar hobbies (Seong, 2006).
However, this conventional wisdom has
changed in recent years, and individualismthe
degree to which individuals are able to achieve
identity and status on their own rather than
through membership in groups (McFadyen et al.,
1998)is on the rise. In this regards, the New
York Times did not reflect the rapidly changing
environment in Korea. With the swift change in
the economy before and after the 1997 financial
crisis as well as recent modernization, Koreas
social and cultural situation has shifted on a large
scale. The younger generation increasingly seeks
Westernized culture, regarding an individuals
success, private property, and nuclear family as
important, instead of the value of the Confucian
culture emphasizing patriarchy and large families
(Jin, 2007). According to a marketing report by
SK Telecom, Those in their 20s and 30s, the first
Internet generation in Korea, gives priority to
their own time and space, rejecting all invasions
of privacy. They are also changing the culture of
collectivism to one of individualism (Joongang

Socio-Cultural Interpretations to the Diffusion and Use of Broadband Services in a Korean Digital Society

Ilbo, 2002, p. C3). The growing individualism

has resulted in a unique culture for developing
broadband services and demanding personal time,
space, and privacy.
The double personality of many Koreans, on
one hand social solidarity and on the other hand
growing individualism, has ironically contributed to the rapid growth of broadband services.
As Chee (2006) observes, young people enjoyed
popular online games sometimes with friends
and other times with nobody, but through highspeed Internet. This shows social solidarity on
the one hand; however, it also explains growing
individualistic mentality, which prefers enjoying games with no virtual friends in cyberspace
to playing with ones peers on the playground.
While seeking social solidarity, many young
people enjoy high-speed cyberspace by creating
their own world, and this changing social cultural
characteristic has become crucial for the rapid
growth of high-speed Internet.
In sum, the explosion of broadband in Korea
has been possible due in large part to various
cultural factors, rooted deeply in Korean society
and its historical context, in particular the 1997
economic crisis. Korean citizens are the most
significant component in broadband diffusion
because they are the main users of the service.
Without high-speed Internet, nobody can have a
normal socio-cultural life, including enjoyable
individual cyberspace from online games to
edutainment, in contemporary Korea, which is
one of the fastest growing information societies.
As Aizu (2002) points out, the social and cultural
factors, the aggressive mentality of the Korean
people, high awareness of the challenges of globalization, and political and historical contexts
played the decisive role in its dynamic acceptance
of the Internet and acceleration to broadband in
Korea.

CONCLUSION
Korea presents a unique example with its rapid
deployment of broadband penetration. Several
significant factors have contributed to the rapid
development of broadband Internet connections.
The government and telecommunications companies, as providers, have played important roles
in the rapid development of broadband Internet,
in particular to providing infrastructure for its
development. Favorable government policies
and competition among telecommunications
companies became driving forces for the rapid
deployment of broadband Internet.
The explosion of broadband in Korea, however,
was greatly made possible against the backdrop
of the 1997 economic crisis and due in large part
to various deeply rooted social, historical, and
cultural factors. Although political factors can
be important driving forces behind broadband
penetration, growth also requires an existing
receptiveness to using the services and applications that can be provided through broadband (Jin,
2007). In this regards, it was the citizens who actually made inroads into the worlds most shrewd
market for broadband services in Korea.
Several socio-cultural factors, which are crucial to the diffusion and use of new technologies,
have played significant roles in the swift deployment of broadband services. Cultural characteristics emphasizing quick communication and quick
responses, as well as enthusiasm for edutainment,
have contributed to the exponential growth of
broadband services. A growing complexity of
characteristics of the younger generation showing double personality of social solidarity and
individualism, which is not easy to find in any
other country, has particularly contributed to the
rapid growth in broadband services. If the younger
generation has only one characteristic of these

1207

Socio-Cultural Interpretations to the Diffusion and Use of Broadband Services in a Korean Digital Society

two, it could not be the strong cultural factor for

the rapid growth of broadband services. In the
midst of globalization that expedited with the
economic crisis in 1997, the younger generation
has developed its unique characteristics, which
resulted in one of the most significant cultural
factors in broadband Korea. The majority of Koreans have been eager to gain new technology, and
their unique cultural characteristics have become
the crucial initiative of the rapid deployment of
broadband services.
In conclusion, the introduction and development of new technology cannot be explained solely
in terms of political structures and economic
policy surrounding technology. As Ellul (1964)
stated: One should be looking at technology in
its sociological aspect because technology is not
an isolated fact in society but is related to every
factor in the life of modern man. Technology
does not develop independently, but is part of a
particular social-economic and cultural setup.
The deployment of broadband in Korea has
been possible due to its specific socio-cultural
dimensions as well as political and economic
environment. Cultural characteristics rooted in
the Korean society greatly influence the growth
of technology and broadband services throughout
the country.

REFERENCES

Bang, H.C. (2005). The second round of speed

war in high-speed Internet. Weekly Chosun.
Belson, K., & Richtel, M. (2003).Americas
broadband dream is alive in Korea. New York
Times, (May 5), C1.
Borgmann, A. (1984). Technology and the character of contemporary life. Chicago: University
of Chicago Press.
Chee, F. (2006). The games we play online and
offline: Making Wang-tta in Korea. Popular
Communication, 4(3), 225-239.
Chinese Cultural Connection. (1987) Chinese
values and the search for culture-free dimensions
of culture. Journal of Cross-Cultural Psychology,
18(2), 143-164.
Cho, H.Y. (2002). 5-day workweek to spotlight
new software sectors: SERI. The Korea Herald,
(November 9).
Cho Han, H.J. (2006). Internet and culture: About
the formation of subject in information society.
Retrieved September 2, 2006, from http://about.
haja.net/bbs/download.asp?ClubNumber=214
7483569&BoardType=3&file=119%08&code_
generation=Ecce_Homo
Choudrie, J., & Lee, H.J. (2004). Broadband
development in South Korea: Institutional and
cultural factors, European Journal of Information
Systems, 13(2), 103-114.

Aizu, I. (2002, April 22). A comparative study

on broadband in Asia: Development and policy.
Proceedings of the RIETI Symposium on Asian
Economic IntegrationCurrent Status and
Prospects.

Christians, C.G. (1989). A theory of normative

technology. In E.F. Byrne & J.C. Pitt (Eds.),
Technological transformation: Contextual and
conceptual implications (pp. 123-139). Dordrecht,
The Netherlands: Kluwer Academic.

An., J.H. (2001), English education special: Lets

play with English online. HanKook Ilbo, 27(August), 26.

Crotty, J., & Lee, K. (2002). A political-economic

analysis of the failure of neo-liberal restructuring in post-crisis Korea. Cambridge Journal of
Economics, 26(5), 667-678.

Bajai, V. (2002). High-sped connections common in South Korea. The Seattle Times, (October
21).

1208

Ellul, J. (1964). The technological society. New

York: Vintage.

Socio-Cultural Interpretations to the Diffusion and Use of Broadband Services in a Korean Digital Society

Feenberg, A. (1991). Critical theory of technology.

New York: Oxford University Press.

Kwon, K.D. (2006). How to stay ahead as digital

powerhouse. The Korea Herald, (February 23).

Franklin, U. (1999). The real world of technology.

Toronto: Anansi.

Lau, T., Kim, S.W., & Atkin, D. (2005). An examination of factors contributing to South Koreas
global leadership in broadband adoption. Telematics and Informatics, 22(4), 349-359.

French, H. (2002). Koreas real rage for virtual

games. New York Times, (October 9), 8.
Han, G.J. (2003). Broadband adoption in the
United States and Korea: Business driven rational model versus culture sensitive policy model.
Trends in Communications, 11(1), 3-25.
Ishii, K. (2003). Diffusion, policy, and use of
broadband in Japan. Trends in Communication,
11(1), 45-61.
ITU (International Telecommunications Union).
(2003a). Broadband Korea: Internet cast study.
Paris.
ITU. (2003b, September 16). Number of global
broadband subscribers grows 72% in 2002. Press
Release, ITU, Paris.
Jin, D.Y. (2005). Socio-economic implications
of broadband services: Information economy in
Korea. Information, Communication & Society,
8(4), 503-523.
Jin, D.Y. (2007). Hands on/hands off: The Korean
state and the market liberalization of the communication industry. Cresskill, NJ: Hampton Press.
Joongang Ilbo (English Edition). (2002). Generation 2030 went from apathy to passion. Joongang
Ilbo, (December 21), C3.
Kim, D.Y. (2006b). Hanaro, share the Wireline
Telecom with KT. Dong-A Ilbo, (January 2), 38.
Kim, P.H. (2006a). Is Korea a strong Internet nation? The Information Society, 22, 41-44.
Kim, T.G. (2005). South Korea remains king
of high-speed Internet penetration. The Korea
Times, (May 29).

Lee, J.S. (2000). Balanced cultural policy. Retrieved September 3, 2006, from http://www.kcpi.
or.kr/database/e_sosik/20000102/news_1.html
Lee, Y.K., & Lee, D. (2003). Broadband access in
Korea: Experience and future perspective. IEEE
Communications Magazine, 30-36.
Lee, H.J., OKeefe, R.M., & Yun, K. (2003). The
growth of broadband and electronic commerce in
South Korea: Contributing factors. The Information Society, 19(1), 81-93.
Lee, H.J., Oh, S., & Shim, Y.W. (2005). De we
need broadband? Impacts of broadband in Korea.
Info: The Journal of Policy, Regulation, and
Strategy for Telecommunications, Information
and Media, 7(4), 47-56.
Lim, H. (2001). Hype of education information
policy. Kookmin Ilbo, (April 25), 3.
McFadyen, S., Hoskins, C., & Finn, A. (1998). The
effect of cultural differences on the international
co-production of television programs and feature
films. Canadian Journal of Communication,
23(4), 523-538.
Mesthene, E. (1970). Technological change. New
York: Signet.
MIC (Ministry of Information and Communication). (2004). Broadband IT Korea vision 2007.
Seoul.
MIC. (2006). 2006 Korea Internet white paper.
Seoul.
Pacey, A. (1983). The culture of technology.
Cambridge: MIT Press.

1209

Socio-Cultural Interpretations to the Diffusion and Use of Broadband Services in a Korean Digital Society

Park, Y.S. (2006). High-speed Internet market is

going to June war. Munhwa Ilbo, (May 29), 16.
Parks Associates. (2005, November 21). Taiwan
and Korea lead in the adoption of consumer
technologies. Press Release.
Reynolds, T., & Sacks, G. (2003, April 9-11).
Promoting broadband: Background paper. Proceedings of the International Telecommunications
Unions Workshop on Promoting Broadband,
Geneva.
Rescher, N. (1969). What is value change? A
framework for research. In K. Baier & N. Rescher (Eds.), Values and the future. New York:
The Free Press.
Seong, N.Y. (2006). How has Korea become
Internet powerhouse? The Korea Times, (September 22).
Shim, M.S. (2006). Similar aspects between Korean soccer and Korean information technology.
HanKook Ilbo, (June 23).
Taylor, C. (2006). The future is in South Korea.
Retrieved June 14, 2006, from http://money.cnn.
com/2006/06/08/technology/business2_futureboy0608/index.htm
Ward, A. (2004). Where high-speed access is going
mainstream the Korean experience: Government
policy, high levels of urbanization. Financial
Times, (June 9), 4.
Webster, F. (2002). Theories of the information
society (2nd ed.). London: Routledge.
Williams, R. (1992). Television: Technology and
cultural form. London: Wesleyan University
Press.
Yun, K., Lee, H.J., & Lim, S.H. (2002). The growth
of broadband Internet connections in South Korea: Contributing factors. Paper, Asia/Pacific
Research Center, Stanford University, USA.

1210

KEY TERMS
1997 Economic Crisis: The Korean economy
has gone from being an example of one of the most
successful development experiences in modern
history to economic stagnation and decline in 1997.
From the middle of 1997, Korea was beset with
a series of financial crises. The trend of decades
of rising incomes reversed, and unemployment
and poverty were reaching alarming levels. In
particular, factors responsible for the decline in
the value of exports then include a dramatic fall
in the prices of some electronic and information
equipment, in particular semiconductors, which
had dire consequences for a number of countries
in the East and South Asia region.
Confucianism: A philosophy of life developed
by Confucius. It stressed the proper relationships
in society, such as father/son and ruler/subject.
The philosophies of Confucius are emphasizing
love for humanity; high value given to learning,
and devotion to family, peace, and justice.
Convergence: Also known as digital convergence. The concept that all modern information
technologies are becoming digital in nature. The
technological trend whereby a variety of different
digital devices such as TVs, in particular highdefinition TVs, and mobile telephones are merging into a multi-use communications appliance
employing common software to communicate
through the Internet.
Edutainment: The combination of education
and entertainment. Many people in the digital
age use the Internet as a form of studying several
subjects, in particular English while enjoying
playing in cyberspace.
Information Technology (IT): Compared to
the labor-led technologies and industries, information technology usually includes semiconductors,
computers, and telecommunications, although
some economists like to use the term knowledgebased industry in order to explain IT.

Socio-Cultural Interpretations to the Diffusion and Use of Broadband Services in a Korean Digital Society

Online Games: Internet games (also known

as online games) are games that are played online
via the Internet. They are distinct from video
and computer games in that they are normally
platform independent, relying solely on clientside technologies. Normally all that is required
to play Internet games are a Web browser and the
appropriate plug-in.
Social Solidarity: The degree or type of
integration of a society. In simpler societies, solidarity is usually based on kinship ties or familial
networks. In more complex societies, solidarity
is more organic, referring to the interdependence
of the component parts. Thus, social solidarity is
maintained in more complex societies through the
interdependence of its component parts.

ENDNOTES
1

For example, Yun et al. (2002) and Lee et

al (2003) mention that the strong emphasis
on education and academic performance has
prompted parents to turn to the Internet for
educational goods and services; however,
they do not extend their discussion further.
Since broadband penetration was above 77%
as of January 2006, Korea almost achieved
one of the major goals; however, the speed
of broadband, another major goal, could not
be fulfilled given that an average speed of
broadband was only 2.5Mpbs in May 2005
(see Kim, 2005).
In 1997, Korea experienced the worst economic recess in modern history, mainly
due to the crisis in financial and corporation sectors. Thirteen large conglomerates,
including Daewoo and Donga, collapsed
or became partially insolvent because of
the shortage of currency. The immediate
result of these companies collapse was the
suffering of small companies and workers.

When they failed, thousands of small-scale

companies collapsed. As of February 1998,
3,377 mid-size and small companies were
bankrupt. When companies went bankrupt,
workers lost their jobs. By the end of June
1999, the number of jobless reached 1.78
million. This means the unemployment rate
rose sharply from 2.6% in the fourth quarter
of 1997 to 8.5% in the second quarter of
1999, and real wages fell by 7.9% (see Jin,
2007).
The government placed an emphasis on
creating Internet-Friendly Classrooms at
every level of schooling by connecting them
to high-speed Internet free of charge as the
main component of educational information
infrastructure. The government spent $1.4
billion for the establishment of school information infrastructure between 1998 and
late 2000. Of this amount, the government
used $600 million to connect classrooms of
10,064 elementary, middle, and high schools
around the country with broadband services
(see Lim, 2001, p. 3).
The relatively low monthly subscription
fees have helped the growth of broadband
services; however, it is almost certain that
a high subscription fee could not deter the
spread of broadband services because the
majority of Koreans have not taken the price
as a factor. For example, most Koreans over
teens have had cell phones regardless of high
monthly subscription fees. The subscription
fees of mobile phones have greatly risen since
1997; however, the penetration rate has also
exponentially increased. In the late 1990s,
the Korean government officially banned
mobile carriers from offering handset
subsidies in an attempt to block excessive
marketing; however, many local mobile users
still wanted to have mobile phones, although
the full price of many handsets were priced
at more than $500 in 2002. There are major
factors other than promotion policies and

1211

Socio-Cultural Interpretations to the Diffusion and Use of Broadband Services in a Korean Digital Society

competition among telecom companies in

the Korean context which would be cultural
factors.
100Mbps broadband services can download
the equivalent of a 32-page newspaper per
second and a film (1.4 GB) within two minutes, compared to one hour with 3.5 Mbps
broadband services. About 80% of customers used broadband services of 1-4Mpbs in
Korea as of December 2005.
Parks Associates surveyed more than 10,000
households in 13 countries and ranked nations according to their proclivity to adopt

and use MP3 players, video-on-demand

(VOD), home networks, computers, online
services, and similar advanced technologies.
The survey found that Taiwan (Digital Living Index: 8.7) and Korea (8.4) lead the world
in the adoption of consumer technologies,
followed by the U.S. (8.1), Canada (8.0), and
Japan (7.9) (see Parks Associates, 2005).
It is in some ways despicable because Korean
people follow others, not because they need,
but because they want to be in the same class
and enjoy showing off.

This work was previously published in Handbook of Research on Global Diffusion of Broadband Data Transmission, edited by
Y. Dwivedi; A. Papazafeiropoulou; J. Choudrie, pp. 78-89, copyright 2008 by Information Science Reference, formerly known
as Idea Group Reference (an imprint of IGI Global).

1212

1213

Chapter 3.13

Mobility and Multimodal User

Interfaces
Christopher J. Pavlovski
IBM Corporation, Australia
Stella Mitchell
IBM T. J. Watson Research, USA

INTRODUCTION
Traditional user interface design generally deals
with the problem of enhancing the usability of a
particular mode of user interaction, and a large
body of literature exists concerning the design
and implementation of graphical user interfaces.
When considering the additional constraints that
smaller mobile devices introduce, such as mobile
phones and PDAs, an intuitive and heuristic user
interface design is more difficult to achieve.
Multimodal user interfaces employ several
modes of interaction; this may include text, speech,
visual gesture recognition, and haptics. To date,
systems that employ speech and text for application interaction appear to be the mainstream
multimodal solutions. There is some work on
the design of multimodal user interfaces for general mobility accommodating laptops or desktop
computers (Sinha & Landay, 2002). However,

advances in multimodal technology to accommodate the needs of smaller mobile devices, such
as mobile phones and portable digital assistants,
are still emerging.
Mobile phones are now commonly equipped
with the mechanics for visual browsing of Internet
applications, although their small screens and
cumbersome text input methods pose usability
challenges. The use of a voice interface together
with a graphical interface is a natural solution
to several challenges that mobile devices present. Such interfaces enable the user to exploit
the strengths of each mode in order to make it
easier to enter and access data on small devices.
Furthermore, the flexibility offered by multiple
modes for one application allows users to adapt
their interactions based on preference and on
environmental setting. For instance, handsfree speech operation may be conducted while
driving, whereas graphical interactions can be

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Mobility and Multimodal User Interfaces

adopted in noisy surroundings or when private

data entry, such as a password, is required in a
public environment.
In this article we discuss multimodal technologies that address the technical and usability
constraints of the mobile phone or PDA. These
environments pose several additional challenges
over general mobility solutions. This includes
computational strength of the device, bandwidth
constraints, and screen size restrictions. We
outline the requirements of mobile multimodal
solutions involving cellular phones. Drawing
upon several trial deployments, we summarize
the key designs points from both a technology and
usability standpoint, and identify the outstanding problems in these designs. We also outline
several future trends in how this technology is
being deployed in various application scenarios,
ranging from simple voice-activated search engines through to comprehensive mobile office
applications.

BACKGROUND
Multimodal interaction is defined as the ability to
interact with an application using multiple sensory
channels (i.e., tactile, auditory, visual, etc.). For
example, a user could provide input by speaking,
typing on a keypad, or handwriting, and receive
the subsequent response in the form of an audio
prompt and/or a visual display. Useful multimodal
applications can cover a broad spectrum including
tightly synchronized, loosely synchronized, and
complementary modes of operation. Synchronization behavior must be defined both for input
(the way in which input from separate modes
is combined) and for output (the way in which
input from one mode is reflected in the output
modes). The W3C distinguishes several types of
multimodal synchronization for input as follows
(W3C, 2003a):

1214

Sequential: Two or more input modalities

are available, but only a single modality is
available at any given time.
Simultaneous: Allows input from more
than one modality at the same time, but each
input is acted upon separately in isolation
from the others.
Composite: Provides for the integration of
input from different modes into one single
request.

A general framework for multimodal systems

is depicted in Figure 1. This diagram elaborates
further on several fundamentals positioned by
W3C.
The interaction manager is responsible for
combining multiple requests, dialog management,
and synchronization. The function of receiving
and combining multiple inbound requests is the
responsibility of the integration manager subcomponent. Conversely, the generation manager
is responsible for distributing multimodal output
to all of the respective output channels (modes)
via an interpretation layer, which may involve
text to speech (TTS) conversion or transcoding
of graphical content to accommodate the needs of
the target modality. Earlier work in multimodal
systems referred to the integration tasks relating
to composition and decomposition of requests as
fusion and fission respectively (Coutaz, Nigay, &
Salber, 1993).
Speech-based telephone interfaces currently
available in the commercial market commonly
use varying levels of directed dialog. Directed
dialog, as the name implies, employs a style of
system prompts that helps to direct the user
in what to say next. Users are often presented
with spoken menu options from which they can
make a selection, thus navigating in a controlled
manner until the task is completed. Much of the
naturalness and power of speech is undermined
when the application relies too heavily on the use
of directed dialogs. A Natural Language speech
interface, which allows the user to phrase their

Mobility and Multimodal User Interfaces

Figure 1. Multimodal framework

Interpretation
ASR

Pen

HR

Keyboard

Interaction Manager
Integration
Manager
(Fusion)

Modality n
Modality n
Text
Speaker

TTS

Graphics

Transcode

request in a wide variety of ways, reduces the

cognitive load since there are no commands to
memorize or hierarchies to navigate. A mixedinitiative interface allows the user to share control over the direction of the dialog, making the
interaction more efficient for the user.
Device manufacturers can install specialized software or firmware on handsets to enable
distributed speech recognition (DSR). DSR technology digitizes the speech signal and sends it
over an error-protected data channel to a speech
recognizer on a server. Thus the processing is
distributed between a terminal client and a server.
The accuracy of speech recognition can be better when using DSR because the degradations
associated with sending speech over the mobile
network, such as low bit rate speech coding and
channel transmission errors, are avoided. In addition, DSR allows for the implementation of
a multimodal, speech and data, application on
devices which do not support simultaneous voice
and data connections.

MULTIMODAL TECHNOLOGY
IN MOBILITY SYSTEMS
Multimodal applications for small mobile devices
must overcome several technical challenges;

Generation
Manager
(Fission)

Dialog,
Session, and
Compound
Management

Multimodal Application

Device
Microphone

these include device capability and computational

strength, functional and bandwidth constraints of
the network, and limitations of the user interface.
Present work in multimodality is focused upon the
support of two modes of interaction, most typically
data (graphics or text) and speech (Kondratova,
2004; Pavlovski, Lai, & Mitchell, 2004a; Hastie,
Johnston, & Ehlen, 2002; Kvale, Warakagoda, &
Knudsen, 2003). Future trends support additional
modes that include visual gesture, lip reading,
and haptic responses. In this section we present
advances in multimodality supporting speech
and data entry, outlining the current state of the
technology used and the outstanding challenges
yet to be addressed.

Multimodal Architectures
Due to the need to support two or more modes of
input and output, the solutions to support multimodal systems are more complex than unimodal
systems. Additional capabilities are required to
support composite input and output requests,
manage multiple application states, and perform
session management between devices and the
multimodal application services.
There are fundamentally two architectural
approaches to constructing multimodal solutions
for mobile devices such as mobile phones and

1215

Mobility and Multimodal User Interfaces

Figure 2. Distributed client multimodal architecture

Server

TTS
GUI

Integration
Manager

Transcode

Interpretation

Channel Decoding

HR

Channel Encoding

ASR

PDAs. The most widely investigated architecture

appears to involve deployment of an application
onto the mobile device, using distributed speech
recognition (DSR) and a GUI client application.
Such a solution may be categorized as distributed
client architecture (i.e., thick client architecture).
The alternative employs the browser-based architecture (thin client), where a standard Web
browser is the only client software used.
Several proposed solutions have been studied that support distributed client architectures
(Kvale, Narada, & Warakagoda, 2005; Klante,
Krsche, & Boll, 2004). Figure 2 provides an
overview of the key sub-systems and components
of the distributed client architecture. The device
is required to host several components of the
solution, generally including a multimodal client
application and distributed multimodal processing technologies. This may include automated
speech recognition (ASR), text to speech (TTS),
and handwriting recognition (HR); however,
typical distributed client designs to date have
such components on the server. Communication
between the client and server occurs over a single
channel; hence the multimodal input is marshaled
(encoded) into one communications stream.
There are several advantages of this type of architecture including true synchronicity in delivery
of the multimodal output, distributed processing,
and richer user interface. Since the multimodal
output is combined before transmission to the

1216

Generation
Manager

Dialog,
Session, and
Compound
Management

Multimodal Application

Device

client, the client application is able to ensure

media synchronizationthat is, synchronized
presentation of content to the user; this is a key
problem to address in multimodal systems. The
presence of a client application also provides a
richer graphical user interface and consistency
in user interface output. Additionally, the capability to distribute some processing tasks to the
client provides greater flexibility for multimodal
applications, particularly in relation to response
time. For instance, a multimodal short message
service is able to perform speech recognition locally within the device, rather than returning to
the server, and display the utterance to the user
before sending as a short text message.
A key drawback to distributed client architectures is the need to manage the client software to
be deployed to the mobile device. This is appears
to be of particular concern to mobile operators for
several reasons. Given the significant numbers of
mobile phones in use, software distribution and
management to the device is costly. The majority
of mobile phone plans are 12 or 18 months, hence
new devices are acquired rapidly and continual
support becomes cost prohibitive. Additionally, it
may not be practical to support all devices available on the market.
Further research suggests that while a limited
graphical user interface is of concern to users, in
fact the major contributor to improved usability
of multimodal user interfaces is the speech inter-

Mobility and Multimodal User Interfaces

Figure 3. Browser multimodal architecture

Microphone

Modality n

Bearer n

face (Hastie et al., 2002; Pavlovski et al., 2004a).

There is also much work on the notion of speechcentric multimodal user interfaces (Johnsen &
Kvale, 2005). These observations also support
the notion that alternative architectures require
consideration.
An alternative approach is a class of multimodal system that does not require client applications to be deployed to the mobile device. Such a
framework is also termed a browser-based, or thin
client, architectureonly requiring a standard
browser on the mobile phone. In order to support
the multiple channels, typically speech and data,
use is made of a Class A mobile phone and the
multi-call supplementary service, which is only
available in 3G networks. Specifically, each device
establishes several bearer channels with the server,
one per mode of user interaction. For example,
simultaneously a voice circuit is established for
speech and a data bearer service for the browsing
graphical and text content (see Figure 3).
Thin client architectures have been studied for
mobile devices. An initial set of building blocks
to support browser architectures for sequential
multimodal processing was outlined in Niklfeld,
Finan, and Pucher (2001). The approach in Chou,
Shan, and Li (2003) and Li, Wong, and Guo (2004)
supports a thin client architecture, where use is
made of the multi-call supplementary service for
simultaneous connections between the mobile
phone and the multimodal application server;

Integration
Manager

Generation
Manager

Dialog,
Session, and
Compound
Management

Multimodal Application

Voice Cct

Interpretation

Speaker

Xcode

Browser

Server
GPRS

Multicall Service Manager

Device

however, only sequential multimodal input and

output is supported. A further solution solves the
problem of supporting composite multimodal
output response (Pavlovski, Wood, Mitchell, &
Jones, 2004b), however only sequential simultaneous multimodal input is supported. The MONA
project provides a similar capability for thin
client, and extends the device range to support
both mobile phones and PDA devices using a
multimodal presentation server (Anegg, Dangl,
& Jank, 2004).
The browser-based architecture would naturally seem more attractive to mobile phone operators
given the dynamic and rapid turnover of mobile
phones, hence eliminating the need to manage
software distribution and complexity attributed to
supporting numerous mobile devices. The disadvantage of course is that coordinating additional
channels increases the complexity and introduces
the possibly of latency or synchronization errors.
One paper points out that media synchronization
is an impact to multimodal applications (Pavlovski
et al., 2004b). This may be accommodated by
delaying one of the responses so that both arrive
simultaneously at the client device. However,
complex multimodal applications are yet to be
studied to comprehensively assess the impact of
synchronization.
Several key challenges remain for browserbased multimodal architectures. This includes
management of channel latency and media syn-

1217

Mobility and Multimodal User Interfaces

chronicity. A full implementation that supports

both input and output multimodal composite
requests remains as an outstanding problem. And,
support for three or more multimodal channels
is yet to be explored. These problems may be
solved by using the distributed client architecture
and have been demonstrated (Johnsen & Kvale,
2005)however, at the cost of introducing additional software management overheads and
restrictions on device supportability.

Standards and Technology

The W3C organization has defined several
multimodal standards, including interaction requirements and an interaction framework (W3C,
2003b). We briefly summarize some further technologies that are specific to multimodal solutions,
outlining their applicability to mobile devices.

Extensible Multimodal Annotation

(EMMA)
EMMA is a set of XML specifications that represents multimodal user input including speech,
pen, and keystroke. The XML is intended to be
generated automatically, hence is suitable for use
between recognition engines and the interaction
manager.

Multimodal Presentation
Markup Language (MPML)
The MPML Language is designed for presenting
multimodal output on browsers supporting XML.
The limited processing capability of phones is
accommodated, with a mobile edition defined
for J2ME applications (Saeyor, Mukherjee, Uchiyama, & Ishizuka, 2003). As such, the architecture
resembles a distributed client solution.

1218

Speech Application Language Tags

(SALT)
SALT is a set of extensions to HTML (or XHTML
and WML) adding speech capability to these
mark-up languages. While it is suggested to
support the thin client architecture, applicability
to mobile phones is limited due to the requirement of browsers support for SALT. The need
to manage multiple application versions, one
for each device, is also observed as a drawback
(Kondratova, 2004).

XHTML+Voice (X+V)
X+V is a proposed standard for multimodal
markup. Designed for clients that support spoken
and visual interaction, X+V technology furnishes
traditional Web pages with further voice tags
for input and output speech tasks. A goal is to
support thin client, however in its current form
this technology is most suited to the distributed
client architecture. Use of VoiceXML has also
been studied (Niklfeld et al., 2001).

Synchronized Multimedia
Integration Language (SMIL)
SMIL is a relevant standard to managing media
synchronization, an important multimodal problem. However, the standard is largely aimed at
conventional browser technology that supports
XML, DOM, and XHTML. Hence, it is restricted
to browsers that may not be deployed to mobile
phones.

User Interface Design

There is some work treating multimodal user
interface design in mobility. A framework for

Mobility and Multimodal User Interfaces

multimodal interface design has been proposed

in Baillie, Simon, Schatz, Wegscheider, & Anegg,
2005). The authors suggest the need to study user
behavior in the natural environment. This follows
the observation that user behavior differs considerably when comparing a controlled environment
to the general freedom of use that mobile devices
provide (Baillie & Schatz, 2005). A key feature
of the method is a table designating the context
in which the multimodal system is used and the
users preferred mode of interaction. Further
work extends these concepts by proposing a tool
to more fully analyze user behavior over an extended period of time (Salembier, Kahn, Calvet,
Zouinar, & Relieu, 2005).
Given that there is little data on user behavior
in a mobile environment, early design efforts will
require careful consideration with users on how
to maximize multimodal capabilities. Several
key factors determine user behavior including
response to error recovery, the course of action
context, and properties of the implementation of
each modality (Calvet, Julien Kahn, Pascal Salembier, & Zouinar, 2003). Further work shows that
the speech interface appears to be the major contributor, when complementing a text or graphical
interface, to the improvements bestowed through
multimodality (Pavlovski et al., 2004b).
Summarizing the literature, the key features
that a multimodal user interface design must
address include the capability to dynamically
alter modes, support for simultaneous composite
output, and the ability to review the output request
on one mode, using an alternative mode prior to
actioning the request; for example, users who may
dictate a message would prefer to see a transcript
of the message prior to sending.

FUTURE TRENDS
There are several industry trends and future areas
of research including the extension of multimodal
solutions to accommodate haptic response, vi-

sual gesture recognition, lip reading, and speech

technologies that support natural language understanding. Oviatt (2003) reviews several of these
emerging technologies.
The idea of haptic response is already in use
with present-day mobile phones, where mobile
phones offer a silent mode that alerts the user
with a vibration event. Such stimuli has been in
use for some time and appear with early flight
control systems that provide feedback to the
pilots control system indicating turbulent flight
conditions. More recently, gaming controllers use
such feedback. There is some recent work on the
design of haptic response (Kaaresoja & Linjama,
2005) in mobile devices, while other work focuses
on haptic input to control a mobile device (Oakley
& OModhrain, 2005). The notion of extending
applications in mobility with such haptic responses
is relatively new. To extend such a capability to
multimodal applications, access to native mobile
services such as vibration is required to initiate
a haptic response.
Lisetti et al. (2003) describe research progress
on a multimodal emotion recognition system.
They have developed a prototype that takes input
from multiple modes (i.e., camera, microphone,
wearable computer) and includes both subjective, as expressed by the user, and physiological
components as inputs to its model. Lip reading
technology, as an alternative interaction mode,
may be used to improve speech recognition. Prior
research suggests that bimodal audiovisual speech
recognition can improve as much as 40% in accuracy (Chan, 2001), in comparison with using
audio input alone. Additional work on the use of
Natural Language understanding to augment the
multimodal user experiences has been proposed
and studied (Pavlovski et al., 2004b).
A final observation is required regarding security and privacy. Traditionally, voice communications are considered relatively secure, due to the
transient nature of circuit establishment between
parties and the real-time exchange of data between
two peer entities. This is no longer the case for

1219

Mobility and Multimodal User Interfaces

multimodal application, where the utterances are

digitally recorded and stored. Furthermore, in
order to improve the speech accuracy and fault
resolution, there is a desire on the operators part
to store such data for a perceived user benefit. Of
course the storage of personal conversations and
requests for information raises several security and
privacy issues that have yet to be addressed.

CONCLUSION
The literature has shown that multimodal user
interfaces are able to provide a superior user
experience when interacting with multimodal
applications on mobile devices such as cellular
phones. Furthermore, the applicability of multimodality is extending beyond traditional mobile
Web applications to a greater range of domains.
For instance, due to the plurality of user interface
modes, multimodal systems provide a natural
improvement for people with disabilities (Kvale
et al., 2005). Although studies demonstrate that
multimodal user interfaces provide a superior user
experience on a mobile phone, it remains unclear
whether such an improvement remains consistent
for all types of applications.
Distributed client architectures are able to overcome several problems identified in the literature,
however the practicality of deployment for mobile
phones requires consideration. Browser-based
multimodal architectures overcome deployment
issues and provide an alternative and easier
mechanism for allowing mobile devices to access
services. In order to support multimodal applications that make use of three or more interaction
modes, the use of thin client architecture is as yet
unproven. In addition, further capabilities that may
not be inherent within the devices are required,
such as local composite management. Hence,
distributed clients would seem a necessary choice
to support these capabilities and appear most appropriate to address the needs of the increasing
complexity due to several modes of interaction,

1220

particularly as the media synchronization of

modes becomes more complex.

REFERENCES
Anegg, H., Dangl, T., & Jank, M. (2004). Multimodal interfaces in mobile devicesthe MONA
project. Proceedings of the Workshop on Emerging Applications for Mobile and Wireless Access
(www2004 Conference), New York.
Baillie, L., & Schatz, R. (2005). Exploring
multimodality in the laboratory and the field.
Proceedings of the 7th International Conference
on Multimodal Interfaces (ICMI 2005) (pp. 100107). Torento, Italy.
Baillie, L., Simon, R., Schatz, R., Wegscheider, F.,
& Anegg, H. (2005). Gathering requirements for
multimodal mobile applications. Proceedings of
the 7th International Conference on Information
Technology Interfaces (ITI 2005), Dubrovnik,
Croatia (pp. 240-245).
Calvet, G., Julien Kahn, J., Pascal Salembier, P.,
& Zouinar, M. (2003). In the pocket: An empirical
study of multimodal devices for mobile activities.
Proceedings of the HCI International Conference,
Crete, Greece, (pp. 309-313).
Chou, W., Shan, X., & Li, J. (2003). An architecture
of wireless Web and dialogue system convergence
for multimodal service interaction over converged
networks. Proceedings of COMPSAC 2003, Dallas, TX, (p. 513).
Coutaz, J., Nigay, L., & Salber, D. (1993). Taxonomic issues for multimodal and multimedia
interactive systems. Proceedings of the Workshop
on Multimodal Human-Computer Interaction
(ERCIM 93) (pp. 3-12). Nancy, France.
Hastie, H., Johnston, M., & Ehlen, P. (2002). Context-sensitive multimodal help. Proceedings of the
4th IEEE International Conference on Multimodal
Interfaces (ICMI 02), Pittsburgh, PA, (p. 93).

Mobility and Multimodal User Interfaces

Johnsen, M. H., & Kvale, K. (2005). Improving

speech centric dialogue systemsThe BRAGE
project. Proceedings of the Norsk Symposium
on Signal Handling (NORSIG 2005), Stavanger,
Norway.
Kaaresoja, T., & Linjama, J. (2005). Perception
of short tactile pulses generated by a vibration
motor in a mobile phone. Proceedings of the 1st
Joint Eurohaptics Conference and Symposium on
Haptic Interfaces for Virtual Environment and
Teleoperator Systems (WHC05) (pp. 471-472).
Pisa, Italy.
Klante, P., Krsche, J., & Boll, S. (2004).
AccesSightsA multimodal location-aware
mobile tourist information system. Proceedings
of the 9th International Conference on Computers for Handicapped Persons (ICCHP 2004) (pp.
287-294). Paris.
Kondratova, I. (2004). Speech-enabled mobile
field applications. Proceedings of the Conference
on Internet and Multimedia Systems and Applications (IMSA 2004), Kauai, HI.
Kvale, K., & Warakagoda, N. (2005). A speech
centric mobile multimodal service useful for dyslectics and aphasics. Proceedings of EuroSpeech
2005, Lisboa, Portugal, (pp. 461-464).

Niklfeld, G., Finan, R., & Pucher, M. (2001). Architecture for adaptive multimodal dialog system
based on VoiceXML. Proceedings of EuroSpeech
2001, Aalborg, Denmark, (pp. 2341-2344).
Oakley, I., & OModhrain, S. (2005). Tilt to scroll:
Evaluating a motion based vibrotactile mobile
interface. Proceedings of the 1st Joint Eurohaptics
Conference and Symposium on Haptic Interfaces
for Virtual Environment and Teleoperator Systems
(WHC05) (pp. 40-49). Pisa, Italy.
Oviatt, S. L. (2003). Advances in robust multimodal interface design. IEEE Computer Graphics
and Applications, 23(5), 62-68.
Pavlovski, C. J., Lai, J., & Mitchell, S. (2004a).
Etiology of user experience with Natural Language speech. Proceedings of the 8th International
Conference on Spoken Language Processing
(ICSLP/INTERSPEECH 2004), Jeju Island, Korea,
(pp. 951-954).
Pavlovski, C. J., Wood, D., Mitchell, S., & Jones,
D. (2004b). Reference architecture for 3G thin
client multimodal applications. Proceedings of
the International Symposium on Communications
and Information Technologies (ISCIT 2004), Sapporo, Japan, (pp. 1192-1197).

Kvale, K., Warakagoda, N., & Knudsen, J. (2003).

Speech centric multimodal interfaces for mobile
communication systems. Telektronikk, (Vol. 2,
pp. 104-117). Temanummer: Spoken Language
Technology.

Saeyor, S., Mukherjee, S., Uchiyama, K., &

Ishizuka, M. (2003). A scripting language for
multimodal presentation on mobile phones.
Proceedings of the 4th International Workshop
on Intelligent Virtual Agents (IVA 2003) (pp.
226-230).Kloster Irsee, Germany.

Kvale, K., Warakagoda, N., & Kristiansen, M.

(2005). Evaluation of a mobile multimodal service
for disabled users. Proceedings of the 2nd Nordic
Conference on Multimodal Communication (Multimod Com 05), Gothenburg, Sweden.

Salembier, P., Kahn, J., Calvet, G., Zouinar, M.,

& Relieu, M. (2005). Just follow me. Examining
the use of a multimodal mobile device in natural
settings. Proceedings of the HCI International
Conference, Las Vegas, NV.

Li J., Wong W., & Guo, W. (2004). Case study of

a multimedia wireless system. Proceedings of the
IEEE International Conference on Multimedia
and Expo, (ICME 2004) (pp. 1815-1818). Taipei,
Taiwan.

Sinha, A. K., & Landay, J. A. (2002). Embarking

on multimodal interface design. Proceedings of
the 4th IEEE International Conference on Multimodal Interfaces (ICMI 2002) (pp. 355-360),
Pittsburgh, PA.
1221

Mobility and Multimodal User Interfaces

W3C (World Wide Web Consortium). (2003a,

January 8). Multimodal interaction requirements.
Retrieved from http://www.w3.org/TR/mmireqs/
W3C (World Wide Web Consortium). (2003b, May
6). Multimodal interaction framework. Retrieved
from http://www.w3.org/TR/mmi-framework/

KEY TERMS
Automated Speech Recognition (ASR): The
use of computer processing to automatically translate
spoken words into a text string.

Directed Dialog: Speech interface where

the user receives system prompts that direct the
user on available options that usually require a
response with simple spoken words.

Fission: The decomposition of output from a

multimodal application, to be distributed to the
different modalities.
Fusion: The composition of input from
multiple modalities into one input request to the
multimodal application.
Multi-Call Supplementary Service: Enables
several simultaneous circuit-switched calls to the
same device, each call using its own dedicated
bearer. Only one circuit-switched bearer can be
used for speech at any one time.
Multimodal Interaction (MMI): W3C term
for a class of human to computer interaction that
involves multiple modes of user interaction, such
as speech, writing, text, and haptics.
Text To Speech (TTS): The use of computer
processing to transform text into spoken words.

This work was previously published in Encyclopedia of Mobile Computing and Commerce, edited by D. Taniar , pp. 644-650,
copyright 2007 by Information Science Reference, formerly known as Idea Group Reference (an imprint of IGI Global).

1222

1223

Chapter 3.14

Foreseeing the Future Lifestyle

with Digital Music:
A Comparative Study Between Mobile
Phone Ring Tones and Hard-Disk
Music Players Like iPod
Masataka Yoshikawa
Hakuhodo Inc., Japan

Abstract
This chapter aims to explore the future trajectory
of enjoying digital music entertainment among
consumers comparing the characteristics of the
usage patterns of digital music appliances in the
U.S. and those in Japan. As the first step of this
research, the author conducted two empirical
surveys in the U.S. and Japan, and found some
basic differences in the usage patterns of a variety of digital music appliances. Next, a series of
ethnographical research based on focus-group
interviews with Japanese young women was done
and some interesting reasons of the differences
were discovered. In Japan, sharing the experiences
of listening to the latest hit songs with friends by
playing them with mobile phones that have the high
quality, ring tone functions can be a new way of
enjoying music contents, while hard-disk music

players like iPod have become a de facto standard

of the digital music appliances in the world.

Introduction: Central
Questions
The November 2001 debut of iPod and the subsequent opening of iTunes Music Store have brought
a rapid expansion of the digital music market
around the world. Some estimate that the market
will be worth $1.7 billion dollars by 2009 (Jupiter
Research). Now, iTunes Music Store service is
available in 30 countries around the world, with
the total number of downloaded songs surpassing
the 500 million mark in July 2005.
The store only opened in Japan in August 2005
and sold over 1 million songs in the first 4 days.
This is an astonishing achievement, consider-

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Foreseeing the Future Lifestyle with Digital Music

ing that Japans largest online music store Mora

has monthly sales of around 450,000 songs. In
March and April 2005, SONY, which has long
led the portable music player market, released
a new digital music player under the Walkman
brand, offering both the hard disk type and USB
flash memory type to launch a marketing drive
against iPod. The developments have finally
begun to provide Japanese music lovers with an
environment whereby digital music contents are
broadly enjoyed in terms of both services and
hardware devices.
One of the major characteristics of Japans
digital music market has been the presence of
digital music contents for use on mobile phones.
The use of digital music contents on mobile
phones, which started as regular ring tones, has
gradually evolved into Chaku-uta (true-tone
ring tones) by December 2002, and to Chaku-uta
Full (mobile-phone-based music distribution
service launched in December 2004 by the mobile
carrier au). Chaku-uta and Chaku-uta Full
have sold over 100 million songs and 10 million
songs respectively, making the digital music
service the largest segment in mobile-phone
content services.
The environment for enjoying digital music
content is set to expand even further into the
future. How would such a development affect
the way Japanese music fans listen to music in
general? This paper examines future ways of
enjoying digital music content in Japan, and the
competition between music players like iPod for
use with personal computers and mobile phones
that have adopted the usage as music players.

Japans Digital Music Content

Market and the
Proliferation of Mobile
Phones Before 2005
Firstly, let us examine past developments of the
digital music content market in Japan. Japans first

1224

digital music distribution service started in April

1997. A company called MUSIC.CO.JP began
offering songs mainly from independent labels.
Coinciding with the launch of numerous music
download services in the U.S., a number of online
music Web sites opened one after another, orchestrated by individual artists and record labels. In
December 1999, SONY Music Entertainment
became the first major record company to start an
online music store bitmusic. Toshiba EMI, Avex
Records, and other major companies followed
suit. Yet, since early 2005, the system for online
distribution of digital music contents has been
underdeveloped, as can be seen in the fact that
Moras supposed largest online music catalog in
Japan contained just 100,000 songs, as opposed
to iTunes Music Stores lineup of 1 million songs
upon its launch in August in Japan.
There is no denying that mobile-phone-related
music services have been the driving force of the
nations digital music market. The launch of the
i-mode service by NTT DoCoMo in February
1999 marked the start of digital content downloading services via mobile phones. The connection
speed of 9600bps in those days made it, initially,
difficult to distribute songs in high audio quality. Faced with the adversity, businesses began
offering Chaku-melo music ring tones, instead
of distributing actual music contents, achieving
dramatic growth. The Chaku-melo market has
rapidly expanded to 80-90 billion yen in 2002.
What makes this development unique was the
fact that this service was initiated not by record
companies rather by major online Karaoke service
providers like GIGA and XING, computer game
companies like SEGA, and other companies operating in the peripheral areas of the music industry
itself. The market size of 80-90 billion yen as of
2002 is among the highest of all mobile-related
digital content services, proving the market-led
proliferation of digital content services for mobile
phones.
Amidst the flourishing success of the Chakumelo market, supported by peripheral music

Foreseeing the Future Lifestyle with Digital Music

businesses, record companies that lead the music

industry initiated a move to provide the Chakuuta service, offering true-tone music as ring
tones, instead of Chaku-melo MIDI-based ring
tone melodies. The service was initially started
solely by Japans second largest mobile carrier au
in December 2002. While the Chaku-melo service
was employed by all mobile carriers rapidly, the
Chaku-uta service was not adopted by the industry leader NTT DoCoMo until February 2004
and by the number three mobile carrier Vodafone
until March 2004. However, the service picked
up substantial support from younger generations.
As the preceding proliferation of the Chaku-melo
service had already familiarized mobile phone
users with the concept of obtaining music over
mobile phone, Chaku-uta sales reached 100
million songs by July 2004, and surpassed 200
million songs by April 2005 to establish a market
of substantial scale. Record companies joined
forces to establish Label Mobile, which currently
provides around 300,000 songs, approximately
three times the catalog size of computer-based
online music stores.
After Chaku-uta came the Chaku-uta Full
service, which provides whole songs as ring tones
to become a de facto digital music store over mobile phones. It reached its fifth million download
in April, just 6 months after its launch in October
2004. The cumulative total of downloaded songs
reached 10 million in June, causing a dramatic
expansion in market size. Although the number
of songs available remains smaller than Chakuuta at 37,000, the catalog is expected to keep
on expanding.
As described thus far, the digital-music-content industry has rapidly mushroomed as one
of mobile phone services, but it has been less
than 1 year since a full-scale music distribution
service (Chaku-uta Full) was launched. Music has been merely distributed as an additional
function to mobile phones, that is, the ring tone.
Consumption has been initiated by mobile phone
use, instead of music itself. In other words, an

explosive proliferation of a new communications

device called mobile phones, has triggered the
consumption of digital music content as a natural
course of evolution. Amidst this situation, a series
of dedicated digital music players called iPod has
emerged with major success, triggering the launch
of the iTunes Music Store offering downloads of
digital music content. With the development of
a fully fledged environment for computer-based
consumption of digital music contents, what has
been the course of competition between different
types of devices in todays digital music content
market? Let us examine the overview based on
the results of a quantitative survey.

Todays Digital Music Content

Markets: Japan and U.S.
Markets
In order to grasp the state of todays digital music
content market in Japan, we have simultaneously
conducted a survey consisting of identical questions for use in both Japan and the U.S. Results
from the two countries were compared against
each other in order to identify characteristics of
the Japanese market. The survey was titled Survey on Digital Contents, and the survey period
it ran online was between February and March
2005.1 The following samples were included:
Japan, N=677 aged 15-59 and in the U.S., N=700
aged 18-59.
First, let us take a look at the rate of musicterminal use in the two countries: 6.9% of Japanese respondents used hard-disk music players
like iPod, whereas the ratio was almost double
at 11.5% in the U.S. The ratio of people using
USB flash-memory music players was 7.2% in
Japan and 16.1% in the U.S., more than double
the Japanese figure. However, the ratio of those
using mobile phones as music players was 19.8%
in Japan, nearly three times the U.S. result of 6.1%.
These figures demonstrated a clear tendency of
U.S. users opting for hard-disk or flash-memory

1225

Foreseeing the Future Lifestyle with Digital Music

devices with music transferred via computers,

and Japanese users choosing mobile phones to
listen to music.
Next, the survey examined how samples typically downloaded digital music contents: 28.9%
of U.S. respondents have downloaded music via
computer, over 10 percentage points higher than
the Japanese ratio of 17.4%. On the other hand,
42.2% have downloaded music (music ring tones)
over mobile phones in Japan, around three times
the equivalent U.S. figure of 14.2%. The ratio
of people who have downloaded true-tone ring
tones was 20.4% in Japan, an astonishing lead of
around seven fold compared to the U.S. result of
3.0%. The clear tendency of computer orientation
in the U.S. and mobile phone orientation in Japan,
observed in the choice of music-playing terminals,
was also evident in terms of the practice of music
downloading.
As explained in the previous section, these
findings are a natural outcome reflecting how the
digital-music-content market emerged and established itself around mobile phones from the early
days in Japan in contrast to market development
that evolved around computers and the Internet
in the U.S. However, there is some interesting
datathe survey asked those who do not own
a portable digital music player which type of
device they would like to possess. The results
indicated almost identical tendencies between
Japanese and U.S. respondents, unlike the stark
differences they demonstrated in previous questions. Those who intend to purchase a hard-disk
music player accounted for 26.7% in Japan and
26.1% in the U.S. The figures for flash-memory
players were also very similar at 21.4% in Japan
and 21.7% in the U.S. Finally, the ratio of those
using a mobile phone as a music player is 5.3%
in Japan and 3.0% in the U.S. Even though the
Japanese figure is slightly higher than the U.S.
figure, they can be viewed as almost at the same
level, in comparison to the ratio gaps observed
in other questions.

1226

This data demonstrates a strong contrast to

previous data, which showed a noticeable computer orientation for the U.S., and mobile phone
orientation for Japan. In both countries, purchase
intention appears higher for computer-based music
players based on either hard disk or USB flash
memory, and relatively low for mobile phones
doubling as music players.
Until now, Japans digital-music-content market has been characterized, in terms of hardware,
with proliferated use of mobile phones, as opposed
to the U.S. market where more users download
music via computer. However, as the results of the
aforementioned survey suggest, computer-based
music players will be used increasingly for the
consumption of digital music content, in addition
to mobile phones, in the future Japanese market.
Then, what changes will emerge in consumers
music playing styles when such hardware competition (spread of computer-based music players)
evolves?

Future Style of Digital Music

Listening in Japan: Overall
Trend
We have projected future changes in the style
of digital music listening in Japan, dividing the
samples of the aforementioned survey into the
following three groups and comparing the profiles, music listening styles, and mentalities of
the current and future users.
1.

2.

3.

Current digital music content usersThose

who own hard-disk/USB flash-memory
music players N=42.
Digital music content potential usersThose
who intend to purchase hard-disk/USB
flash-memory music players N=307.
Nonusers of digital music contentsThose
excluding the above two groups N=319.

Foreseeing the Future Lifestyle with Digital Music

We made a particular comparison between

current digital music content users who have
constituted the computer-oriented digital music
market, and digital-music-content potential users
who intend to join the market from now, so as to
identify how the listening style of this market is
likely to change, and what impact such changes
will have on the market, which has evolved through
downloading by means of mobile phones thus far.
First, we compared samples demographic profiles
and basic indicators in music consumption.

Gender Comparison
Current digital-music-content users mainly consist of men, accounting for 66.7%, as opposed to
women at 33.3%. Digital-music-content potential
users have a more even gender distribution, consisting of men and women at respectively 54.4%
and 45.6%. Nonusers of digital music contents
have a greater proportion of women at 58.4%,
compared to men at 43.6%.

Comparison by Gender and

Generation
The generation factor was then incorporated to
characterize the three groups more clearly. Among
current digital-music-content users, men in their
20s claimed the largest proportion at 29.6%, followed by women in their 20s at 23.8%, and men
in their 30s at 16.7%. These three groups alone
represent over 70% (70.1%), indicating that digital
music content is primarily enjoyed among younger
peopleboth men and women. In comparison,
among digital-music-content potential users, men
in various age groups accounted for around 10%
each, that is, men in their 20s at 11.1%, men in
their 30s at 10.7%, men in their 40s at 15.3%, and
men in their 50s at 11.7%. Women in their 20s and
30s also represented, around the same proportion,
at 11.4% and 8.5% respectively. Compared to current digital-music-content users, there is a more
even distribution of age and gender groups. As

for nonusers of digital music contents, women in

the middle to high age groups made up over 40%,
including women in their 30s at 14.7%, women
in their 40s at 12.3%, and women in their 50s at
16.3%. The data analysis incorporating generation factors highlighted distinctive characteristics
among the three user categories.

Comparison of the Number of CDs

Owned
When asked how many CDs they own, 18.7%
of current digital-music-content users said 50 to
99, followed by 23.8% owning 100-199 CDs and
11.9% owning over 200 CDs. These three groups
represent over 50% (54.4%). Among digital-music-content potential users, 18.2% own 0-9 CDs,
whereas those owning 10-19 CDs, 20-29 CDs,
and 30-49 CDs accounted for 14.0%, 13.7%, and
16.3% respectively. Combined, over 60% (62.2%)
owned less than 50 CDs. Almost 70% (69.8%) of
nonusers of digital music contents also own less
than 50 CDs, broken down into those with 0-9
CDs, 10-19 CDs, 20-29 CDs, and 30-49 CDs at
respectively 31.7%, 14.7%, 12.2%, and 11.6%.
As the figures show, current users have a large
proportion of people with a substantial CD collection, where as nonusers have a large proportion
of people with limited CD ownership.

Comparison of Monthly Music

Spending
Similarly to the former, the ratio of those spending over 3,000 yen (equivalent to the cost of one
CD album) per month was 61.8% among current
digital-music-content users but less than 40%
(39.1%) among digital-music-content potential
users. Over 70% (75.9%) of potential users spent
at least 1,000 yen (equivalent to the cost of one
CD single) per month. Nonusers of digital music
contents demonstrated a similar tendency to
potential users, with 28.8% spending over 3,000
yen, and 66.1% spending over 1,000 yen. As the

1227

Foreseeing the Future Lifestyle with Digital Music

figures indicate, current users have a large proportion of people who spend more on CDs, whereas
nonusers have a large proportion of people who
spend less on them.
Summarizing the results thus far, current
digital-music-content users are mainly young
men and women in their 20s, with substantial CD
ownership and high music-related spending per
month. They can be described as music fans with
substantial music-related consumption. Potential
users of digital music content, who are expected
to enter this market, are distributed across both
genders and broad generations, from youth to
those in middle age. They are characterized as
middle-level users in music consumption. Nonusers of digital music content are mainly women in
higher age groups, with relative inactiveness in
terms of music consumption. The results illustrate
clear differences in demographic characteristics
and music consumption behavior. There are major
differences between consumers who have bolstered the computer-based, digital-music-content
market until now, and those who will support the
market from now on. These facts alone point to
the possibility that the current market is set to
undergo substantial changes in its nature. In order
to examine details of anticipated changes, we have
compared the three groups in their attitude and
mentality in listening to music.

Formats of Music Ownership

Of current digital-music-content users 61.9%
acknowledge the desire to store all of their CD
collection on the computer, a significantly higher
ratio than digital-music-content potential users
at 26.7% and nonusers of digital music contents
at 17.2%. Current users appear to have a strong
desire to manage their music by computer and
use computers as the main device for handling
music content. In comparison, such desire is not
as strong among the other two groups.

1228

Intention Regarding Songs Available

for Downloading
Next, in order to examine the number of songs
that are available for downloading, we looked
at whether people want a greater selection from
download services via computer or those via
mobile phone. When asked whether the number
of songs downloadable via computer on the
Internet should be increased, 45.2% of current
digital-music-content users said yes, much
greater than 30.0% among digital-music-content
potential users and 15.0% among nonusers of
digital music content. As for whether they want
to see the number of songs available via mobile
phone increased, just 7.1% of current users agreed,
whereas the ratio was more than double at 15.0%
among potential users, and 9.7% among nonusers.
Although with not as stark a difference as the last
paragraph, these groups clearly demonstrated
different preferences in catalog enhancement
between downloading services via computers or
those via mobile phones. In short, current users
want to see enhancement of the computer-downloadable catalogs, while potential users want
enhancement of mobile-phone-based catalogs
just as much as of computer-based catalogs. The
results, once again, indicate a strong preference
among current users on computer-based services.
In comparison, potential users are requesting
catalog enhancement to both computer-based
and mobile-phone-based services. In other words,
potential users and nonusers wish to use both
computers and mobile phones to listen to music
rather than mere computers.

Style of Using Songs

We also asked several questions on how people
wish to use songs they own. Of current digitalmusic-content users 35.7% said they want to store
all CDs they own on a computer and edit them, for

Foreseeing the Future Lifestyle with Digital Music

example, compiling a collection of favorite songs.

The ratio was 22.5% among digital-music-content
potential users and 11.0% among nonusers of digital music contents. These figures again confirmed
the computer-oriented style of current users and
highlighted another of their characteristics, that
is, actively using downloaded songs for their
personal enjoyment. This characteristic became
even more evident in the next question.
People were asked whether they like to compile
a collection of songs from CDs they own according to specific themes and use the original CD as
a gift for friends on a suitable occasion of some
sort. Of current users 11.9% said yes, whereas
the ratio was 15.3% among potential users and
6.0% among nonusers. A greater proportion of
potential users expressed preference to this style
than current users.
The third question was on whether they
wanted to burn their favorite songs on CD-R
or DVD more casually to give away to friends
and acquaintances. The results showed a similar
tendency to the results for the second question.
Of current users 7.1% agreed, while the ratio
among potential users was greater at 12.7%.
Even nonusers had a greater proportion at 9.1%.
Looking at the results to these three questions,
current users have a self-contained approach in
enjoying music with a preference to downloading
via computers, whereas potential users are more
inclined towards exchanging and distributing
music with others.
Finally, we asked whether they wanted to give
away or exchange songs, downloaded via mobile
phone, to friends over the mobile phone. Again,
only 7.1% of current users, who have the preference to computer-based song purchase, agreed to
the concept, whereas the ratio was greater among
potential users (9.1%), with even nonusers reaching the same level as potential users (8.8%). All
the numbers point to the computer-oriented and
self-contained nature of current users, and the
potential users tendency of combined computer
and mobile phone use and a strong inclination
towards distributing and exchanging songs.

When these analysis results are combined

with the demographic characteristics and basic
indicators in music consumption, current digitalmusic-content users can be defined as those with a
strong computer preference, wishing to use digital
music content for their personal enjoyment in a
self-contained approach. In contrast, digital music content, potential users who are entering the
market from now, are combining computers and
mobile phones for this purpose and are inclined
towards enjoying music with others in addition
to appreciating it by themselves. This group has
a particular tendency of using music as one of
the tools for communicating with other people
around them.
Let us take a closer look as the results to enable
us to explore the direction of how Japans digital
music market may change, while reflecting upon
the trends of both hardware and peoples music
listening styles.
The digital-music-content market in Japan
originally evolved from the distribution of mobile
phone ring tones. Then, music content was merely
one of the functions or menus available in using
mobile phones. They did not go beyond the ring
tone boundary. Amidst this situation, the market
embraced the emergence of a new type of music
device that contains a hard disk or USB flash
memory, designed to be used with a computer.
Contemporary music fans were among the first
to adhere to such devices, consisting of men and
women in their 20s that are most active consumers of music. They stored and managed all music
content they already had in a computer, thereby
converting them into digital content, and began
carrying songs in portable music players and in
so doing they were enjoying music for themselves
in a rather self-contained fashion.
Today, digital music content that takes the form
of mobile phone ring tones exists alongside digital
music content that can be carried on hard-disk
or USB flash-memory music players. We have
investigated the future course of the market in
view of the profile and music mentality of potential

1229

Foreseeing the Future Lifestyle with Digital Music

users of digital music content, who are making

a full-scale entry into this market in the future.
Future users will be combining computers and
mobile phones, and, unlike contemporary users,
enjoying music as both a communication tool with
others and for personal entertainment purposes.
Digitizing music contents gives music a new
function as a communication promotion factor,
in addition to the current functions as ring tones
and personal enjoyment.
In order to further clarify this new style of enjoying digital music content, we conducted an oral
qualitative survey on two groups, that is, current
iPod users who represent those enjoying digital
music content via computers, and Chaku-uta and
Chaku-uta Full users who represent those enjoying digital music content via mobile phones. The
survey clarified their styles in listening to digital
music content, so as to obtain an insight into the
future direction of music-listening styles.

Future of the Digital Music

Content Market in Japan:
Changes in Music Content
Consumption
The Survey on the usage of iPod and Chaku-uta
was conducted in the period between December
2004 and February 2005. In-depth interviews
were held with three male and female iPod users
in their 20s and 30s, and with three male and
female Chaku-uta Mobile users in their 10s
and 20s.2

Comments From iPod Users

What follows are typical comments made by iPod
users on their style of enjoying music:
I find the Random Play function to be very refreshing. I can rediscover songs on CDs that
I did not pay much attention previously. [] I
can store a lot of songs without having to worry

1230

about how much space is left. Now, I am storing

whatever songs I have, even ones that I would
skip if I am playing it on CD. [A 37-year-old man
who has used iPod for 1.5 years, SOHO, 6,000
songs are held]
I now realize how much I have restricted myself
with frameworks of genres and artists when listening to music. The Shuffle function highlights the
raw power of individual songs. [] Occasions
like that have broadened the range of music genres
I listen to, making me feel like trying out CDs I
would never have dreamed of listening to before.
[26-year-old woman who has used iPod for 1 year,
office worker, 1,500 songs are held]
I never used to carry music around, but now I
do not go anywhere without my iPod. This has
widened the variety of occasions I listen to music, for example, while on a train or on the way
home after a drink. [] There are times when a
song I frequently listened to on CD sounds very
different on a portable player, because of various
situations you are in at the time. That gives me
fun. [39-year-old man who has used iPod for 6
months, office worker, 700 songs are held]
As suggested in the results of the quantitative
survey, they typicallyto some extenthave a
self-contained approach in music entertainment.
Their remarks illustrate new ways of enjoying
music (consumption styles) they have attained
through hard-disk music players like iPod. For
them, a hard-disk music player is a device that
allows them to randomly enjoy music out of a
greater selection of songs than previously possible in conventional devices (cassette player,
MD player), loaded from their CD collection. A
hard-disk music player is a true portable music
player strictly for personal use. In order for the
device to be self-contained, it must be able to
carry a massive number of songs, which in turn,
facilitates random playing. This random playing
then releases listeners from the boundaries of

Foreseeing the Future Lifestyle with Digital Music

existing music context (by genre, by artist, etc.)

and invites the creation of new contexts, thereby
enhancing the self-contained nature even further.
They are enjoying music in this cycle.
For them, consumption of music content is not
about listening to each individual song, but about
enjoying a stream of music. It must always be a
fresh string or stream of music different from
what they have already experienced previously.
Their consumption of digital music content is
characterized as self-contained context consumption. This is an emergence of a new style of music
consumption, only possible for hard-disk music
players like iPod. The style is facilitated with the
concept of play list in iPod and other devices.
The ability to compile a new string or stream of
music, has diluted the concept of album, presented
conventionally from package providers and artists
as producers, and encouraged individual users to
compile their own music streams. Consequently,
music is increasingly evaluated on the merit of
each song. One example is the way iTunes Music
Store presented its proliferation scale in the unit
of individual songs downloaded, rather than albums. This kind of presentation appears to depict
a transition of mentality, with listeners focusing
more on individual songs, rather than embracing
the supplier-defined unit of album.
The following comments are derived from
Chaku-uta Mobile users on their style of enjoying music:
During a break time at work, I just leave my mobile
phone to play songs to provide some background
music. [] They are all songs that everyone knows,
and will not trigger any music talk. However, it is
better than having no music, and stimulates conversation. [] I dont mind if each song may be
just 30 seconds long. It is actually better to have
short tunes to enjoy them with my colleagues and
build up a lively atmosphere. [25-year-old man
who has used Chaku-uta Mobile for 6 months,
office worker, 10 songs are held]

I like upbeat Techno music. I use these types of

songs as Chaku-uta, and play them during break
time when my friends are around, so that I can
show them how to dance to the tunes. [] The other
day, I had my mobile hanging around my neck, and
playing Chaku-uta, as I danced across a Shinjuku
intersection with my friend. [21-year-old woman
who has used Chaku-uta Mobile for 6 months,
university student, three songs are held]
I might listen to and check out music with my
friends, but I am more likely to use it as the Sleep
Timer when I go to bed. [] I can operate it by
hand, and put it on charge at the same time. It is
very convenient. [] I dont care about (each song
being the ring tone length of 30 seconds and) not
having the complete song. I fall asleep as I listen
to the same songs repeatedly. [19-year-old woman
who has used Chaku-uta Mobile for 1 year, vocational school student, five songs are held]
The analysis of the quantitative survey results
also indicated that persons entering the digitalmusic-content market from now use both computer-based music players and mobile phones, and
use music to enhance their relationship or communication with their friends and acquaintances
instead of merely enjoying music by themselves.
Comments from Chaku-uta Mobile users substantiate the tendency, that is, using music as a
tool for sharing various occasions with friends.
As part of the quantitative survey described
earlier, people were asked how they use ring tones
and Chaku-uta on mobile phones. The top three
answers were as ring tones (87.3%), alarm clock
(60.4%), and other alarm sounds (46.3%), The
fourth highest ranked answer, however, was to
enjoy them alone as music (44.0%), and 41.7%
said they enjoy them together with friends or use
them to entertain othersas such indicating that
people are beginning to enjoy mobile-downloaded
tunes as stand alone songs with friends.
What can be noted in these comments is that
songs are enjoyed in the ring tone length of 30

1231

Foreseeing the Future Lifestyle with Digital Music

seconds, rather than in their entirety, which is

quite different to that of hard-disk music player
users, who consume a massive amount of randomly replayed music in various contexts. Their
consumption style is summarized as sharing
the occasion of playing popular songs, rather
than personal favorites with others to magnify
enjoyment. What counts is how good each NETA
(=song as conversation topic) is, rather than how
many songs you have in store. Their consumption
of digital music content is characterized as NETA
consumption to be shared among friends. For
them, Chaku-uta Mobile is perceived as a music
player that plays 30-seconds of everyones favorite
songs for a shared experience. The emergence
of a new service called Chaku-uta has brought
about this new style in music consumption, while
now, the style seems to transform formats of
music content.

Conclusion
As we have examined, Japans digital-musiccontent marketwhich started off with the
distribution of ring tones as one mobile phone
servicehas embraced the arrival of fully fledged
digital music players and online stores, both
designed to be used via computers, such as iPod
and iTunes Music Store. From the viewpoint of
hardware competition, the market has now entered
a stage of combined development of mobile-phonebased devices and computer-based devices. It has
brought about two contrasting consumption styles
with distinctive characteristics (computer-based
and mobile-phone-based consumption of digital
music content), and diversified peoples styles in
enjoying music at the same time.
People who use a computer-based means to
enjoy digital music content, have a self-contained
style of consuming music in a specific context,
loading a hard-disk music player with a greater
amount of music from their personal CD collection
than previously possible and enjoying songs in

1232

random order. In contrast, people who use mobilephone-based devices, employ a mobile phone as
a communal music player for playing 30-second
tunes of high popularity and consume music as
topics (information) for sharing various occasions
with friends or enhancing the atmosphere.
At present, these styles are separate tendencies
and can be observed among users of hard-disk
music players and users of mobile phones as music players as two extremes. However, a steady
proliferation of hard-disk or USB flash-memory
music players may cause these styles to merge
on the side of individual users. Competition
between two types of devices has created two
distinctive styles of listening to music. Now, each
user may start using both of these devices at the
same time, hence adopting both styles alongside
each other. Such a user may eventually begin to
seek both of the styles in one of the two types of
devices, which may amount to hardware integration, brought about by the symbiosis of the two
different music-listening styles. Closely paying
attention to consumer behavior and practices in
the future will then give way to rich empirical data
to be used to develop and elaborate the stream of
thought outlined in this study further.

Further Reading
Institute for Information and Communications
Policy. (Eds.). (2005). Henbou suru contents
business. (Contents business.) Tokyo: Toyo keizai
shinpo sha.
Masuda, S. (2005). Sono ongaku no sakusha toha
dare ka. Tokyo: Misuzu shobo.
Masuda, S., & Taniguchi, F. (2005). Ongaku mirai
kei. Tokyo: Yosen sha.
Ministry of Internal Affairs and Communications.
(Eds.). (2005). Information and communications
in Japan 2005. Tokyo: Gyousei.

Foreseeing the Future Lifestyle with Digital Music

Tsuda, D. (2004). Dare ga ongaku wo korosu no

ka. Tokyo: Shoei sha.
Yoshimi, S. (2004). Media bunka ron. (Invitation
to media cultural studies.) Tokyo: Yuhikaku.

Endnotes

The survey was conducted by Macromill

Inc. in Japan and Zoomerang, Inc. in the
U.S. And, it was organized by Hakuhodo
Institute of Life and Living and Hakuhodo
DY Media Partners Media Environment
Laboratory.
The survey was conducted by Oval Communication and was organized by Hakuhodo
Institute of Life and Living and Hakuhodo
DY Media partners Media Environment
Laboratory.

This work was previously published in Information Communication Technologies and Emerging Business Strategies, edited
by S. van der Graaf, pp. 59-75, copyright 2007 by IGI Publishing, formerly known as Idea Group Publishing (an imprint of
IGI Global)

1233

1234

Chapter 3.15

Widely Usable User Interfaces

on Mobile Devices with RFID
Francesco Bellotti
University of Genoa, Italy
Riccardo Berta
University of Genoa, Italy
Alessandro De Gloria
University of Genoa, Italy
Massimiliano Margarone
University of Genoa, Italy

Abstract
Diffusion of radio frequency identification (RFID)
promises to boost the added value of assistive
technologies for mobile users. Visually impaired
people may benefit from RFID-based applications that support users in maintaining spatial
orientation (Mann, 2004) through provision of
information on where they are, and a description
of what lies in their surroundings. To investigate
this issue, we have integrated our development
tool for mobile device, (namely: MADE, Bellotti,
Berta, De Gloria, & Margarone, 2003),
with a
complete support for RFID tag detection, and
implemented an RFID-enabled
location-aware
tour-guide
.
We have evaluated the guide in an

ecological context (fully operational application,

real users, real context of use (Abowd & Mynatt,
2000)) during the EuroFlora 2006 international
exhibition (EuroFlora). In this chapter, we describe
the MADE enhancement to support RFID-based
applications, present the main concepts of the
interaction modalities we have designed in order
to support visually impaired users, and discuss
results from our field experience.

Introduction
Starting from the European Union cofounded
E-Tour project, we designed the tourist digital assistant (TDA) concept and developed multimedia

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Widely Usable User Interfaces on Mobile Devices with RFID

tour guides on mobile devices (PocketPC and

Smartphone devices) for a number of European
tourist sites, such as the Costa Aquarium of Genoa,
Strada Nuova architectonical area and the city
of Genoa, the Castellon region in Spain, and the
city of Uddevalla in Sweden (Bellotti, Berta, De
Gloria, &

Margarone,
, 2002).
The tour guide provides multimedia contents,
added-value information, and location-based
services to the tourists. Added-value services are
implemented by integrating the mobile devices
with additional hardware and software tools such
as GPS, electronic compasses, wireless connectivity, digital cameras, written text input, databases,
and so forth.
See Figure 1 for snapshots of tourist guide
applications.
Relying on the argument that play is a powerful mediator for learning throughout a persons
life, we developed the educational territorialgaming concept in VeGame (Bellotti, Berta, De
Gloria, Ferretti,

& Margarone, 2003),

a computersupported educational wireless team-game played

along Venices narrow streets to discover the art
and the history of the city (see Figure 2), and in
ScienceGame (Bellotti, Berta, De Gloria, Ferretti,

& Margarone, 2004),

a sort of treasure-hunt game

inviting players to discover the mysteries and the
marvels of the science (see Figure 3) during the

Festival della Scienza exhibition held in Genoa

every year.
These applications were developed from
scratch. From these first experiences, we identified common needs and came up with a system
to support design of multimedia applications
for mobile devices, called Mobile Applications
Development Environment (MADE) (Bellotti
et al., 2002).
MADE includes M3P (MicroMultiMedia
Player), a network-enabled multimedia player
easily programmable through the micromultimedia services language (MSL). MSL provides
high-level components encapsulating advanced
services (e.g., positioning, database query, path
search, etc.) that can be easily integrated in multimedia applications. This allows building modular
software programs that provide information-rich
services to the general public through a coherent
and homogeneous HCI that can be learned with low
mental workload. On the other hand, MADE hides
the low-level aspects of multimedia and service
management, allowing designers to focus on the
modalities of presentation of information and on
user interaction, reducing learning, development,
and code maintenance time.
In this chapter, we describe the latest MADE
enhancement: we have integrated it with a complete support for RFID detection to allow develop-

Figure 1. Snapshots from the Aquarium and Strada Nuova tour guides on PocketPC device

1235

Widely Usable User Interfaces on Mobile Devices with RFID

Figure 2. Snapshots from VeGame

Figure 3. Snapshots from ScienceGame

ment of multimedia mobile applications directly

connected with the physical world (Want, Fishkin,
Gujar, &

Harrison, 1999).

All low-level aspects of

the hardware tag-detection system that are necessary to identify and locate physical objects with
attached small RF tags (Want, 2004) are hidden
to MSL programmer by the MADE system.
This chapter will also show the use of MADE
with the RFID support in a real context such as

1236

EuroFlora 2006 international exhibition. This

guide differs from others because it has been
ad-hoc developed in order to meet strict usability
needs. In particular, the novel interface design
assists visually impaired people in maintaining
spatial orientation (Mann, 2004) through provision of information on where they are, hazards
that might be in the way, and a description of what
lies in their surroundings.

Widely Usable User Interfaces on Mobile Devices with RFID

MADE support of RFID

technology
Location-Aware Computing
Recent research has developed several systems,
to determinate physical location, that differ by
accuracy, cost, and coverage (Boriello, Chalmers,
La Marca, & Nixon, 2005). The global positioning
system (GPS), which uses signal from satellite to
estimate position (Djuknic & Richton, 2001), is
the most used system, but only for applications
in outdoor areas. In indoor areas and urban areas
with poor sky visibility, the system does not work
properly. Moreover, it has a long start-up time.
To overcome these limitations, the first indoor
positioning system was the active badge system
(Want, Hopper, Falco, & Gibbons, 1992), which is
based on sensors that receive infrared ID broadcast
from tags worn by people. This system gives a
poor (room-grained) localization precision. After
the active badge system, typical indoor location
systems are based on radio frequency and on the
estimation of position computed from the measured signal strength. Various technologies can be
used: Wi-Fi (Howard, Siddiqi, & Sukhatme, 2003),
Bluetooth (Bruno & Delmastro, 2003) and nowadays RFID (Liu, Corner, & Shenoy, 2006).
The first two solutions can give an accuracy
of around some meters, but require expensive
fixed base stations. RFID tags, instead, are very
inexpensive and have the same performance. The
literature reports also of many location estimation
algorithms based on cellular radio networks (Xu &
Jacobsen, 2005). However, there is not a generally
agreed solution today, and each algorithm has pros
and cons, depending on environmental issues.
Finally, some vision-based algorithms (Lpez de
Ipia, Mendona, & Hopper, 2002) are promising because they do not require infrastructure
(like tags, satellite, or base station). However,
it is difficult to set up a system to locate a user
with a 1-meter precision. In the selection of the
best methodology for our system, we have taken

into account three major issues: the possibility

to have a system for outdoor/indoor spaces (like
the EuroFlora 2006 exhibition area), a technology
with a low cost for the deployment of the infrastructure, and a likely pervasive availability of the
system in the near future. All these requirements
are satisfied by the RFID technology.

RFID Application Fields

Major RFID application domains include monitoring physical parameters, such as temperature or
acceleration, during fragile or sensitive products
delivery, monitoring product integrity from factory to retail locations (Siegemund & Floerkemeier,
2003), utilities for home and office automation
(Langheinrich, Mattern, Romer
& Vogt,
, 2000).
Nowadays we have passive or active inexpensive
RFID (approaching 35 cents today, with a goal
of 5 cents (Quaadgras, 2005)) that makes these
kinds of sensors practical for tourist applications.
For example, a museum exposition can place tags
attached to each point of interest so that tourists
can receive information about exposition in the
right moment at the right place; when near to the
object. The research community has actively
explored this possibility at the Exploratorium, the
interactive science museum in San Francisco. The
HP Laboratories researchers have implemented
a system that uses three types of identification
technology: infrared beacon, barcodes, and RFIDs
(Fleck, Frid, Kindberg, OBrian-Strain,

Rajani, &
Spasojevic,
2002). In Goker et al. (Goker, Watt,
Myrhaug Whitehead,

Yakici, Bierig, et al., 2004),

a special tag that can work with mobile devices

to provide ambient information to users on the
move is described. In the Cooltown project (Kindberg & Barton, 2001), RFIDs are used to attach
pointers from everyday objects to entities in the
computational world. A full exploitation of RFID
potentials requires study and implementation of
humancomputer interaction (HCI) modalities
able to support usability of the enhanced mobile
tool by the general public. This implies the neces-

1237

Widely Usable User Interfaces on Mobile Devices with RFID

sity to resort to programming methodologies and

tools specifically dedicated to support the RFID
technology. Thus, we have extended the MADE
toolkit to support a link between applications and
physical world through RFID sensors.

MADE Architecture
A typical MADE application consists of a set of
pages containing multimedia and service objects.
The micromultimedia services language (MSL)
script specifies pages layout and objects appearance, synchronization, and user-interaction
modalities. MSL scripts are interpreted at runtime
by the M3P player that manages presentation of
contents and user interaction according to the
instructions specified in the input MSL script.
M3P player relies on two-layer architecture
(see Figure 4) involving a high-level, platformindependent director and a low-level driver. The
director is responsible for creating, initializing,
and managing the objects that implement the
language functionalities. In order to support
incremental development of the player, M3P is
composed by a set of modules. In particular, the

director has been designed to be independent of

the components it manages. According to the object-oriented methodology, this has been achieved
by encapsulating the functions of the components
in the code of their class, and by structuring the
hierarchy so that the director can simply keep a
reference to the presentations pages and convey
them events.
According to the instructions specified by the
MSL developer in the script, events (either from
the system or from user interaction) are conveyed
to the director that simply redirects them to the
components of the page currently on show, which
is the higher-priority choice or, with lower priority,
to the other pages of the presentation.
Events are implemented as string messages
that are to be interpreted by the target objects.
This design choice allows the directors code
to be independent of the components and the
components to be independent of each other.
The basic assumption of this schema is that each
component exports a well-defined interface (i.e.,
a set of messages to which it is able to react) and
implements this interface (i.e., implements the
reaction to such messages).

Figure 4. MADE architecture

MSL
Application

M3P Core

Data Structure

Director

A C++ object
corresponds to each
MSL components
instance

Driver

Operating System
(WinCE)
User

1238

Widely Usable User Interfaces on Mobile Devices with RFID

Thus, components can be seamlessly added

and interchanged (in this last case, as long as they
comply with the same interface). Adding a new
component (i.e., a new functionality) does not
involve any change either in the directors code,
or in the other components code.
Such a design choice supports easy incremental
development, allowing seamless integration of services within a single application framework. This
implies that a homogeneous HCI can be applied to
an application that hosts several different services
that have been developed independently of each
other (e.g., intelligent tour planning, interactive
maps, positioning, and database access).
MSL relies on a component-based data structure. That is, an MSL file specifies creation of
components, attributes of components, and their
reaction as a consequence of user interaction.
Components are organized in three main libraries: multimedia (e.g., audio, image, video, button),
synchronization (utilities like timers that can be
used to implement synchronization and scheduling
of contents), and services (objects that encapsulate
services such as positioning, shortest path search,
tour planning, database query, etc).
Every different type of component has its own
kind of attributes (fields). The fields record data
for specifying the appearance (such as position)
and the behaviour (i.e., reactions to events) of the
component. In general, components are contained
in a special container component, called CARD,
that can be thought of as an empty page on which
the developer can add components.
The core of M3P involves a platform-independent director and a platform-dependent
driver. The director manages the multimedia
objects that implement the presentation. Objects
are compounded in hierarchical structures. For
instance, a CARD (i.e., a multimedia page) may
include several images, buttons, and mpeg players.
The driver implements the functions to access the
hardware, while the director deals with the logic
of the multimedia presentation.

Integration of RFID Subsystem

A major feature of MADE consists in the possibility of incrementally adding new hardware
and software modules, that are integrated into
the HCI framework with no need for modifying
the M3P core, since every components class is
responsible for interpreting its receivable messages, independent of the others. MADE can
integrate, into a common framework, various
hardware modules independently developed to
augment the mobile device potentiality. M3P
drivers classes, which have to be developed to
integrate every new hardware subsystem, manage
low-level aspects of the hardware modules, while
the MSL interface to the application developer
abstracts the services at high level. This implies
that a homogeneous HCI can be applied to an
application that hosts several different services
that have been developed independently of each
other (e.g., automatic positioning, intelligent tour
planning, and database access can be integrated
in an interactive map), and the MSL developer
can simply exploit the service modules focusing on the integration of the HCI. Examples of
hardware modules already integrated in MADE
are a positioning and orientation module that an
MSL developer can exploit to get geographical
position from a GPS receiver and direction from
a digital compass, and the remote communication
module able to exploit the hardware available for
connection with the external world (e.g., wired/
wireless LAN, Bluetooth, GSM/GPRS cellular
networks).
In order to enable applications to react to objects in the physical world, our new M3P module,
called RFID sensing module (RfidSM), detects
presence of RFID tags in the surrounding space
and notifies the M3P run-time objects with events
implemented as string messages (see Figure 5).
The script interface of the RfidSM is a new
MSL component, called RFID, that exposes the
fields shown in Table 1.

1239

Widely Usable User Interfaces on Mobile Devices with RFID

Figure 5. Integration (within MADE) of the RFID sensing module (RfidSM)

Table 1. RfidSM fields description

Component
Field
Target

Description

Period
Repetition
Id
Delay
dBm

A time period in milliseconds between two consecutive environmental scans to detect tags
A number of tag detection operations executed consecutively on each scanning action
A list of RFID tags that are of interest for the component
A list of time frames, one for each interesting tag, in which tags are not identified again
A list of signal strength values, one for each interesting tag, that specify thresholds for tag
identifications
A list of events, one for each interesting tag, that RFID component launch when a tag is
identified
If a component launch this event on a RFID component starts the scanning of tags
If a component launch this event on a RFID component stops the scanning of tags

onFound
Start
Stop

List of identifiers of the components to which information about identified tags are sent

When the RfidSM component is started, and

until it receives a stop event, it scans the surrounding environment to check the presence of tags

1240

every period of time. The list of detected tags

is then sent with the MADE message-exchange
modalities to the components specified in the

Widely Usable User Interfaces on Mobile Devices with RFID

target field. In addition, the component has an

id field to allow programmer expressing interest
in a set of tags, and defining (through the field
onFound) the corresponding events list that
should be executed. Each interesting tag is also
featured with a signal strength threshold (through
the field dBm) that specifies a limit under which
the tag is considered in range.
There is the problem of collisions, since the
scan results are typically imperfect due to not
all tags are detected in every scan. To solve this
problem, a tag typically awaits a random number
of time slots before it answers the RF pulse sent by
the reader. However, the problem still remains and
grows as the number of tags in the surrounding
environment grows. The MADE RFID sensing
module tackles this issue, allowing the programmer to specify, through the field repetition, a
number of times that the reader should repeat the
scanning before returning to the founded tags. The
list of founded tags is the collection of all tags
observed in each scan. A small value of repetition
results in a fast scan with high risk of collision,
whereas large repetition value results in a slow
scan with few collisions. This trade-off should be
resolved by the programmer basing his decision
on application constrains: long delays can result
in human-computer interaction problems if the application allows a user expectation for immediate
reaction to tags. This is the case of applications in
which user a voluntarily accosts the mobile device
to tagged objects to obtain information. Similar
problems arise if the application has a short time
frame to detect tags, for example, in applications
where the user moves at relatively high speed in
the environment, like in territorial games. Instead,
others type of applications can gain advantage
from precise but slow detections. It is the case
of a tourist mobile guide for a naturalistic park
in which the user moves along a path with some
points of interest largely spaced, like tree species
or rare flowers.
The other problem affecting the RFID technology is the tag detection flickering (Rmer,

Schoch, & Mattern, &

Dubendorfer, 2003):

due to
the collision problem, some tags can appear and
disappear from sequential scanning, generating a
fast list of tag identifications. The MADE RFID
sensing module allows the programmer to decide
how to convert scan results into applications events
handling this problem. Programmer can specify,
through the delay field, a time period (for each
interesting tag) starting from the detection. During this time, subsequent detection events of the
same tag are discarded; also, the exact definition
of this delay is application dependent. Applications with events that occur only one time, like
tourist guide for museums with linear path, can
have delay values set to infinite. Instead, in applications with events generated multiple times
closer each other, like territorial games, the delay
should be short or zero.
Currently, we have implemented the low-level
driver support for the iCARD Identec reader in
a PCMCIA card format (IDENTEC). This card
can be integrated in handheld, portable, or laptop computers to communicate with the iQ and
iD active RFID tags at a distance of up to 100
meters. The RF signal is in the UHF radio band
(915 MHz or 868 MHz), providing long-range
communication and high-speed transmission
rates for reliable data exchange.

The EuroFlora guide

In order to assess the possibility of using RFID
technology to develop widely usable interfaces,
we present a real-world application (see Figure
6) developed through the MADE toolkit and
deployed in an ecological environment (fully operational, reliable, and robust application, used by
real users and in a real context of use) at EuroFlora
2006 (the international flower exhibition that is
held in Genoa every 5 years). With over 500,000
visitors in 10 days, EuroFlora is one of the most
important exhibitions of Europe.

1241

Widely Usable User Interfaces on Mobile Devices with RFID

Figure 6. Snapshot of the cover page of EuroFlora

Guide application

The developed application concerns the research area of assistive technologies for visually
impaired people. Such assistive applications have
the potential to improve the quality of life of a
large portion of population (by 2020, there will
be approximately 54 million of blind persons over
age 60 worldwide (WHO, 1997)).

In this field, maintaining spatial orientation is

a major challenge for people with visual impairment. There is the need of systems in providing
blind people with information on where they are,
hazards that might be in the way, and a description
of what lies in their surroundings (Mann, 2004).
The notion of spatial orientation refers to the
ability to establish awareness of space position
relative to landmarks in the surrounding environment (Guth & Rieser, 1997). The goal of our
application is to support functional independence
to visually impaired people, providing support to
indoor awareness of elements in the surroundings
(Ross, 2004).
The EuroFlora guide is organized in two parts.
One part provides general information about the
exhibition, the guide, and their services. The
other part provides the description of the selected
interest points. While first part is directly accessible by the user at any moment, the second one
is event driven. More precisely, every interest
point description is associated to an RFID tag,
and when a user enters that area (i.e., her/his
handheld device recognizes the RFID tag), the
software asks the user whether to launch the
corresponding description.
We placed 99 RFID sensors on an area of
30,000 mq of exhibition, covering 99 points of

Figure 7. a) The packaging of the multimedia guide in a leather case; b) Snapshots from the tests: users
visit EuroFlora 2006 supported by the guide and touch some dedicated plants

1242

Widely Usable User Interfaces on Mobile Devices with RFID

interest, services, and major areas (see Figure

7). RFID sensors were IP65 compliant in order
to resist to water and dust, and self-powered.
Power level of sensors could be set in two levels,
low and high.

enhance the direct perception of the visitor,

leading to a better and more pleasant comprehension of her/his surrounding environment.
For example, the guide should make use of
environmental sound (e.g., waterfall) and
scent (e.g., flower smell) to connect content
information and the objects in the space.

Design Methodology
The necessity for combining the flexibility and
multimedia potential of a mobile device with the
extreme simplicity of interaction, required for use
by a wide audience (also visually impaired people),
involves facing three main HCI issues:

Usability by general users: The tourist has

little time and willingness to learn how to
use the new technological tool, since she or
he is there to visit the exhibition and not to
learn a tool. Most of the tourists use such
a tool for the first time and just for a short
time (typically, from 30 to 90 minutes).
Thus, the complexity of the platform should
be hidden from visitors, making the guide
immediately usable, with no effort by users.
This implies that the interface is to be as
simple and intuitive as possible.
Usability by visually impaired people:
Visiting an exhibition is a critical task for
the blind, mainly for the combination of
several reasons: the site is often crowded and
unfamiliar to the visitor, it may be noisy, it
is difficult to orientate in a highly dynamic
place. In this context, the guide should be
not intrusive, with few and very recognizable
input interface elements (also with tactile
feedback), and should give information in
a proactive modality when needed by the
user.
Presentation of information: Added-value
information (e.g., how the various specimens
live in their natural environment) should
be synergistic with the direct experience
of the visitor at the exhibition. Provision of
information has to be structured in order to

We have tackled such issues resorting to the

methodologies of the user-centric design (Carroll,
1997), in an iterative development of the guide
involving participatory design (Beck, 1993), definition of usability specifications, and contextual
design, as shown in the following:

Participatory design consisted of the participation of botanists at the design decisions,

authors skilled in writing for blind people and
visually impaired end-users, together with
technical developers. The most significant
contribution of the first three categories
consisted in the definition of the targets and
in the concrete perspective they brought into
the project.
Usability specifications provide explicit and
measurable targets to verify the suitability of
the work done. Examples of such goals are
90% of the users should be able to operate
the guide without asking questions to the
personnel, 90% of the users should be able
to use the interface with no errors, 90%
of the users should be able to understand
the meaning of all the touchable controls
within 120 seconds. All these objectives
were verified in early lab and field tests in
order to take the appropriate corrective actions.
Contextual design involved early field tests
with experts and users at the exhibition in
the preopening days when the infrastructure
of EuroFlora was being built. Field tests
have been helpful to highlight problems and
shortcomings that had been overlooked or
ignored in lab.

1243

Widely Usable User Interfaces on Mobile Devices with RFID

Structure of the Interface

The interface of the EuroFlora guide has been
designed to support immediate use by the general public, also by visually impaired people.
To this end, we used general design principles
(we already described them in the introduction)
such as overall simplicity, low intrusiveness, and
support for natural interaction and knowledge
acquisition. Moreover, we added further features
in order to meet the specific needs of visually
impaired people:

Tactile feedback in the control interface

Tutorial stage
Event-driven interface
Facilities to support orientation

The basic element of the interface is the multimedia card. A multimedia card corresponds
to each subject of a presentation (e.g., a flower
species). Each multimedia card provides, in an
audio format, texts specifically written for visually
impaired people (i.e., highlighting olfactive and

Figure 8. Snapshot of the guide instruction

1244

tactile sensorial information, providing detailed

ambient descriptions).
The tactile feedback is necessary to allow
impaired people to easily understand the position
of the controls and give her/him feedback. Our
previous multimedia guides had the interface
embedded in the graphic contents, exploiting the
touch screen of a pocket-pc device. During the
early field tests, visually impaired people pointed
out some important shortcomings in these solutions. They felt that the screen was too large and
their fingers were lost in a space without roughness. Since most of such users are well acquainted
with the common cell phones relief keyboard, we
studied a new solution exclusively based on the
hardware buttons (Figure 8).
The hardware buttons of the interface are
highlighted. The choice of this area as navigation
control allows visually impaired people to have
a tactile feedback. The meaning of the button is
up to accept a content description (which is
automatically triggered when the user enters a
new cell), down to reject it, right to exit from
a section of the guide, and back to the main menu,
left to have contextual information about users
current position.
The tutorial stage is an initial guide section
in which users could freely experiment withthe
interface of the tool in order to allow people to
use the guide in an independent way. In this stage,
users are invited to freely press buttons. A speech
description briefly explains the meaning of each
pressed button. This tutorial stage prevents the
necessity for providing papers or long explanations when users rent the guide.
The event-driven interface allows a user to
get information about points of interest (POIs)
and orientation when they are in the proximity
of a POI. For example, in Figure 9, a user near
the azalea stand is told about the presence of this
flower by a pop-up window (the guide is to be
usable by everybody) and a corresponding jingle
sound. If she/he wants to listen to this content,
she/he can press the up hardware button. By

Widely Usable User Interfaces on Mobile Devices with RFID

default, the system skips the presentation. This

operational mode has low intrusiveness (users are
asked whether to listen to a content), but it also
provides a certain degree of proactivity. Infor-

Figure 9. Snapshot of the event-driven interface

Figure 10. Snapshot of the map section of EuroFlora Guide

mation is not only botanical, as in the example

of azalea, but also concerns the positioning of
the user. Many tags are placed in the proximity
of facilities, such as lavatories, cafs, exits, and
intersections. This localization system lowers
the mental workload necessary for the tourist to
synchronize the physical space of the exhibition
with the virtual space of the program.
This message (accompanied by a jingle) is
shown to the user when she/he has just entered a
POI area. The combination of audio and graphics
is due to the fact that the guide may be used also
by not visually impaired people. In the example
in this figure, the user is near to the azalea flower,
and if she/he is interested in the description she/he
can press the up hardware button to access the
related content.
One of the main tasks of the guide is to assist
the visitor in her/his exploration of the exhibition space. A facility to support orientation (not
useful for visually impaired people) is a section
with the map that helps the tourist to orient
herself /himself in the exhibition. The map (see
Figure 10) shows the structure of the EuroFlora,
including lavatories, caf, exits, and so forth, and
the location of the points of interests. In order to
enhance users orientation, the map is centered
on the position of the user, as determined by the
currently perceived RFID tags.

Field evaluation
Experimental Framework
Real evaluation of advanced mobile device applications and of the impact on their intended
population is difficult and costly. Evaluation
requires analysis of real users, in a real context
of use. In order to adequately evaluate interaction
with computation resources, test-people should
use a fully operational, reliable, and robust tool,
not just a demonstration prototype (Abowd &
Mynatt, 2000). Hence, it is important to perform

1245

Widely Usable User Interfaces on Mobile Devices with RFID

early tests in the authentic context of use in order

to verify end-user acceptance and overall usefulness of the system, and to receive feedback to
inform future design.
In this chapter, we describe the early stage
analysis of acceptance and usefulness of the
developed multimedia guide. The tests were
performed at EuroFlora 2006, the international
flower exhibition that is held in Genoa every 5
years. With over 500,000 visitors in 10 days,
EuroFlora is one of the most important exhibitions of Europe.
The exhibition area (around 90,000 squared
metres) was equipped with an infrastructure of
99 RFID tags. The experimentation involved 120
visually impaired tourists who used the tour guide,
and were observed and interviewed by expert of
disabilities and HCI designers. Subjects were
visually impaired (25%) or blind persons (75%)
representing a wide range of age (28% age <30;
32% age between 30 and 50; and age >50 40%).
Moreover, the tests involve 64 females and 56
males.
The tour guide consisted of a PocketPC
equipped with a special leather package with a
lace dropping from the neck for a more comfortable use. Headphones were also used in order to
isolate the user from the highly noisy surrounding
environment (see Figure 1).

Preexhibition Tests
In an early test session-performed 2 days before
the official opening of the exhibition, when some
stands were already readyenabling a realistic
test-we prepared a prototype software version
that was used by five selected visually impaired
users visiting 30% of the total exhibition area. We
followed and interviewed the users in this phase,
in order to understand shortcomings, defects and
weaknesses, and strong points of the product.
In this phase, we understood and solved some
problems on user interface and contents, such as
the most suited assignment of buttons to presentation control functionalities and the length of
the descriptions. Some test-users found the long
silence time between a presentation activation
and the next one (i.e., the period of time in which
the user is walking through areas not covered
by RFID tags) frustrating. We partially tackled
this issue by periodically providing a message
saying that the user is currently in an area not
close to a POI.

Ecological Tests
One hundred and twenty blind people used
the guide during the exhibition. Sixty of them
(aged from 12 to 78 years old) participated in

Table 2. Overall survey results

Issue
Usability
Usefulness
Support for spatial orientation
Session length time

1246

Average
4.00
4.25
4.20

Standard Deviation
0.64
0.75
0.66

201 minutes

30 minutes

Widely Usable User Interfaces on Mobile Devices with RFID

an ecological test conducted by the authors. We

interviewed the users at the return of the guide.
We evaluated three main performance factors:
usability (including effectiveness, efficiency and
pleasantness of use), usefulness, and capability
to support spatial orientation (in particular the
approach to the POIs). We asked users to give a
general comment on the guide and a 1-5 grade for
each factor (which was carefully explained and
defined by the interviewers). An overall survey
of results is reported in Table 2; it clearly shows
the high acceptance by the users.
Analyzing the variables correlations based on
the chi-square test, we observed that usability is
correlated with the perceived support for spatial
orientation (=25.3, df (degree of freedom) =
16, 90% confidence), and that perceived utility
of the tools is strictly correlated with perceived
support for spatial orientation (=30.2, df=16,
99.9% confidence). This suggests the importance
of our design choice to use mobile technology to
support orientation of visually impaired people.
Moreover, test results also show that the tool is
perceived as useful and usable.
Considering the free comments, the guide was
judged as an excellent tool for users to orientate
themselves inside the exhibition. Several people
expressed a similar concept, which we can synthesize with the words of one visitor: after always
having been guided, for the first time I myself
have been able to guide my wife and to explain
the exhibition! Such positive comments were also
confirmed by the blind assistance experts, who
highlighted the significant degree of independence
the blind could reach through the guide.
Shortcomings in the interface were reported
by some elderly users, while some people asked
for more extended descriptions, though each
point of interest included at least one. The high
performance and reliability of hardware, software,
and batteries assured long sessions of use with
no troubles for the user.

Future trends and vision

The research community is envisaging a new
model of a tagged world as an intelligent environment that allows providing visually impaired
people with information about architectonical
barriers, safe paths, points of interest, potential
danger areas, and other useful information. A
sample scenario description may give an idea of
this likely future.
Maria is visually impaired. She is in a foreign
city on a business trip. Maria owns a mobile
device with a mobility-assistance system (MAS:
it is similar to the EuroFlora Guide, but with a
much larger action range). The MAS accompanies her in her path to her destination office, and
signals pedestrian crossings, traffic lights, safe
paths in work-in-progress areas, and so forth.
All objects in the world send their signals, but
Marias wearable device has an intelligent reasoning algorithm (based on user preferences and
interpretation of the users current activity) and
a suitable human-computer interaction (HCI) in
order to provide her only with the needed information. This information is extracted from a mass
of data that are continuously received from the
close-by RFID tags. Thus, the wearable device
notifies Maria about a pedestrian crossing only if
it knows that this is useful for her current activity
(i.e., going to office). Not useful information will
not be provided, in order not to distract Maria.
Along her path to her destination, Maria passes
by a newsagent. The World Guide scans all the
magazines and identifies todays issue of Marias
favourite magazine. It queries Marias database,
which replies that Maria has not purchased this
issue yet; so, it notifies her about the opportunity
to buy the magazine.

Conclusion
The ubiquitous presence of smart tags will offer,
in the near future, a critical mass of information,

1247

Widely Usable User Interfaces on Mobile Devices with RFID

embedded in the world, that will be exploitable to

rethink the relationships between people involved
in their daily-life activities and the surrounding
world.
With MADE we have designed a system that
continuously scans the tagged world, interprets
the large amount of information coming from the
surrounding objects, and provides it to the users
through multimedia human-computer interaction.
Moreover, the application in the future will filter
the raw data coming from the environment (with
artificial intelligence behaviour) taking into account the user needs, preferences, and profile.
The field test at EuroFlora 2006 has demonstrated the feasibility of our vision, by deploying
the system in a real-world setting (an exhibition
area with indoor and outdoor instrumented environments), and performing extensive field tests
with real users. In a longer-term view, with such
an application, we intend to investigate the future
scenarios that will be enabled by a massive presence of RFID tags in our environments. This early
prototyping has allowed us to understand, as early
as possible, costs, limits, strengths, and benefits
of the new technology. We have also obtained a
significant positive feedback on user acceptance.
Usability results show that the guide is perceived
as highly usable and useful, in particular because
of its ability to support spatial orientation.
The next step towards a tagged world will
require integration of data and services, and
capability of interpreting a variety of sources according to the specific and dynamic user needs.
Achieving these goals will involve a huge research
effort that will be successful only if it will lead
to the deployment of compelling applications
that will be perceived as useful by the users. In
a user-centered design view, this implies a rapid
prototyping of applications and extensive user
testing in the real context of use, which was our
inspiring principle in the EuroFlora project.

1248

References
Abowd, G. D., & Mynatt, E. D. (2000). Charting
past, present, and future research in ubiquitous
computing. ACM Transaction in Computer-Human Interaction, 7(1), 29-58.
Beck, A. (1993). User participation in system
design: Results of a field study. In M. J. Smith, Human-computer interaction: Applications and case

Elsevier.
studies (pp. 534-539). Amsterdam:
Bellotti, F., Berta, R., De Gloria, A., Ferretti, E.,
& Margarone, M. (2003).
VeGame: Field exploration of art and history in Venice. IEEE Computer,
26(9), 48-55.
Bellotti, F., Berta, R., De Gloria, A., Ferretti, E.,
& Margarone, M. (2004 ). Science

game: Mobile
gaming in a scientific exhibition. eChallenges
e2004. Fourteenth International Conference on
eBusiness, eGovernment, eWork, eEurope 2005
Vienna.
and ICT.
Bellotti, F., Berta, R., De Gloria, A., & Margarone,
M. (2002). User

testing a hypermedia tour guide.

IEEE Pervasive Computing, 1(2), 33-41.
Bellotti, F., Berta, R., De Gloria, A., & Margarone,
M. (2003).
MADE: Developing edutainment applications on mobile computers. Computer and
Graphics, 27(4), 617-634.
Borriello, G., Chalmers, M., La Marca, A., &
Nixon, P. (2005). Delivering

real-world ubiquitous
location systems. Communications of the ACM,
Special Issue on The Disappearing Computer,
48(3) 3641.
Bruno, R., & Delmastro F., (2003). Design and
analysis of a bluetooth-based indoor localization
system. Personal Wireless Communications,
2775, 711-725.
Carroll, J. M. (1997). Human-computer interaction: Psychology as a science of design. Inter-

Widely Usable User Interfaces on Mobile Devices with RFID

national Journal of Human-Computer Studies,

46(4) 501-522.
Djuknic, G. M., & Richton, R.E. (2001). Geolocation and Assisted GPS. IEEE Computer, 34(2),
123-125.
EuroFlora. (s.d.). EuroFlora 2006 Home Page.
Retrieved from http://www.fiera.ge.it/euroflora2006/index_eng.asp
Fleck, M., Frid, M., Kindberg, T., OBrian-Strain,
E., Rajani, R., & Spasojevic, M. (2002). From
Informing to Remembering: Ubiquitous Systems
in Interactive Museums. IEEE Pervasive Computing, 1(2), 11-19.
Goker, A., Watt, S., Myrhaug, H. I., Whitehead,
N., Yakici, M., & Bierig, R. (2004). An ambient,
personalised, and context-sensitive information
system for mobile users. Second European Union
Symposium on Ambient Intelligence. Eindhoven,
The Netherlands.
Guth, D. A., & Rieser, J. J. (1997). Perception and
the control of locomotion by blind and visually impaired pedestrians. In Foundation of orientation
and mobility (2nd ed.) (pp. 9-38). AFB Press.
Howard, A., Siddiqi, S., & Sukhatme, G. (2003).
An experimental study of localization using wireless Ethernet. International Conference on Field
and Service Robotics
IDENTEC. IDENTEC Solution Ltd Home Page.
Retrieved from http://www.identec.com
Kindberg, T., & Barton, J. (2001). A Web-based
nomadic computing system. Computer Networks,
35(4), 443-456.
Langheinrich, M., Mattern, F., Romer, K., & Vogt,
H. (2000). First steps towards an event-based
infrastructure for smart things. PACT 2000.
Philadelphia.
Liu, X., Corner, M.D., & Shenoy, P. (2006, September). RFID Localization for pervasive mul-

timedia. In Procedings of the 8th International

Conference on Ubiquitous Computing (UbiComp
'06). California, USA.
Lpez de Ipia, D., Mendona, P., & Hopper A.
(2002). TRIP: A low-cost vision-based location
system for ubiquitous computing. Personal and
ubiquitous computing, 6(3), 206-219
Mann, W. C. (2004). The aging population and its
needs. IEEE Pervasive Computing, 3(2), 12-14.
Quaadgras, A. (2005). Who joins the platform?
The case of the RFID business ecosystem. 38th
Hawaii International Conference on Systems
Science (pp. 855-864). Gig Island (HI): IEEE
Computing Society Press.
Rmer, K., Schoch, T., Mattern, F., & Dubendorfer,
C. (2003).
T-smart identification frameworked for
ubiquitous computing applications. PerCom 2003.
Fort Worth: IEEE Press.
Ross, D. A. (2004). Cyber crumbs for successful
aging with vision loss. IEEE Pervasive Computing, 3(2), 30-35.
Siegemund, F., & Floerkemeier, C. (2003). Interaction in pervasive computing settings using
Bluetooth-enabled active tags and passive RFID
technology together with mobile phones. PerCom
2003. Fort Worth: IEEE Press
Want, R. (2004). Enabling ubiquitous sensing
with RFID. IEEE Computer, 84-86.
Want, R., Fishkin, K., Gujar, A., & Harrison,
B. (May 1999). Bridging physical and virtual
worlds with electronic tags. ACM Conference on
Human Factors in Computing Systems (CHI 99).
Pittsburgh, PA.
Want, R., Hopper, A., Falco, V. & Gibbons, J.
(1992). The active badge location system. ACM
Transactions on Information Systems, 10(1),
91-102.
WHO, W. H. (1997). Who sounds the alarm:
Visual disability to double by 2020. Retrieved

1249

Widely Usable User Interfaces on Mobile Devices with RFID

from http://www.who.int/archives/inf-pr-1997/
en/pr97-15.html
Xu, Z., & Jacobsen, H. A., (2005). A framework
for location information processing. 6th International Conference on Mobile Data Management
(MDM05). Ayia Napa, Cyprus.

Key terms
Chi-Square Test: The Chi-square is a test of
statistical significance for bivariate tabular analysis (crossbreaks). This test provides the degree of
confidence we can have in accepting or rejecting
a hypothesis.
Ecological Context: The ecological context is
a set of conditions for a user test experiment that
gives it a degree of validity. An experiment with
real users to possess ecological validity must use
methods, materials, and settings that approximate
the real-life situation that is under study.
Human-Computer Interaction: Human
computer interaction (HCI), also called manmachine interaction (MMI) or computerhuman
interaction (CHI), is the research field that is
focused on the interaction modalities between
users and computers (interface). It is a multidisciplinary subject, relating to computer science
and psychology.
Location-Aware Computing: Locationaware computing is a technology that uses the
location of people and objects to derive contextual
information with which to enhance the application
behaviour. There are two ways to acquire infor-

mation about user context: requiring the user to

specify it or by monitoring users and computer
activity. Sensor technology, such as RFID, could
enable mobile devices to extract information from
user position automatically.
Mobile Tourist Guide: A mobile tourist
guide is a software application with an intuitive
interface, that provides users with multimedia
information when and where needed during their
visit to museums, city centres, parks, and so forth.
Such an application runs on PDA-type terminals
or on cellular phones, and could be augmented
with GPRS (general packet radio service), GPS
(global positioning system), and Bluetooth wireless technology. The guide allows tourists to plan
routes according to preferences and ambient
conditions (weather, timetables, sites of special
interest, etc).
Radio Frequency Identification: Radio frequency identification (RFID) is an automatic identification method based on storing and remotely
retrieving data using small and cheap devices
called RFID tags or transponders. An RFID
tag is an object that can be attached to objects,
products, or persons to identification using radio
waves. Passive tags (with a few centimeter range
of sensitivity) require no internal power source,
whereas active tags (with more long range of
sensitivity, 100 meters) require a power source.
User-Centric Design: User-centric design is
a design process that aims at realizing products
that meet users expectations. The key idea of this
design methodology is to start the design strategy
taking into account the users perspective.

This work was previously published in the Handbook of Research on User Interface Design and Evaluation for Mobile Technology, edited by J. Lumsden, pp. 657-672, copyright 2008 by Information Science Reference, formerly known as Idea Group
Reference (an imprint of IGI Global).

1250

1251

Chapter 3.16

Plagiarism, Instruction,
and Blogs
Michael Hanrahan
Bates College, USA

Abstract
This chapter takes as its point of departure the
Colby, Bates, and Bowdoin Plagiarism Project
(http://ats.bates.edu/cbb), which sought to approach the problem of undergraduate plagiarism
as a pedagogical challenge. By revisiting the decision to publish the projects content by means of
a weblog, the article considers the ways in which
weblogs provide a reflective tool and medium
for engaging plagiarism. It considers weblog
practice and use and offers examples that attest
to the instructional value of weblogs, especially
their ability to foster learning communities and
to promote the appropriate use of information and
intellectual property.

Introduction
Alarmist news accounts of student dishonesty and
cheating abound. More often than not, such stories
describe how universities, colleges, and even high
schools have resorted to plagiarism detection

services to fight a veritable epidemic of student

cheating. The preferred method of combating
academic dishonesty, after-the-fact detection, is
not the only and is perhaps not the best way to address the problem of student plagiarism. Instead of
fighting the lost cause of plagiarism retroactively,
technologists and librarians at Colby, Bates, and
Bowdoin colleges (CBB) collaborated to develop
a program of instruction to educate students about
the principles of academic honesty. The resulting
plagiarism resource site (http://ats.bates.edu/cbb)
includes an introduction to plagiarism, an online
tutorial that tests ones understanding of plagiarism and that provides guidance in the conventions
of citation, and a dedicated weblog that publishes
links to newsworthy articles, notices, and projects
dedicated to plagiarism.
Conceived as a case study, this chapter discusses and evaluates the projects reliance on a
weblog to develop, manage, and publish learning
resources dedicated to plagiarism. In the matter of technical choices, the project developers
were influenced by their commitment to Open
Source Software as well as Creative Commons

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Plagiarism, Instruction, and Blogs

licensing. The former influenced the choice of

weblog software, Drupal (http://www.drupal.org),
and the latter informed the decision to make all
of the projects learning objects and resources
available under an Attribution-Non-CommercialShare-Alike Creative Commons license. These
decisions, it turns out, have allowed the project
to model the appropriate use of online materials
and have retrospectively provided an occasion
to reflect on weblogs as an effective medium for
engaging plagiarism.

Background
Over the past several years, national, regional,
local, and campus newspapers across the globe
have regularly featured articles on student cheating. While academic dishonesty takes any number
of forms (using a PDA, cell phone, or crib notes
during an exam; submitting unoriginal work
copied from an existing publication, cut and
pasted from an online source, or purchased from
a paper mill; or simply peering over a classmates
shoulder during a quiz), plagiarism has emerged
as the most visible form of student cheating. In
many ways, the term threatens to subsume all
other categories of academic dishonesty. A passing visit to the statistics page at Turnitins Web
site (plagiarism.org) reinforces this tendency.
Turnitin, the worlds leading plagiarism detection
service, claims that A study by The Center for
Academic Integrity (CAI) found that almost 80
percent of college students admit to cheating at
least once. Besides generalizing and rounding
up the centers published summary (On most
campuses, over 75 percent of students admit to
some cheating), Turnitins claim isolates a common tendency to conflate a number of dishonest
behaviors with plagiarism. Donald McCabe
(personal communication, August 4, 2004)
explains that the 75 percent figure published
by the CAI represents about a dozen different
behaviors and was obtained in a written survey.

1252

Plagiarism is certainly one form of cheating, but

not all cheating is plagiarism.
Reports of plagiarism in the media tend to
indulge in hyperbole: it is consistently described
as nothing less than an epidemic on campuses.
McCabe (1996), who conducted extensive surveys
between 1996 and 2003, repeatedly found that the
facts do not correspond with the dramatic upsurge
in cheating heralded by the media. McCabe (2000)
has elsewhere observed: Even though Ive stated
on previous occasions that I dont believe these
increases have been as great as suggested by the
media, I must admit I was surprised by the very
low levels of self-reported Internet-related cheating I found. McCabe has subsequently further
qualified his view of the problem: Although
plagiarism appears to have remained relatively
stable during the past 40 years, . . . it is actually
far more prevalent today because many students
dont consider cut-and-paste Internet copying as
cheating (Hansen, 2003, p. 777). More recently,
McCabes evaluation of his 2002-2003 Survey
of U.S. Colleges and Universities identifies an
increase in certain kinds of cheating and a continued misunderstanding of plagiarism among
undergraduates: The past few decades have seen
a significant rise in the level of cheating on tests
and exams. . . . While the data on various forms
of cheating on written assignments do not reflect
the same trend, this may be due to a change in
how students define cheating (2004, p. 127).
To complicate matters further, statistical estimates of academic dishonesty seem to vary due
to contexts (including education level and geography). For example in a recent survey of graduate
students enrolled in 32 business programs in the
United States and Canada, McCabe, Butterfield,
and Trevio (2006) have reported that business
students tend to cheat more than other graduate
students: Fifty-six percent of graduate business
students, compared to 47 percent of their nonbusiness peers, admitted to engaging in some form of
cheating . . . during the past year (p. 299). The
level of self-reported cut-and-paste plagiarism in

Plagiarism, Instruction, and Blogs

this survey, in turn, was 33 percent of the graduate

business students . . . compared to 22 percent for
nonbusiness students (p. 300). A recent study
conducted by the University of Guelph and coadministered by McCabe and Christensen Hughes
(2006) has estimated that 53 percent of Canadian
undergraduate students engage in serious cheating on written work (Gulli, Kohler & Patriquin,
2007). According to Christensen Hughes, Serious cheating on written work includes copying a
few sentences without footnoting, fabricating or
falsifying a bibliography, or turning in a paper
that someone else has written (Cooper, 2007). To
help put matters in a global perspective, a recent
survey of British higher education conducted
by Freshminds.co.uk (with the assistance of the
JISCs Plagiarism Advisory Service and the
Center for Academic Integrity) found that 75
percent of respondents have never plagiarized.
This figure in turn approximates what Turnitin
representatives have elsewhere estimated: in an
interview for the student newspaper at University
of California, Santa Barbara, Paul Wedlake, director of sales for iParadigms, the parent company
of Turnitin.com, is reported to have claimed that
approximately 30 percent of all students in the
United States plagiarize on every written assignment they complete (Ray, 2001).
Regardless of the figures and statistics, the
Internet very much lies at the center of the current fascination with plagiarism. As a result,
the fundamentally ethical nature of the offense
often gets confused with a technological one.
As Patrick Scanlon of the Rochester Institute of
Technology has acknowledged: Plagiarism is not
a technological problemits a problem that has
to do with ethical behavior and the correct use of
sources. And it existed long before the advent of
the Internet (Hansen, 2003, p. 791).
Whether attributed to hype or misperception,
plagiarism and the Internet remain entangled in
the popular and the academic imaginations. The
association is further reinforced by student study
habits, especially their research practices. A recent

Pew report found that nearly three-quarters (73

percent) of college students in the United States
claim to use the Internet more than the library
(Jones, 2002, p. 3). An even greater percentage
of students no doubt resorts to the Internet for
leisureto game, surf, IM, and share music
files. This reliance on the Internet for study and
entertainment has blurred the lines between appropriate and inappropriate cyberpractice and has
promoted the intentional as well as unintentional
misuse of intellectual and creative property.
The Internet is not the sole source of undergraduate plagiarism. The current manifestation
of the problem also can be attributed to non-technological developments, including the increased
tendency among students and their parents (at
least in the English-speaking world) to perceive
higher education as a service industry. That is,
the relegation of higher education to a service for
which one pays has created a scenario in which
students-as-consumers readily expect performance (in the form of good grades) as something
to which they are entitled. This sense of entitlement, in turn, overrides concerns about academic
honesty. Plagiarism, in this light, emerges as
symptomatic of wide-ranging cultural shifts that
are not simply or easily reducible to technological
shifts and developments. Recent commentary on
student plagiarism has provoked observations on
this phenomenon. For example, Frank Furedi,
professor of sociology at the University of Kent,
has observed that In the customer-client culture,
degrees are seen as something you pay for rather
than something you have to learn. Its the new
ethos of university life (A Quarter of Students
Cheating, 2004). This cultural shift and attendant
ethos may very well lie at the root of the misrecognition of plagiarism among undergraduates
that McCabe has observed (Hansen, 2007, p. 777;
McCabe, 2004, 127).
Another significant contributing factor to the
rise of plagiarism is an educational culture that
resists adapting its instructional methods in the
face of advances in technology. This resistance

1253

Plagiarism, Instruction, and Blogs

is forcefully demonstrated by the widespread

adoption of plagiarism detection services. In an
ostensible attempt to counter technology with
technology, schools have settled for a punitive
solution to what is a basically an instructional
problem, and in doing so have escalated rather
than engaged the problem. Turnitin, for example,
adds each assignment submitted to its service to
its databases. This ethically questionable practice
of collecting content has been widely criticized
as ignoring the intellectual property rights of
students: the issue was raised several years ago
by Howard (2001); it surfaced in 2003 at the
center of a controversy at McGill University
(McGill Student, 2006); more recently Mount
Saint Vincent University in Halifax, Nova Scotia, has banned Turnitin for this reason (MSVU
bans anti-plagiarism software, 2006); and high
school students in suburban Washington, D.C.,
have protested their schools subscription to
Turnitin on the same grounds (Glod, 2006). In
most of these cases, iParadigms has defended its
product against this allegation. In a surprising
move, however, the company recently took the
issue into account when renegotiating its contract
with the University of Kansas: Because Turnitin.com retains student papers, the service has
raised intellectual property and copyright issues
Turnitin.com addressed the issue by agreeing
to remove papers from the database if requested
by the KU Writing Center, which administers the
service for KU (Maines, 2006).
Intellectual property matters aside, the
discourse of combating and surveillance that
commonly attend the use and promotion of plagiarism detection technology seems ill-suited in
an instructional setting. Colleges and universities,
after all, have the luxury of privileging learning
in their approach to problem solving. Recognizing that after-the-fact detection of plagiarism is a
lost cause, faculty, educational technologists, and
librarians at Colby, Bates, and Bowdoin jointly developed a plagiarism resource site that attempts to
discourage student plagiarism through a program

1254

of instruction.1 The project takes for granted that

plagiarism is an inescapable condition of learning.
Such a view is by no means unique: Howard (1999,
p. xviii), who has published widely on the subject,
likens plagiarism to imitation: that is, while trying
to find their own voices as writers, inexperienced
students invariably adopt and imitate the voices of
others and rarely in accordance with the scholarly
conventions of attribution. With this view of the
problem in mind, instruction would seem to be
the desirable as well as the necessary solution to
plagiarism. Many educators share this view, and
few have been more vocal over the years than
librarians, including Burke (2004).
Plagiarism certainly has caught the attention
of instructors, librarians, and administrators, but
students by-and-large continue to have a vague
grasp of it. As Jackson (2006) recently discusses,
there is clearly evidence to support the notion
that students, in fact, do not understand plagiarism
and lack the necessary skills to avoid it Many
authors agree that students lack understanding
of what constitutes plagiarism, how to properly
paraphrase, what needs to be cited, and how to
cite sources (p. 420). The many acts of negligence
or ignorance that constitute plagiarism also vary
in degrees of magnitude: failure to observe accurately the rules for citing sources, for example, is
a different order of offense than the inadvertent,
unattributed incorporation of anothers language
or ideas into a written assignment. These lapses,
in turn, are potentially more easily remedied
than the conscious, pre-meditated submission of
anothers work or ideas as ones own.
With this range of plagiaristic practices in
mind, Howard (1995, pp. 799-800) has usefully
identified three categories: outright cheating;
non-attribution as a result of unfamiliarity with
the conventions of citing sources; and patchwriting, or stringing together someone elses
language or ideas without proper attribution.
The CBB plagiarism project seeks to promote
instruction as the best remedy to help teachers
and librarians prevent the last two categories of

Plagiarism, Instruction, and Blogs

plagiarism, which inexperienced students are

especially prone to commit. Based on responses
to the projects instructional materials, these goals
are being met. For example, Suffolk Community
College has used the projects online tutorial in
library workshops on Understanding Plagiarism
and Documenting Sources. Students there have
found the tutorial helpful, andthey are always
particularly interested to learn about the need to
cite paraphrases (Beale 2006). In a recent survey
of an online tutorial on plagiarism, Plagiarism:
The Crime of Intellectual Kidnapping, created
by San Jose State University, Jackson (2006)
has produced convincing evidence that students
need more instruction and practice with proper
paraphrasing (p. 426).
To achieve its goal of providing an instructional
solution to plagiarism, the project takes full advantage of the Internet and responsible cyberpractice: its developers chose an open source content
management system to store, manage, and publish
resources; and its resources are freely available
not only to be viewed and used via the WWW,
but also to be shared, adapted, and re-published
under a Creative Commons copyright license.2 The
resources include a general overview of academic
honesty, an introduction explaining different
kinds of plagiarism, an online tutorial for testing
ones understanding of the various practices that
constitute plagiarism, and dynamic examples of
citations and paraphrasing. The projects Web
site also boasts a dedicated weblog that serves
as a clearinghouse on all matters plagiaristic,
including news items from around the world and
notices on resources, tools, activities, and events
concerning plagiarism in higher education. Taking advantage of Web syndication, the projects
weblog makes its content available via RSS feeds.
As a result, anyone can import the projects news
updates into individual, departmental, or institutional Web sites or weblogs by means of prepared
JavaScripts.3
The CBB Plagiarism Project promotes the
responsible use, re-use, and re-purposing of its

resources so instructors and librarians can address

the problem of plagiarism at the level of local institutional practices, values, and concerns. While
plagiarism undoubtedly is a global problem, its
solution might best be sought at the local level,
where definitions, policies, and expectations
vary widely. The decision to publish content by
means of a weblog has in retrospect leveraged
a technology that has unexpectedly provided a
reflective tool and medium for engaging plagiarism. A consideration of weblog practice and use,
guided by the concept of plagiarism, provides a
framework for understanding the instructional
value of weblogs, especially their ability to foster
and promote learning communities that discourage plagiarism.

Issues, Controversies, and

Problems Associated with
Weblogs
Weblogs basically aggregate meta-data: that is,
they compile information about information in the
form of chronological postings and do not generally publish original content per se. More often
than not, weblogs refer and link to other weblogs
or Web sites, and the result is a highly interconnected network of communication. The resultant
mode of disseminating information has reinforced
certain practices that are commonly understood
as plagiaristic. Researchers at Hewlett-Packard
(HP) Labs have tracked the flow of information
in what they call blogspace and have identified how ideas, regularly unattributed, spread
among blogs (Asaravala 2004). The RSS feeds,
moreover, that enable blogs to publish content in
various ways are often understood as contributing
to plagiarism because they allow unscrupulous
users to capture content (specifically textual data)
and re-purpose it without attribution. Dishonest
practices aside, the HP researchers assert that the
dynamic flow of information in blogspace has a
generative function: individual weblogs link

1255

Plagiarism, Instruction, and Blogs

together to create a complex structure through

which new ideas and discourse can flow. The HP
researchers, Adar, Zhang, Adamic, and Lukose
(2004), conceive of the circulation of information
among blogs as ultimately creative rather than
iterative and original rather than plagiaristic. This
interpretation of blogs isolates tensions that have
attended the reception of the World Wide Web
from its earliest days. Such tensions similarly
inform cultural perceptions of our students use
of the Internet. Their habitual cutting and pasting
and sampling and repurposing are commonly dismissed as purposeless, narcissistic self-expression
and are censoriously viewed as indicative of their
disregard for intellectual and creative property
rights and laws. In a recent article, Ellis (2003)
productively has situated youth cultures creative
as well as plagiaristic practices in contemporary
contexts.
High school and college students operate
with the conscious or unconscious understanding
(based on a lifetime of practice) that any content
available on or accessible via the Web is public
property and free. By re-using and re-purposing
what they find online, students not only contribute to and reproduce a sub-culture founded on
pastiche, but they also develop and acquire the
transferable skills that Ellis (2003) suggests will
enable those interested to join the ever-growing
ranks of knowledge workers in post-industrial
economies. There are drawbacks as well as
benefits to what Ellis envisions as the evolving
new knowledge environment chunks up human experience into multiple, cross-referenced
nuggets dispersed in oceanic cyberspace. Stripped
of our distinctively human purposes, the new
knowledge environment is what George Trow
famously called the context of no context.
This cutting adrift of knowledge results in its
circulation without respect to historical or cultural
contexts and creates a number of potential abuses
and ethical problemsplagiarism among them.
According to Ellis (2003), however, the new
knowledge environment has some potential

1256

benefits that he describes in terms similar to the

HP researchers description of blogspace: This
environment favors those who can apprehend the
interconnectedness of things, create bridges and
connections, spark associations and create the
clat of montage. . . . Social network analysis,
network topology and other new perspectives are
being framed to help us understand the natural
dynamics of this new environment.
The dynamics of the blogosphere represent
potentially exciting developments in cyber-communication, but they simultaneously revisit many
of the criticisms commonly invoked to condemn
the WWW. The Web is many things to many
people: a commerce tool for business; a recruitment tool for new religions; a protest space for
political activism; a play space for dedicated
gamers; and so on. Regardless of its intended use
or unintended abuse, the WWW has provided
interested parties with a readily available means
to publish content of all sorts, and its users have
responded by taking advantage of its publishing
capabilities: according to a recent Pew report,
practically half (or 44 percent) of adult users of
the Internet in the United States have created and
published content (Online Activities and Pursuits,
2004). The value, usefulness, and originality of
that content are an endless source of debate, and
the popularity of weblogs has provided additional
fodder for critics who question the informational
and instructional value of the WWW.
Weblogs tend to promote a confessional mode
of discourse that celebrates self-referentiality
(McDonald 2004). This tendency has fueled the
criticism that blogs have ushered in a new era of
navel-gazing. The form and content of many personal blogs reinforce this view, but virtual personal
diaries do not exhaust the uses and applications
of blogs. Adar and Adamic (2005) have suggested
that beyond serving as online diaries, weblogs
have evolved into a complex social structure, one
which is in some ways ideal for the study of the
propagation of information. Their observation
posits an interrelationship of information and its

Plagiarism, Instruction, and Blogs

circulation that previous scholars have variously

notedfrom McLuhans The Medium is the
Message to Clanchys From Memory to Written
Record to Brown and Daguids The Social Life
of Documents.
In their approach to the interconnectedness
of information and its circulation, Brown and
Daguid (1996) have considered the ways in
which developing technologies have historically supported social relations in new ways. A
wide range of disciplines and historical examples
inform their understanding. Enlisting Anselm
Strausss notion of social worlds, Brown and
Daguid (1996) describe a dynamic of group formation that can further the understanding of the
culture of weblogs. Following Strauss, Brown
and Daguid (1996) observe that once formed,
social worlds continually face disintegration (as
dissenting members split off into sub-worlds).
The blogosphere seems largely populated by such
sub-worlds that all too often appear to celebrate
a community of one; that is, if one is viewing
weblogs as repositories of content rather than as
nodes within a network. The flow of information
that populates many weblogs, as tracked by the
HP researchers, establishes a social matrix that
assumes both implicit and explicit communities.
Dedicated weblog writers can be roughly divided
into two main types: political bloggers and technobloggers. This overly simplistic distinction falls
short of capturing the full range of representative
blogging sub-worlds (edubloggers, for examples),
but it conveniently describes two influential communities of bloggers.
Drawing on the theory of the imagined community proposed by political scientist Anderson
(1991), Brown and Daguid (1996) further consider
the ways in which popular cultural items, such
as journals, novels, pamphlets, lampoons, ballad
sheets, and so forth contributed to the formation
of national identity in the American colonies
leading up to the Revolution. Citing daily newspapers, in particular, they point out that it was their
widespread circulation and not just their content

that helped foster Colonial Americas sense of

nationhood. The similarities between newspapers
and weblogs are instructive. Many observers have
noted that blogs have greatly contributed to if not
forever changed journalism. McDonald (2004), for
example, understands blogging to be a genuinely
positive development in mass communication,
and particularly in publishing and journalism.
He attributes the popularity of weblogging to its
adoption by the journalistic establishment. I
would attribute their popularity to their embrace
by alternative journalists, especially the proliferation of warblogging in the aftermath of 9/11
and the subsequent events leading up to the U.S.
invasion of Iraq in 2003.
Weblog pioneer and advocate Dave Winer,
moreover, has speculated that newspapers will
ultimately be replaced by weblogs as news sources
in the not too distant future. This prediction is
based in no small part on the publishing ability of weblogs, which has greatly extended the
publishing capacity of the WWW. According to
Winer, In a Google search of five keywords or
phrases representing the top five news stories of
2007, weblogs will rank higher than The New
York Times Web site (Long Bet).

Solutions and
Recommendations:
Building Learning
Communities via Weblogs
Blogs are powerful and flexible publishing tools:
they publish content rapidly and easily; they provide an archive for content that is readily searchable by date, subject, or keyword; and they can
also publish their content in a number of ways,
including dedicated Web sites as well as RSS
feeds that can populate other Web sites, weblogs,
aggregators, e-mail clients, and Web browsers.
That which has secured their popularity and
wide reception (the rapid creation, publication,
and circulation of information) also represents

1257

Plagiarism, Instruction, and Blogs

their greatest potential for instruction. Librarians,

technologists, and instructors can capitalize on
blogs for making available a range of resources
and information to targeted usersstudents, staff,
faculty, and colleaguesboth on their own as well
as on other campuses. They can do so, moreover,
with their own content or with content developed
entirely by other institutions. This latter ability,
importing content from elsewhere, demonstrates
how blogs can reinforce the responsible and productive use and circulation of information.
The hallmark features of weblogs (the rapid
creation and dissemination of content) are extremely useful for fostering learning communities
whose members resort to various methods and
media for instruction and information. Certain
integral aspects of weblogs further promote
their instructional potential. Weblogs have not
only made publishing content easier but more
socialthey open content development up to a
group by means of their ability to allow various
levels of access to different groups of users; and
they invite dialogue between creators and readers
of content by permitting the exchange of comments
within the blog as well as between blogs. Weblogs
are dynamic in a couple of ways: the content
posted on them changes as information is added
and they allow users to interact by carrying on
a dialogue. This dialogic aspect of blogs enables
content developers to work towards breaking
down the distinction between the creator and the
user of content. This feature of blogs participates
in the trend already discerned by Pew: that the
consumers of Web content are also largely the
producers of it.

Future Trends: Engaging

Plagiarism via Multimedia
The controlled dissolution of boundaries between
producers and users of content (or between instruc-

1258

tors and students, for that matter) has emerged

as a valuable lesson of the CBB projects use of a
weblog. Successful instruction in plagiarism must
strive to increase the awareness of the difference
between the creation of new and the appropriate
use of existing content. The project has sought to
promote this awareness in practice by example
and in theory by instruction. The content is freely
available to be used and re-purposed according
to an Attribution-Non-Commercial-ShareAlike
Creative Commons Deed. The project developers
have also sought to create learning objects that help
socialize students into the culture of academics,
which is founded on what Green (2002, p. 171)
has described as the norm of attribution. Most
teachers take for granted the scholarly conventions used to avoid plagiarism. Recognizing
the profound difference that exists between the
initiated and the uninitiated, the CBB Plagiarism
Project has set out to provide students with guidance and instruction in the standards of academic
integrity. In doing so, it strives to facilitate our
students initiation into the norms and practices
of the academic community.
Looking ahead to further development, the
projects next phase will involve creating a more
adaptive learning environment for engaging
plagiarism. While the weblog provides a valuable
means to deliver, create, and respond to content,
the text-based nature of that content may reinforce some of the limitations of online tutorials
as instructional resources. Jackson (2006, pp.
423-26) has recently considered the effectiveness
of plagiarism instruction online. By developing
media-rich content about the subject (including
audio, video, and animation), the project would
create a range of resources that better suit diverse
learning styles. In doing so, the project would be
more responsive to the needs of its users and would
further realize its goal of helping to integrate
students into academic cultural practice.

Plagiarism, Instruction, and Blogs

Conclusion
An increased use and understanding of media in
the curriculum, moreover, may very well allow
faculty to harness the creative energies of students in a way that deals with plagiarism in both
practical and theoretical terms readily understood
by students. Current wisdom on how to avoid
plagiarism has emphasized the need to rethink
written assignmentsfor example, essays should
be conceived of as ongoing processes consisting
of specific, discrete stages or components, all of
which are submitted for review, evaluation, and
assessment, rather than a single finished product
submitted in its entirety only once. In rethinking assignments, instructors may also want to
begin to rethink what writing is and to encourage
non-traditional forms of writing. I have in mind
here the creation of fictional and non-fictional
narratives, reports or accounts by means of multimediadigital video and audio or computer animation and graphics or any combination of these
and other media. Just as the weblog has emerged
as a reflective tool for considering plagiarism, a
media-rich learning environment would allow
students to begin to understand plagiarism in new
and perhaps more compelling ways. In a recent
essay on plagiarism, the novelist Jonathan Lethem
(2007) describes what it is like to be cut adrift in
our contemporary media environment:
The world is a home littered with pop-culture
products and their emblems. I also came of age
swamped by parodies that stood for originals yet
mysterious to me Im not alone in having been
born backward into an incoherent realm of texts,
products, and images, the commercial and cultural
environment with which weve both supplemented
and blotted out our natural world. I can no more
claim it as mine than the sidewalks and forests
of the world, yet I do dwell in it, and for me to
stand a chance as either artist or citizen, Id probably better be permitted to name it.

In the academy, students are encouraged to

name and when appropriate cite their sources,
influences, and inspirations. However, finding
themselves, like Lethem, in a world already created and populated with signs, they need to learn
how to negotiate the conventions and practices of
that world and to decode its constituent signs. Educators should begin to make use of the multitude
of media that figures our manifold experiences of
the world. The energy and creativity generated by
such a diversely constituted learning environment
would permit powerful models for rethinking our
engagement of plagiarism.

References
Adar, E. & Adamic, L. A. (2004). Tracking
information epidemics in blogspace. Retrieved
October 5, 2006 from http://www.hpl.hp.com/research/idl/papers/blogs2/index.html
Adar, E., Zhang, L., Adamic, L. A., & Lukose, R.
M. (2004). Implicit structure and the dynamics
of blogspace. Retrieved October 2, 2006 from
http://www.hpl.hp.com/research/idl/papers/
blogs/index.html
Anderson, B. (1991). Imagined communities: Reflections on the origin and spread of nationalism.
London: Verso.
Asaravala, A. (2004). Warnings: Blogs can be
infectious. Wired News. Retrieved October 2,
2006 from http://www.wired.com/news/culture/0,1284,62537,00.html
Beale, P. (2006). E-mail, October 6.
Blog Epidemic Analyzer. Retrieved October 5,
2006 from http://www.hpl.hp.com/research/idl/
projects/blogs/index.html
Brown, J. S. & Daguid, P. (1996). The social
life of documents. First Monday, 1(1). Retrieved

1259

Plagiarism, Instruction, and Blogs

October 2, 2006 from http://www.firstmonday.

dk/issues/issue1/documents/index.html
Burke, M. (2004). Deterring plagiarism: A new
role for librarians. Library Philosophy and Practice, 6(2), 1-9.
Center for Academic Integrity. Retrieved October
2, 2006 from http://www.academicintegrity.
org/cai_research.asp
Clanchy, M. T. (1979). From memory to written
record, England, 1066-1307. Cambridge: Harvard
University Press.
Cooper, L. (2007). Cheating a big problem. At
Guelph. Retrieved February 22, 2007, from
http://www.uoguelph.ca/atguelph/06-10-25/featurescheating.shtml
Ellis, L. (2003). MP3s, Plagiarism, and the Future
of Knowledge. The ClickZ Network. Retrieved
October 2, 2006, from http://www.clickz.com/
showPage.html?page=3080961
Freshminds.co.uk. Retrieved October 2, 2006,
from http://www.freshminds.co.uk/FreshMinds_
plagiarism_survey.pdf
Glod, M. (2006). Students rebel against database
designed to thwart plagiarists (p. A01). Washington Post. Retrieved October 2, 2006 from http://
www.washingtonpost.com/wp-dyn/content/article/2006/09/21/AR2006092101800.html
Green, S. (2002). Plagiarism, norms, and the
limits of theft law: Some observations on the use
of criminal sanctions in enforcing intellectual
property rights. Hastings Law Journal. 54(1),
167-242
Gulli, C., Kohler, N., & Patriquin, M. (2007).
The great university cheating scandal. MacLeans. Retrieved February 22, 2007 from http://
www.macleans.ca/homepage/magazine/article.
jsp?content=20070209_174847_6984

1260

Hansen, B. (2003). Combating plagiarism: Is the

Internet causing more students to copy? The CQ
Researcher, 13(32), 773-796.
Howard, R. M. (1995). Plagiarisms, authorships,
and the academic death penalty. College English,
57(7), 788-806.
Howard, R. M. (1999). Standing in the shadows
of giants. Stanford, CT: Ablex.
Howard, R. M. (2001). Point of view: Forget about
policing plagiarism. Just teach. The Chronicle
Review. Retrieved October 2, 2006 from http://
chronicle.com/prm/weekly/v48/i12/12b02401.
htm
Jackson, P. A. (2006). Plagiarism instruction
online: Assessing undergraduate students ability to avoid plagiarism. College and Research
Libraries, 67(2), 418-427.
Jones, S. (2002). The Internet goes to college:
How students are living in the future with todays
technology. Washington, DC: Pew Internet and
American Life Project.
Lethem, J. (2007). Ecstasy of influence: A plagiarism. Harpers Magazine. Retrieved February
22, 2007 from http://www.harpers.org/TheEcstasyOfInfluence.html
Maines, S. (2006). KU renews anti-plagiarism
software subscription. LJWorld.com. Retrieved
October 7, 2006 from http://www2.ljworld.
com/news/2006/oct/04/ku_renews_antiplagiarism_software_subscription/?ku_news
McCabe, D. L. (2000). New research on academic
integrity: The success of Modified honor codes.
College Administration Publications. Asheville,
NC. Retrieved October 2, 2006 from http://www.
collegepubs.com/ref/SFX000515.shtml
McCabe, D. L. (2004). E-mail August 4, 2004.
McCabe, D. L., Butterfield, K. D. & Trevio, L.
K. (2004). Academic integrity: How widespread

Plagiarism, Instruction, and Blogs

is cheating and plagiarism? In D. R. Karp & T.

Allena (Eds.), Restorative justice on the college
campus: Promoting student growth and responsibility, and reawakening the spirit of campus
community (pp. 124-35). Springfield, IL: Charles
C. Thomas.
McCabe, D. L., Butterfield, K. D. & Trevio, L.
K. (2006). Academic dishonesty in graduate business programs: Prevalence, causes, and proposed
action. Academy of Management Learning &
Education, 5(3), 294-305.
McCabe, D. L. & Christensen Hughes, J. (2006).
Understanding academic misconduct. The
Canadian Journal of Higher Education, 36(1),
49-63.
McCabe, D. L. & Trevio, L. K. (1996). What we
know about cheating in college. Change, 28(1).
Academic Search Premier.

Plagiarism.Org. Retrieved October 2, 2006, from

http://www.plagiarism.org/plagiarism_stats.
html
A Quarter of Students Cheating. (2004). BBCNews.
Retrieved February 16, 2007 from http://news.bbc.
co.uk/2/hi/uk_news/education/3852869.stm
Ray, D. (2001). Web site may catch, deter plagiarized work. Daily Nexus Online: UC Santa
Barbaras Student Newspaper. Retrieved October
3, 2006 from http://www.ucsbdailynexus.com/
news/2001/1252.html
Winer, D. Long Bet. Retrieved October 2, 2006
from http://www.longbets.org/

Endnotes
1

McDonald, N. (2004). The future of weblogging.

The Register. Retrieved October 2, 2006, from
http://www.theregister.co.uk/2004/04/18/blogging_future
McGill student continues fight against anti-plagiarism Web site. (2003). CBC.com. Retrieved
October 6, 2003 from http://www.cbc.ca/news/
story/2003/12/27/plagiarism031227.html
McLuhan, M. (1964). The medium is the message.
Understanding Media: The Extensions of Man.
New York: McGraw Hill.
MSVU bans anti-plagiarism software. (2006).
CBC.com. Retrieved October 2, 2006, from
http://www.cbc.ca/canada/nova-scotia/story/2006/03/08/ns-msvu-plagiarism20060308.
html
Online Activities and Pursuits. (2004). Retrieved
October 2, 2006, from http://www.pewinternet.
org/report_display.asp?r=113

Original project members included Judy

Montgomery and Sue ODell, Bowdoin College; Zach Chandler and Marilyn Pukkila,
Colby College; and Thomas Hayward, Bates
College. Jim Hart at Bates College served
as a technical consultant from the projects
inception and generously provided extensive support by administering the Linux
server that hosted the projects weblog and
resources.
The site is driven by Drupal, a PHP-MySQLbased open-source content management
system, which is freely available to download
at http://www.drupal.org. For further details
on Creative Commons, see http://creativecommons.org.
See http://leeds.bates.edu/cbb/rss/rss.html
for instructions and guidelines. For an example of the feed in action, see the Missouri
State University Libraries, http://library.missouristate.edu/resources/cheating.shtml.

This work was previously published in Student Plagiarism in an Online World: Problems and Solutions, edited by T. Roberts,
pp. 183-193, copyright 2008 by Information Science Reference, formerly known as Idea Group Reference (an imprint of IGI
Global).
1261

1262

Chapter 3.17

Twin Wiki Wonders?

Wikipedia and Wikibooks as Powerful

Tools for Online Collaborative Writing
Meng-Fen Grace Lin
University of Houston, USA
Curtis J. Bonk
Indiana University, USA
Suthiporn Sajjapanroj
Indiana University, USA

Abstract
Web 2.0 technologies empower individuals to
contribute thoughts and ideas rather than passively survey online content and resources. Such
participatory environments foster opportunities
for community building and knowledge sharing,
while encouraging the creation of artifacts beyond
what any single person could accomplish alone.
In this chapter, we investigate the emergence and
growth of two of such environments: the highly
popular Wikipedia site and its sister project,
Wikibooks. Wikipedia has grown out of trends for
free and open access to Web tools and resources.
While Wikipedians edit, contribute, and monitor
distinct pieces of information or pages of documents, Wikibookians must focus on larger chunks

of knowledge, including book modules or chapters

as well as entire books. Several key differences
between these two types of wiki environments
are explored. In addition, surveys and interviews,
conducted with Wikibookians, shed light on their
challenges, frustrations, and successes.

Introduction
Thomas Friedman, in his 2005 book, The World
is Flat, talks about 10 forces that have flattened
the world in terms of economic globalization.
The word flat acts as a metaphor to symbolize
the leveled playing field on a global scale. In
Friedmans (2005) view, when the playing field
is leveled, everyone can take part. And he means

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Twin Wiki Wonders?

everyone! Talking from a business perspective,

Friedman charts the progress of globalization
from what he describes as 1.0 to 3.0. Globalization
1.0 focused on country to country relationships,
such as treaties and trade. In Globalization 2.0,
such relationships moved down to a companyto-company level. We are now entering an age of
Globalization 3.0, where the rise of the individual
comes into focus. It is in this third phase of globalization, wherein individuals obtain new powers
and freedoms to participate socially, economically,
politically, and educationally with others around
the world. Time Magazine recognized this trend,
and in 2006 named You as the person of the
year (Time Magazine, 2006/2007). In effect, the
Year 2006 signified the trend toward empowering technology users with what was called Web
2.0 technologies, which allowed individuals to
generate ideas online instead of just reading and
browsing through someone elses Web pages
(Grossman, 2006/2007a, 2006/2007b).
Lessig (2005) called this phenomenon the
read-write Web, to contrast the read-only Web,
where users were merely passive consumers of
information. Most fundamentally, the read-write
Web dramatically enhances the power of individuals, and fosters a participatory culture of building,
tinkering, learning, and sharing (Brown, 2006).
Typically included in Web 2.0 technology lists
are wikis, podcasts, blogs, online photo albums,
and virtual worlds such as Second Life. In particular, wikis offer an innovative mechanism of
computer-supported collaboration. This highly
unique social phenomenon utilizes armies of
volunteers who work without pay, recognition, or
much acclaim to continually contribute, update,
or edit resources online (Lin, 2006).
In this chapter, we primarily offer insights
into the recent emergence of two wiki developments; namely, Wikipedia and Wikibooks. First,
we detail what a wiki is. Subsequently, we briefly
review existing wiki-related research, and then
document the emergence and growth of what is
the most popular wiki today; Wikipedia. After

presenting our research findings in Wikibooks,

we compare and contrast some of the underlying
principles and features of Wikipedia to those
found in Wikibooks.

Background
Brandon Hall (2006) defines a wiki as a collection of Web pages that can be easily viewed and
modified by anyone, providing a means for sharing
and collaboration. These are open-ended, generative, and unstructured environments (Honegger,
2005; Leuf & Cunningham, 2001; Lio, Fraboni,
Leo, 2005). Pioneered by Ward Cunningham
in 1995, wikis are online spaces for recording
information, sharing knowledge, typically in
collaboration with others. Each modification is
recorded as the history of a document. The history page records the time of change, the person
who made the change, and the changes that were
made. Such a mechanism not only permits page
retraction by anyone, it also behaves as a podium
for reputation management. In addition, the history page permits open examinations of each
revision, allowing each version to be compared
and contrasted by anyone.
Many universities have picked up the wiki
fever and started using its functions for information sharing. For example, Stanford has an institutionalized wiki wherein students can contribute
information on places to eat, workout, study,
socialize, and so forth. (Campus Technology,
2006b). As this Web site indicates, there is now a
wave of student-contributed wiki resources. Similarly, MIT has created the Center for Collective
Intelligence, where people from around the planet
could come and solve huge scientific, social, and
business problems (Campus Technology, 2006a).
The underlying belief of these wiki projects
indicates that, collectively, the human race can
act more powerfully than it can at an individual
level. As a prime example of this principle, in
early February, 2007, Penguin books announced

1263

Twin Wiki Wonders?

A Million Penguins, a Web site where people

from around the planet will help write a novel
(Reuters, 2007). Similarly, in the academic world,
faculty at MIT and Wharton School of Business
invited thousands of individuals or potential authors to help write a collaborative online textbook
at the site We are Smarter than Me (Campus
Technology, 2007).
There is scant academic research on wiki
technology and resources even though we are in
the midst of a proliferation of wiki-based projects. In particular, the research on Wikibooks,
which emerged with much enthusiasm in 2003
(Wikibooks, 2007a), is particularly thin. Some
Wikipedia-related research has explored the accuracy of the content (Lih, 2004; Rosenzweig,
2006), historical flow visualizations of contributions (Viegas, Wattenburg, & Dave, 2004),
the formality of the language used (Emigh &
Herrring, 2005), hackers and trolls (Schachaf &
Hara, 2006), contributions within collaborative
authoring environments (Korfiatis, Poulous, &
Bokos, 2006), differences between experts and
novices (Bryant, Forte, & Bruckman, 2005), and
cultural differences in Wikipedia contributions
and editing behaviors (Pfeil, Zaphiris, & Ang,
2006). Such studies are, of course, just the starting
point for a long line of research on wikis.

A brief history of the brief

history of Wikipedia
Wikipedia is a free online encyclopedia to which
anyone can contribute. Created in January 2001,
Wikipedia started out as a side experiment of a
failing project called Nupedia.com, created by
Larry Sanger and Jimmy Wales, that also aimed
to create a free online encyclopedia. However,
Nupedia.com employed a traditional prepublishing review process that demanded an elaborate
and often long procedure of coordination among
submission, reviewing, and negotiation. Few articles were published as a result of this arduous

1264

process. Wikipedia was created as an experiment

to bypass this prepublishing review process and
empower the postpublication, real-time, peerreview procedure among volunteer collaborators
(Ciffolilli, 2003; Voss, 2005).
Today, Wikipedia is perhaps the largest instance of a wiki. With easy to explore and open
technology at its core, Wikipedia quickly became
an online volunteer community with its devotion
to the creation of a free encyclopedia where the
division of labor is facilitated by information
technology. That volunteer community now
includes more than 50,000 individuals who have
made at least 10 contributions to Wikipedia
(Rosenzweig, 2006).
According to Alexa, a Web ranking service, as
of February 7th, 2007, Wikipedia had become the
12th most visited Web site across all languages
(alexa.com, 2007). During the past 6 years, Wikipedia has enjoyed tremendous, if not exponential
growth in terms of registered users, number of
articles, and number of languages. Not surprisingly, Wikipedia quickly overshadowed Nupedia.
com and caused Nupedia.com to close in 2002.
As of February 2007, the combined Wikipedia
site included articles in 250 languages, 6.4 million articles, and almost 3.5 million contributors
(Wikipedia, 2007d). The English language version
of Wikipedia continues to be the largest site at
more than 1,624,000 articles, while some other
language versions remain quite small (Wikipedia,
2007b, Wikipedia, 2007c). There are more than
300,000 articles in German, French, Polish, and
Japanese languages, with the German Wikipedia site now pushing over 500,000. This is an
impressive amount of work to accomplish since
inception 6 years ago. It is also quite obvious that
a vast number of willing volunteers awaited just
such an opportunity to contribute their time and
knowledge.
Wikipedias exceptional growth was perhaps
best captured and described by Voss (2005), who
conducted one of the first quantitative analyses of
the structure of Wikipedia. From the elaborated

Twin Wiki Wonders?

numbers, charts, and mathematic formulas, Voss

presented a comprehensive picture of the overall
Wikipedia community. As would be expected,
Wikipedia experienced a period of linear growth,
while the more spiked or exponential growth that
is often alluded to in the media did not occur
until about 15 months after inception in April
2002. Since that time, the number of new articles
in Wikipedia has increased at various speeds
among different languages. Even though anyone
with Internet access on this planet is invited to
contribute to Wikipedia, according to the Voss
study, the average number of authors per article
remains four to five, a number that is perhaps
much lower than most people would assume.
Even more interesting is that about half (48%)
of the articles had less than five authors and
about one-third (28%) of the articles in German
Wikipedia had only been edited by one author. In
addition, one-third of the Wikipedia authors had
only contributed one article, and just one in five
Wikipedians had been involved in more than 16
articles. Granted that Voss excluded anonymous
edits in his calculation, these numbers indicate that
not everyone will contribute, or at least continue
to contribute, just because they can.
In summary, then, the Wikipedia community
has the following characteristics. It is a volunteerbased project to which everyone can contribute.
Wiki technology facilitates certain divisions of
labor, enables the ease of contributing and participation, and records all change histories for
easy examination and recovery. It is a rapidly
growing community where authority is gained
through individuals active participation. Its
social structure does not necessarily recognize
people based on their expertise, even though it
does employ a promotion mechanism for various
roles in the community. Positioning itself as an
encyclopedia, Wikipedia inherits the perception
of being a reliable reference source. However, its
credibility continues to be questioned.

Wikibooks and other wiki

sites
Sister projects to Wikipedia coordinated by
the Wikimedia Foundation include Wikibooks,
mentioned previously, as well as Wikispecies,
Wikiquote, Wikinews, Wikiversity, Wiktionary,
Wikisource, Commons, and Meta-Wiki. Wiktionary, for instance, is a free, community created
online dictionary and thesaurus with more than
300,000 entries in nearly 400 different languages
(Wiktionary, 2007). Wikiquotes contains an index
of thousands of quotations from famous people
from Laozi, William Shakespeare, Aristotle,
Hellen Keller, and Martin Luther King, as well
as from literacy works, television shows and
movies, news pages, and other creative efforts
(Wikiquotes, 2007). Wikinews, like other online
news services, contains current events making
headlines. But as folks like Rosenzweig (2006)
point out, it often can break leading stories before
the other news services since its reporters cover
the planet.
In our own work, Wikibooks has been the primary focus. The Wikibooks Web site, originally
named the Wikimedia Free Textbook Project and
Wikimedia Textbooks, was initiated on July 10,
2003 (Wikibooks, 2007a). It was the brainstorm
of Karl Wick, who was searching for a place to
host his books on organic chemistry and physics
as a means to bring such educational resources to
the people of this planet while addressing the high
costs and other limitations of existing books and
materials on these topics. By the end of that month,
there were 123 modules or chapters posted and
47 registered users at that site. Two months later,
there were 530 modules and 150 registered users.
Since it began, controversies and criticisms have
been posed regarding the scope of the Wikibook
project, copyright and licensing issues, the quality of the books posted, and the large number of
unfinished book projects.

1265

Twin Wiki Wonders?

The coordinator of one or more Wikibook

projects is called a Wikibookian. Given that in
less than 4 years Wikibooks has become a collection of over 1,000 free online book projects
containing more than 20,000 modules or chapters (Wikibooks, 2007b), there are many active
Wikibookians. The Wikibooks Web site contains
collaboratively written textbooks and nonfiction
books, study guides, and booklets are socially
negotiated documents. There are both books for
adult learners as well as the more recent emergence
of junior books for younger learners, ages 8 to
11 (Wikibooks, 2007c) (see Figure 1). The junior
Wikibooks site is currently creating full-color
booklets on topics such as bugs, big cats, the kings
and queens of England, and dinosaurs.
Such environments offer hope for someday
providing access to educational books, study
guides, and other documents to every connected
learner and in any language. Nevertheless, there
are myriad issues, questions, and problems related
to online collaboratively authored books, such

Figure 1. Screen shot of the Wikibooks Web site

1266

Figure 2. The Wikibook environment

Twin Wiki Wonders?

as those found at the Wikibooks site, that need

to be better understood. Figure 2 depicts the key
aspects of the Wikibook environment.

an online book could ever be complete, and suggestions for improvement.

Studying Wikibookians

Findings

During a 4-month period in early 2006, we

observed how Wikibookians communicated
and shared ideas with each other. Based on our
observations and literature review, we designed
a close-ended survey that encompassed basic
demographic data, as well as a set of questions
geared toward issues involved in the process of
creating a Wikibook from a sociocultural point
of view (Bruns & Humphreys, 2005). Through
communication channels frequented by active
members, we obtained a user list providing
around 45,000 Wikibookians names who had
previously registered an account at the Wikibook
Web site. Using this list, in the spring of 2006,
we randomly sent survey invitations to around
1,500 Wikibookians whose status was shown as
active via the contacting function found at the
Wikibook Web site. We used an online survey
tool, SurveyShare, to distribute the survey and
collect the data. We received 80 responses to our
survey. Ninety percent of the survey respondents
had been active in Wikibooks for more than 6
months, 72% for more than one year, and nearly
one-third for 2 years or more. In addition, 82%
were under age 35, 97% were men, and more
than half did not have a four-year college degree
(Sajjapanroj, Bonk, Lee, & Lin, 2007).
After tallying the surveys, we randomly selected 15 people from the survey respondents for
follow-up e-mail interviews. The e-mails were sent
in July 2006, and eight Wikibookian participants
joined this part of the study. Participants were
asked to answer a 12-question e-mail interview
related to how they had started in Wikibooks,
previous experiences in such an online collaboration process, the future of Wikibooks, advantages
and disadvantages of this environment, whether

The survey and interview data helped us understand the challenges, frustrations, and successes
of Wikibookians. From the 8 participants and
12 e-mail interview questions, five key themes
emerged. These themes were as follows:
1.
2.
3.
4.
5.

The expression of authority in Wikibooks

The nature of collaboration process in Wikibooks
Completeness and success of ones Wikibook
project
Advice to Wikibook wannabes
The future of Wikibooks

The following sections describe the themes and

present the different views participants offered.
Based on these findings, as well as existing wiki
literature, we offer a comparison of the Wikipedia
and Wikibook environments.

The Expression of Authority

in Wikibooks
Authority
Working on a wiki as an editable Web platform
could be rewarding and challenging at the same
time. The number one critique of Wikipedia has
been its authoritative value. It is not surprising
that such questions and concerns about authority
also appeared in our Wikibook findings. Some
participants seemed to disregard the traditional
view of an expert or authority:
people can work together on a wiki and come up
with a result that is better than something written
by one or a couple of experts. There isnot one

1267

Twin Wiki Wonders?

person in charge who can make the harddecisions

that everyone will respect. (Participant 1)
Nevertheless, some seemed to also demand
the skill of experts in order to create a good
quality book:
I expected it to be not very effective, as its hard
to let people work together at all, especially on
this large amount, and even more when its spread
over multiple pages with a special subject. You
should need at least two specialists, as both have
to understand the subject very good. That appeared
to be hard indeed. (Participant 6)
Wikibookian views about community or multiple contributor authority over the knowledge of
experts revealed the nontraditional belief that
each one of us possesses our unique knowledge
that could contribute to the creation of a Wikibook.
On the other hand, there seemed to be a demand
for such knowledge to be able to qualify at an
acceptable level as an expert.

Control
If looking at the issue of authority at a deeper level,
we further discovered the concerns regarding
control over contribution. Who has the right to
contribute? Who can decide who has the right to
contribute? Who can decide whose contribution
is superior to others? In other words, is it true that
in the world of Wiki, everyone stands equally?
Certainly some did not believe so:
The biggest problem I see are the control guilds
that have sprung up. They can be the Christian
groups, or the anti-metaphysicists that go around
labeling everything pseudoscience because they
lack the ability to see the difference between the
two. (Participant 2)
Most participants addressed the issue about
control over contribution from the perspective

1268

of being able to monitor changes. For instance,

Participant 7 wrote to us about the importance of
being able to track the contributions and changes
of each contributor. Participant 3 elaborated
further:
I think the fact that Administrators monitor the
content and make sure that if it is inaccurate or
copyright protected that it cannot be posted is a
very important aspect of WikiBooks/Wikipedia that
should be emulated by others. (Participant 3)
One other aspect of control over contributions
surfaced as well. As shown in the following quote,
Participant 6 expressed the desire to create a
temporarily safe environment during the creation
of a Wikibook:
Make a special area where one set group of people
can take over a book for a time, for example,
to enable one class or onegroup ofprofessors
[to] develop materials in a protected environment where, at least for a time, they have the
final authority of whatever happens in that area.
(Participant 6)
As indicated, Participant 6 believed that having
authority even for a period of time was deemed
important. Participant 4 shared that the biggest
disadvantage is a sacrifice in control over the direction of the book, which seemed to indirectly
echo Participant 6s point. Here, the control issue
seems to center on setting the overall goal of the
book and the initial creation of the first draft.
As demonstrated, even though most participants did not directly bring up the question
about control, through interview excerpts, we
observed how multiple dimensions of authority
were realized among Wikibookians. Our survey
data indicated that an overwhelming majority of
the surveyed Wikibookians believed that Wikibooks promote individual accountability in terms
of ones contributions to an online book project
(15% strongly agree, 54% agree). However, still

Twin Wiki Wonders?

31% of them disagreed that individuals could be

held accountable for the accuracy of contents
contributed. It is not surprising to see that Wikibookians wrestled with these questions, which
leads to the next section about how Wikibookians
collaborate.

Coordination is key. So [is] communication between editors, contributors, etc. is important. It

must also be easy to use, fast and one must quickly
be able to see what was changed, who changed
the document and if necessary, change it back.
(Participant 8)

The Nature of Collaboration Process

in Wikibooks

Collaboration

Coordination
Working in a Wikibook project requires coordination, communication, an understanding of the
collaborative writing process, and a great deal
of time management. In particular, Participant 4
shared many of his perspectives about the inner
workings of a Wikibookian and the process of
Wikibook creation:
Development has been slower than expected. I
expect to be able to finish the bookin about
two years.
My experience has been that Wikimedia projects
promote much more formal rules of etiquette, but
a much less structured writing process.
Maintaining a wiki is much more a challenge for
social issues than for technological issues. Above
all, a collaborative writing community must have
a common vision, specifically a vision to create
a written work.
Interestingly, he suggested that software be
created that makes it easy to post a positive contribution and more difficult to post something
negative.

Communication
Clearly, issues about communication stood out in
many of the Wikibookian observations and suggestions. Some of the comments included:

The concept of a wiki assumes a collaborative

process. While many participants clearly showed
their agreement, two indicated their disappointment when no other took interest in their book:
I also hoped that collaborators would take an
interest in the book as it developed. Although there
have been isolated contributions, no one else has
taken an interest in making a real investment in
the book. I still hope this will eventually happen.
(Participant 4)
My expectation was to help create a free, collaboratively written textbook. At this point I have done
almost all of the work on the book. Admittedly,
word about the book hasnt gotten out, but hopefully once it does others will begin to contribute.
Until that happens, my expectations will not,
technically, have been met. (Participant 7)

Resolving Disagreements
Even though some seemed to be lone writers,
disagreements are bound to happen when working with strangers from around the world. We
asked the Wikibookians about how they resolved
differences. Again, communication provided the
bridge to reconcile different opinions, including
mechanisms such as the talk page, a mediator,
private message, or revert the changes. However,
Wikibookians differed in how they would approach or resolve the differences. As the following
quotes indicate, some would revert the changes,
some would discuss them first, and still others

1269

Twin Wiki Wonders?

did not know since they had yet to experience

such situations:
So revert it :) Its a Wiki, so everybody can edit it.
When s.o.s edit doesnt apply to my standards, I
can revert it. And that person can revert me too.
When we both find it important, we can start talking through the talkpage, and in the end, maybe
get a mediator, or ask the community what they
think about it. (Participant 1)
This happens frequently. I usually either talk to
them via the Talk pages or send them a private
message regarding the change(s) if they are not an
anonymous user. I leave or qualify their changes
if they are correct, but if not, I may revert them
back to their original form. (Participant 3)
Perhaps such quotes reflect the degree of expertise of the Wikibookian. Byrant, Forte, and
Bruckman, (2005), for example, found that expert
Wikibookians have a policy not to revert changes
or additions in Wikipedia if at all possible. They
tend to post their concerns to the talk page first.
From the nature of the collaboration process,
we see the complexity of not only the communication process among authors, but a philosophical question about the definition of a finished
Wikibook emerged. As an emerging phenomena,
Wikibooks draw interest and attention, especially
for their education and self-learning value.

Completeness, Success, and

Happiness Regarding Ones
Wikibook Project
Wikibook Completion
Given the comments and concerns about the quality of Wikibooks and length of time to complete
them, it was important to ask Wikibook contributors about Wikibook completion and overall
successes experienced. Using all the methods

1270

mentioned here, will the revision process ever

be completed? In other words, will a Wikibook
ever be finished? We raised this question to all
of our participants and had rather interesting
findings. In our survey responses, nearly 60% of
Wikibookians indicated that a Wikibook could
be completed. However, as the following quotes
exemplify, respondents to our interview questions
were quite varied in their opinions regarding
online book completion:
It will probably depend on the subject. When you
will try to explain a limited subject, itcan theoretically be complete. However, it will be hard to have
it finally completed indeed. (Participant 1)
No wiki is ever complete, because it is ever evolving. Thats one of the best things about wikis. I
personally think that paper is dead and in many
ways the ideas contained within them too. I want
my ideas and thoughts evolved and allowing
others to improve them makes the work alive.
(Participant 2)
Wikibooks can eventually become complete,
when the authors decide that there is no more to
writeso a book is substantially complete once
the last proper author leaves....It is by dedicated writers that books are written, not casual
one- sentence contributors or people who drop
in for an afternoon. (Participant 4)
As demonstrated, some Wikibookians thought
that completion of the book would naturally be
indicated by participant interest in the topic or the
scope of the project. One Wikibookian explained
that I think a Wikibook becomes complete when
the participants loose interest in the topic. On the
other hand, another Wikibookian suggested that
a Wikibook could be complete if all the relevant
information on a topic was included or known
such as the Iran-Contra scandal of the early to
mid 1980s. Overall, however, there was some

Twin Wiki Wonders?

agreement that a Wikibook is not as a product but

a process, because a Wikibook is always evolving
and allowing others to improve them makes the
work alive. From this notion, a more refined question surfaced: who decides whether a Wikibook
is finished? Can we say a Wikibook is finished
when nobody is contributing to it anymore?

Wikibook Success and Happiness

According to our online survey, most Wikibookians felt that their Wikibook was a success
(see Figure 3). We did not, however, ask what
determines that success. This issue relates back
to the previous one regarding completion; if
Wikibookians focus on the process and not the
final product, then success is an extremely fleeting concept. At the very least, success is then
focused on social interaction and collaboration
processes, instead of simply on the quality and
use of the final document.
Some Wikibookians seem truly inspired by
the Wikibook process and working in any type
of wiki environments. For example, Participant

1 simply stated, Go rockin on! when asked for

additional comments, while Participant 2 noted,
I love wikis theyre truly the closest example of
the purest form democracy.

Advice to Wikibook Wannabes

Technology Skills and Realistic Goals
Obviously, a Wikibookian needs to have the
necessary technology skills to operate within a
wiki environment. Beyond technological savvy,
understanding the magnitude of creating a book
and setting realistic goals could help start one on
the right foot:
A little bit is better than nothing, it can always be
improved later. Often times there is no book for
a topic because everyone thinks Oh, someone
else will make it. (Participant 6)
Get help. Dont try to do it on your own, its a too
big amount of work and you will definitely loose
the overview. (Participant 1)

Figure 3. Wikibookians perceptions of Wikibook success

My recent wikibook project was successful

70%

60.56%

60%
50%
40%
30%

19.72%

20%
10%

15.49%

4.23%

0%

Strong Disagree

Disagree

Agree

Strong Agree

1271

Twin Wiki Wonders?

Wikibooks are Demanding

As the following quote from Participant 5 indicates, experienced Wikibookians also warned us
about the difficulty of chapter organization:
Id suggest getting several coauthors from the get
go and deciding on a template for the book chapters
so it is uniform from the beginning. Its bound to
change over time, but you may as well start with
a plan. Also, Id mention that, for the most part,
they shouldnt expect many people to contribute
to the book. Writing a book is challenging; most
people will never attempt it. (Participant 5)
In addition to getting the flow and direction of
the book, Participant 4 advised us the importance
of understanding the book-writing process. In this
opinion, joining an existing book might be the best
way to become involved in Wikibooks:
I would first offer a warning that writing can be
very difficult. Writing a textbook is an involved
task, demanding the full scope of your expertise,
and requiring a substantial investment before
any payoff can be realized. Yet its extremely rewarding......I would encourage users to consider
working on an existing book rather than starting
a new one. Starting off with a more manageable
goal will let a user test the water without making
a big commitment. (Participant 4)

Wikibooks are Different from Wikipedia

Most Wikibookians seem to have initial wiki
backgrounds from editing or contributing to
Wikipedia. They may have shifted to Wikibooks
on the advice of others that their textbook-related
ideas belonged in Wikibooks not in Wikipedia.
Still, their prior history in Wikipedia shaped
their expectations. Among all the suggestions,
Participant 5 brought up the idea that involvement
with a Wikibook is different from working on a
piece of an article in Wikipedia:

1272

I think authors have to make concepts before they

write a book and think about it, because its not
only an article like in Wikipedia. (Participant 5)
As a result of this difference, in order to work
in a Wikibook, more communication, organization, and compromise is required. Participant 3
shared with us his view about how Wikibooks are
different from other collaborative productions:
I expected WIKIBOOKS to be more in-depth
about specific subject areas and to provide more
details than Wikipedia articles. In the case of
video games, for example, I expected WikiBooks
to provide background stories and walkthroughs.
(Participant 3)
Future studies might explore the need for
additional planning processing and tools within
Wikibooks. Would outlining and planning tools,
for instance, enhance the final product and the
percentage of Wikibooks that are, in fact, completed? These and other differences between
Wikipedia and Wikibooks are explored later on
in Table 2.

The Future of Wikibooks

We were interested in how these experienced
Wikibookians viewed the future of their Wikibook
work. Based on their experience, did Wikibooks
even have a future? And what tools and features
might enhance it? As Participant 6 rightfully
pointed out, the spirit of a wiki aims to make
the world a juster place by putting the tools of
education into the hands of every person.

Future Growth and Success of

Wikibooks
Some seemed concerned about the lack of reader
interests at this early stage of Wikibooks:

Twin Wiki Wonders?

Its difficult to tell at this stage. Wikibooks is faced

with a sort of chicken and egg problem -- it
needs readers to attract writers and writers to
attract readers. The project will recruit more and
more talented writers as it picks up steam, and
will have better and better books in its library.
But Wikibooks may never take off at all. (Participant 4)
The so-called chicken and egg issue was
shared by Participant 1, who believed Wikibook
will be either successful or die. As he stated,
There is no some people will use it. Because
when you want your book to become used, it has
to be used by a large amount of people, not by a
few. He also felt that a single dedicated person
could make a huge difference in determining the
ultimate success or failure of Wikibooks. Now
that is some incentive to participate!
Then the next logical question arises. Why do
few people use the books at the Wikibook Web
site? Is it because there has not been an example
yet of a book that goes all the way through and
is put into use or perhaps it is because People
have not seen it work enough for them to believe
in it, and there are not mechanisms in place that
convince educators of the value of Wikibooks as
a teaching or learning tool?

Academic Acceptance
In their responses to our online survey questions,
Wikibookians seemed to strongly believe in Wikibooks as an online library (64%), a learning tool
(40%), and a supplement to classroom or training
resources (36%). In addition, it was a place for
communities of writers (60%) and learners (34%)
to form. However, making Wikibooks a teaching
and learning tool and enabling the educational
community to embrace it proved a difficult task.
Academics might be the largest potential users of
Wikibooks; they might hold the key to Wikibooks
success. The following quotes indicate the shared
concerns and hopes of Participant 6 and 7:

I dont think the concept will catch on except

among small niches until there are some very
well-developed textbooks available on the site.
If academics turn to using it regularly for free
introductory textbooks, Wikibooks could explode
with interest. But we arent there yet. Maybe in
the next 5 years or so. (Participant 7)
This is a very valuable project with very important
goals, and any educator who is able to contribute
a bit or to draw his or her students or colleagues
into the process will be doing groundbreaking
work in education which is worthy of praise.
Things can only continue to go up from here.
(Participant 6)

Future trends
Each of the previously mentioned themes provides
interesting data for further wiki-related research
and development; especially that related to Wikibooks. Table 1 summarizes our major findings
according to the five themes.

Comparing Wikipedia and

Wikibooks
As indicated in the literature review, there are
many distinct differences between Wikipedia
and Wikibooks. As shown in Table 2, they differ
in terms of the types of communities that they
foster, the resources created, size of user community, technology supports required, the goals
and needs of both experts and novices, and many
other variables.

Future research directions

After their birth in 1995, wikis proliferated quickly
into the mainstream of human life. From users

1273

Twin Wiki Wonders?

Table 1. Key themes from interviews with Wikibookians

Theme 1: The Expression of Authority in Wikibooks
Authority: Each one of us possesses unique
knowledge that could contribute to the creation of
a Wikibook.
Control: Some liked to be able to monitor changes
and most believed in personal accountability.

Theme 2: Collaboration Process in Wikibooks

Coordination: Writing a Wikibook together might
be more of a social issue than a technical one.
Communication: Communicating among writers
proved essential.

not one person in charge

need at least two specialists, as both have
understand the subject very good
When a lone writer is trying to open peoples
minds to the reality of things and closed minded
indoctrinated individuals thatin all reality
could be a 13 yr old who has been given admin
powers can be judge and jury over your work..

Above all, a collaborative writing community

must have a common vision
requires communication in order to get
the project off the ground and into something
tangible
Collaboration: Finding someone else who is also
there have been isolated contributions, no
interested in your book might not be so easy.
one else has taken an interest in making a real
investment in the book
Resolving Disagreements: One would revert
I usually either talk to them via the Talk
the changes, some would discuss them first, and
pages or send them a private message regarding
still others did not know since they had yet to
the change(s) if they are not an anonymous
experience such situations.
user.
Theme 3: Completion, Success, and Happiness Regarding Ones Wikibooks
Completion: Agreement seemed to be that a
No wiki is ever complete, because it is ever
Wikibook is not a product but a process because a
evolving.
Wikibook is always evolving.
Wikibooks can eventually become complete,
when the authors decide that there is no more
to write
Success and Happiness: Success is then focused
Go rockin on!
on social interaction and collaboration processes.
I love wikis theyre truly the closest example
of the purest form democracy.
Theme 4: Advice to Wikibook Wannabes
Technical Skills and Realistic Goals: Be realistic Dont try to do it on your own, its a too big
regarding your skill level and writing goals.
amount of work and you will definitely loose
the overview.
Wikibooks are Demanding: Writing a book
Writing a textbook is an involved task,
chapter is a difficult task.
demanding the full scope of your expertise, and
requiring a substantial investment before any
payoff can be realized.
Wikibooks are Different from Wikipedia.
See Table 2 for comparison.
Theme 5: The Future of Wikibooks
Growth and Success: Lack of readership at early
it needs readers to attract writers and writers to
stage.
attract readers.
Academic Acceptance: Academics might be the
educator who is able to contribute a bit or to
largest potential users of Wikibooks; they might
draw his or her students or colleagues into the
hold the key to Wikibooks success.
process will be doing groundbreaking work in
education which is worthy of praise.

1274

Twin Wiki Wonders?

Table 2. Comparison of Wikipedia and Wikibook environments

Issue or Characteristic
1. Date Launched
2. Community

Wikipedia
January 15, 2001
Wikipedia is a community of
practice for millions of people.
There are myriad subcommunities
within it for different languages,
topics, resources, and tools.

3. Resources Created

Wikipedia is an information
resource for people to look up. It
is comprised of many linked, yet
individual pieces.
7,483,939 pages
6.4 million articles
1,629,257 articles in English
250 languages
110,836,256 edits
14.81 edits per page
700,001 media files
3,511,411 registered users
1,111 system administrators
(Wikipedia, 2007d).

4. Historical Statistics (as of

February 9, 2007):

5. Views of Contributors

6. Goals and Focus

7. Participation or Contribution
Criteria

One could think of the

Wikipedians as a global ant
colony, except that there is no
queen, and users choose their
own roles. Some Wikipedians
welcome newcomers, while
others determine awards, upload
images, clean article grammer,
work on article histories,
revert vandalism, and so forth.
(Wikipedia, 2007e).
Wikipedians are concerned with
the quality of Wikipedia. They
maintain or monitor a set of
watch pages (Bryant, Forte, &
Bruckman, 2006).
Enter Wikipedia site with
generalized knowledge but add
specific pieces of information.

Wikibook
July 10, 2003
Wikibooks has communities
of practice for each book
project. There is also an overall
community of Wikibookians
at the staff lounge within
Wikibooks.
Wikibooks creates usable texts,
guidebooks, and reference
materials. The final product
should be coherent.
67,399 pages
23,790 modules or chapters
Over 1,000 books, the largest
category in English
120 languages
759,033 page edits
11.26 edits per page
50,582 registered users,
33 system administrators
(Wikibooks, 2007b).
A Wikibookian is someone who
coordinates or contributes to a
Wikibook project.

Wikibookians are concerned with

the quality of the particular books
and study guides that they are
preparing.
Enter book project with
specialized knowledge.

1275

Twin Wiki Wonders?

Table 2. Comparison of Wikipedia and Wikibook environments (continued)

Issue or Characteristic
8. Technology Tools, Features,
and Resources

Wikipedia
Tools for tracking the history of
document changes, talk pages,
edit pages, hyperlinking, lounges,
site statistics, and so forth.

9. User Community

Massively millions.

10. Qualification and Participant

Filters
11. Novices

None, open to all to contribute or

edit a document.
Wikipedia is more of a collection
of articles with random people
adding information here
and there, then as a way to
communicate, collaborate, and
protect good work (Bryant, Forte,
& Bruckman, 2005).
Wikipedia is a community of
coauthors who use their talents to
help establish and maintain the
quality of Wikipedia; Wikipedia
is more important than any
single article or set of articles
or resources (Bryant, Forte, &
Bruckman, 2005).
Many to many.

12. Experts

13. Generation, Distribution, and

Maintenance of Ideas
14. Speed of Content
Development

15. Ownership

1276

Tends to be fast. Assisted

by a huge army of volunteer
contributors as well as by news
and media features or stories.
Tends to be fast.
Free. No one owns.

Wikibook
Same tools as Wikipedia but
could also use book planning,
outlining, and overview tools,
enhanced discussion tools,
mark-up and commenting tools,
enhanced tracking of book
contributions and contributors,
and ways of fostering
collaboration and interaction
among those in a Wikibook
project.
Extensive yet far more limited
membership than Wikipedia
None, open to all to contribute or
edit a document.
Wikibooks is a collection
of books to which one can
contribute to or read.

Wikibooks is a place wherein one

can coordinate a book project that
can contribute to the movement
of making open knowledge more
accessible.

One to many as well as many to

many.
Tends to be relatively slow.
Book projects take longer than
Wikibookians often expect; many
incomplete or unfinished books
and resources (Sajjapanroj et al.,
2007).
Free. No one owns.

Twin Wiki Wonders?

perspectives, wikis offer common knowledge

resources, whether they be dictionaries, encyclopedias, books, or sets of morals, fables, and
quotes. At a basic knowledge level, then, wikis
provide community-driven access to information resources when needed. From a contributing
writers viewpoint, on the other hand, wikis are
a penultimate example of collaboratively written and socially negotiated texts. Contributions
in a wiki can be small pieces of information or
media for a presentation, train schedules for a
trip to the UK, places to eat at the riverwalk in
San Antonio during a conference, and synonyms
for a word being defined in Wiktionary. At the
same time, they might also be entire chapters or
modules of a book or an entire book as in the case
of Wikibooks.
As the scope of wiki-work increases from
an informational focus at the word, sentence, or
paragraph level, as is often the case in Wikipedia
or Wikiquotes, to something much more substantive, as in the case of Wikibooks or other online
book sites, the problems, issues, and challenges
also significantly change. The same wiki tools,
approaches, and procedures that work so effectively for Wikipedians may not to be sufficient for
Wikibookians. In addition, the level of expertise
required to help write a book is at a higher level
of specificity and depth than most arriving at a
wikibook type of site would likely have. At the
same time, the communities of practice that form
around such a book have to negotiate issues of
chapter coherence, book organization, and topical
relevance that those in other types of wiki-related
sites do not typically need to consider. Such issues
place serious challenges as well as opportunities in
the face of wiki researchers as well as designers of
new generations of wiki-related technologies.

Conclusion
As we pointed out in this chapter, wikis offer a
unique window into knowledge negotiation and

collaboration processes. The coming decade will

undoubtedly witness new tools and resources to
track and better understand such negotiation and
interaction processes. Now is the time to begin
organizing a set of research initiatives, agendas,
and questions that will allow for responsive research that helps in the formation of new tools, as
well as enhances understanding of the power of
wikis. Wikpedia and Wikibooks and many other
wiki sites are wondrous tools and resources for
twenty-first century working, learning, and living. With these wiki wonders, YOU can now all
participate in the human race in new and exciting
ways! Engage.

References
Alexa. (2007). Results for Wikipedia. Retrieved
February 7, 2007, from http://www.alexa.com/
search?q=wikipedia
Brown, J. S. (2006). Relearning learningApplying the long tail to learning. Presentation at
MIT iCampus. Retrieved February 9, 2007, from
http://www.mitworld.mit.edu/video/419
Bruns, A., & Humphreys, S. (2005). Wikis in
teaching and assessment: The M/Cyclopedia
project. Paper presented at the WikiSym 2005.
Retrieved February 5, 2007, from http://www.
wikisym.org/ws2005/proceedings/paper-03.pdf
Bryant, S. L., Forte, A., & Bruckman, A. (2005).
Becoming Wikipedian: Transformation of participation in a collaborative online encyclopedia. In M. Pendergast, K. Schmidt, G. Mark, &
M. Acherman (Eds.), Proceedings of the 2005
International ACM SIGGROUP Conference on
Supporting Group Work, GROUP 2005, Sanibel
Island, FL, November 6-9 (pp. 1-10). Retrieved
February 7, 2007, from http://www-static.
cc.gatech.edu/~aforte/BryantForteBruckBecomingWikipedian.pdf

1277

Twin Wiki Wonders?

Campus Technology (2006a, October 10). News

Update: MIT launches center for Collective (Wiki)
intelligence. Campus Technology. Retrieved
February 2, 2007, from http://campustechnology.
com/news_article.asp?id=19384&typeid=150

Hall, B. (2006). Five innovation technologies.

Chief Learning Officer, 5(7), 13. Retrieved
February 2, 2007, from http://www.clomedia.
com/content/templates/clo_article.asp?articleid
=1443&zoneid=190

Campus Technology (2006b, October 10). News

update: Stanford debuts Wiki of all things Stanford. Campus Technology. Retrieved February
2, 2007, from http://campustechnology.com/
news_article.asp?id=19384&typeid=150

Honegger, B. D. (2005). Wikis A rapidly growing phenomenon in the German-speaking school

community. Paper presented at the WikiSym 2005.
Retrieved February 5, 2007, from http://www.
wikisym.org/ws2005/proceedings/paper-10.pdf

Campus Technology (2007, January 30). News

update: MIT, Wharton to publish collaborative
textbook by Wiki. Campus Technology. Retrieved
February 2, 2007, from http://campustechnology.
com/news_article.asp?id=20096&typeid=150

Korfiatis, N., Poulos, M., & Bokos, G. (2006).

Evaluating authoritative sources using social
networks: An insight from wikipedia. Online
Information Review, 30(3). Retrieved February
4, 2007, from http://www.korfiatis.info/papers/
OISJournal_final.pdf

Ciffolilli, A. (2003). Phantom authority, selfselective recruitment and retention of members

in virtual communities: The case of Wikipedia
[Electronic Version]. First Monday, 8(12). Retrieved February 10, 2007, from http://firstmonday.
org/issues/issue8_12/ciffolilli/index.html
Emigh, W., & Herring, S. C. (2005). Collaborative
authoring on the Web: A genre analysis of online
encyclopedias. Proceedings of the 38th Hawaii
International Conference on System Sciences
(HICSS-38). Los Alamitos: IEEE Press. Retrieved
February 3, 2007, from http://ella.slis.indiana.
edu/~herring/wiki.pdf
Evans, P. (2006, January/February). The Wiki
factor. BizEd, 28-32.
Friedman, T. L. (2005). The world is flat: A brief
history of the twenty-first century. NY: Farrar,
Straus, and Giroux.
Grossman, L. (2006/2007a). You: Time Magazine
Person of the Year. Time Magazine, 168(26), 3841. December 25, 2006/January 1, 2007.
Grossman, L. (2006/2007b). Power to the people.
Time Magazine Person of the Year, 168(26), 4247, 50, 53-56, 58.

1278

Lessig, L. (2005). Creatives face a closed net.

Financial Times. Retrieved February 14, 2007,
from http://www.ft.com/cms/s/d55dfe52-77d211da-9670-0000779e2340.html
Leuf, B., & Cunningham, W. (2001). The Wiki
way: Quick collaboration on the Web. Addison
Wesley Longman.
Lih, A. (2004). Wikipedia as participatory journalism: Reliable sources? Metrics for evaluating
collaborative media as a news resource. Paper
presented at the 5th International Symposium on
Online Journalism, April 16-17, UT Austin.
Lin, M. (2006). Sharing knowledge and building community: A narrative of the formation,
development, and sustainability of OOPS. Unpublished dissertation, University of Houston,
Houston, TX.
Lio, E. D., Fraboni, L., & Leo, T. (2005). TWikibased facilitation in a newly formed academic
community of practice. Paper presented at the
WikiSym 2005. Retrieved February 5, 2007,
from http://www.wikisym.org/ws2005/proceedings/paper-09.pdf

Twin Wiki Wonders?

Pfeil, U., Zaphiris, P., & Ang, C. S. (2006).

Cultural differences in collaborative authoring of
Wikipedia. Journal of Computer-Mediated Communication, 12(1), article 5. Retrieved February
7, 2007, from http://jcmc.indiana.edu/vol12/issue1/pfeil.html
Reuters (2007). Publisher launches its first wiki
novel. Yahoo News. Retrieved February 5, 2007,
from http://news.yahoo.com/s/nm/20070201/
tc_nm/penguin_wiki_dc_4
Rosenzweig, R. (2006). Can history be open
source: Wikipedia and the future of the past. The
Journal of American History, 93(1), 117-146. Retrieved February 4, 2007, from http://chnm.gmu.
edu/resources/essays/d/42
Sajjapanroj, S., Bonk, C. J., Lee, M., & Lin,
M.-F. G. (2007). The challenges and successes
of Wikibookian experts and Wikibook novices:
Classroom and community perspectives. Paper
presented at the American Educational Research
Association, Chicago, IL.
Schachaf, P., & Hara, N. (2006). Beyond vandalism: Trolls in Wikipedia. Presentation at the Rob
Kling Center for Social Informatics, School of
Library and Information Science, Indiana University, Bloomington, IN.
Time Magazine. (2006/2007). Time Magazine
Person of the Year, 168(26), December 25, 2006/
January 1, 2007.
Vigas, F. B., Wattenberg, M., & Dave, K.
(2004).
Studying cooperation and conflict between authors with history flow visualizations.
In E. Dykstra-Erickson & M. Tscheligi (Eds.),
Proceedings from ACM CHI 2004 Conference
on Human Factors in Computing Systems (pp.
575-582). Vienna, Austria. Retrieved February
3, 2007, from http://web.media.mit.edu/~fviegas/
papers/history_flow.pdf

for Scientometrics and Informetrics. Retrieved

February 10, 2007 from http://eprints.rclis.org/
archive/00003610/01/MeasuringWikipedia2005.
pdf
Wikibooks (2007a). Wikibooks: History of
Wikibooks. Retrieved February 9, 2007, from
http://en.wikibooks.org/wiki/Wikibooks:History_of_Wikibooks
Wikibooks (2007b). Wikibooks: Statistics. Retrieved February 9, 2007, from http://en.wikibooks.
org/wiki/Special:Statistics
Wikibooks (2007c). Wikibooks: Wikijunior.
Retrieved February 16, 2007, from http://
en.wikibooks.org/wiki/Wikijunior
Wikipedia (2007a). Wikipedia: Larry Sanger. Retrieved February 7, 2007, from http://en.wikipedia.
org/wiki/Larry_Sanger
Wikipedia (2007b). Wikipedia: List of Wikipedias.
Retrieved February 7, 2007, from http://meta.wikimedia.org/wiki/List_of_Wikipedias#All_Wikipedias_ordered_by_number_of_articles
Wikipedia (2007c). Wikipedia: Main page. Retrieved February 7, 2007, from http://en.wikipedia.
org/wiki/Main_Page
Wikipedia (2007d). Wikipedia: Statistics. Retrieved February 9, 2007, from http://en.wikipedia.
org/wiki/Special:Statistics
Wikipedia (2007e). Wikipedia: Wikipedians. Retrieved February 4, 2007, from http://en.wikipedia.
org/wiki/Wikipedia:Wikipedians
Wikiquotes (2007). Wikiquotes: Main page.
Retrieved February 16, 2007, from http://
en.wikiquote.org/wiki/Main_Page
Wiktionary (2007). Wiktionary: Main page.
Retrieved February 16, 2007, from http://
en.wiktionary.org/wiki/Main_Page

Voss, J. (2005). Measuring wikipedia [Electronic

Version]. Proceedings for International Society

1279

Twin Wiki Wonders?

Additional reading
Bargh, J. A., & McKenna, K. Y. A. (2004). The
Internet and social life. Annual Review of Psychology, 55, 573-590.
Bezroukov, N. (1999b). A second look at the cathedral and the bazaar [Electronic Version]. First
Monday, 4(12). Retrieved March, 15, 2007, from
http://firstmonday.org/issues/issue4_12/bezroukov/index.html.
Bonk, C. J. (2001). Online teaching in an online
world. Bloomington, IN: CourseShare.com.
Bonk, C. J., & Cunningham, D. J. (1998). Searching for learner-centered, constructivist, and sociocultural components of collaborative educational
learning tools. In C. J. Bonk & K. S. Kim (Eds.),
Electronic collaborators: Learner-centered
technologies for literacy, apprenticeship, and
discourse (pp. 25-50). NJ: Erlbaum.
Bonk, C. J., & Kim, K. A. (1998). Extending
sociocultural theory to adult learning. In M. C.
Smith & T. Pourchot (Ed.), Adult learning and
development: Perspectives from educational
psychology (pp. 67-88). Lawrence Erlbaum Associates.
Bonk, C. J., Wisher, R. A., & Nigrelli, M. L.
(2004). Learning communities, communities of
practice: Principles, technologies, and examples.
In K. Littleton, D. Miell & D. Faulkner (Eds.),
Learning to collaborate, collaborating to learn
(pp. 199-219). Hauppauge, NY: Nova Science
Publishers.
Brown, J. S. (2006, December 1). Relearning
learningApplying the long tail to learning.
Presentation at MIT iCampus. Retrieved February 9, 2007, from http://www.mitworld.mit.
edu/video/419
Cole, M., & Engestrom, Y. (1997). A culturalhistorical approach to distributed cognition. In G.
Salomon (Ed.), Distributed cognitions: Psycho-

1280

logical and educational considerations (pp. 1-46).

Cambridge, UK: Cambridge University Press.
Coles, R. (1993). The call of service: A witness
to idealism. Mariner Books.
Desilets, A., Paquet, S., & Vinson, N. G. (2005).
Are Wikis usable? Paper presented at the WikiSym 2005. Retrieved February 26, 2006, from
http://www.wikisym.org/ws2005/proceedings/
paper-01.pdf
Evans, P. (2006). The Wiki factor. BizEd. Retrieved
April 1, 2006, from http://www.aacsb.edu/publications/Archives/JanFeb06/p28-33.pdf
Fichter, D. (2006). Using Wikis to support online
collaboration in libraries. Information Outlook,
10(1), 30-31.
Garlick, M. (2005). A review of Creative Commons and Science Commons. EDUCAUSE Review, 40(5), 78-79.
Giles, J. (2005). Internet encyclopaedias go head
to head [Electronic Version]. Nature, 438, 900901. Retrieved December 15, 2005 from http://
www.nature.com/nature/journal/v438/n7070/
full/438900a.html
Hayes, J. R., & Flower, L. S. (1986). Writing
research and the writer. American Psychologist,
41(10), 1106-1113.
Howe, J. (2006/2007). Your Web, your way. Time
Magazine Person of the Year. Time Magazine
Person of the Year, 168(26), 60-61. Lamb, B.
(2004). Wide open spaces: Wikis, ready or not.
EDUCAUSE Review, 39(5), 36-48. Retrieved
February 26, 2006, from http://www.educause.
edu/ir/library/pdf/erm0452.pdf
Langer, J. A., & Applebee, A. A. (1987). How writing shapes thinking. Urbana, IL: National Council
of Teachers of English.
Lave, J., & Wenger, E. (1991). Situated learning:
Legitimate peripheral participation. Cambridge,
UK: Cambridge University Press.

Twin Wiki Wonders?

Lin, M., & Lee, M. (2006).E-learning localized:

The case of the OOPS project. In A. Edmundson
(Ed.), Globalization in education: Improving
education quality through cross-cultural dialogue
(pp. 168-186). Hershey, PA: Idea Group.
Mann, C., & Stewart, F. (2000a). Internet communication and qualitative research: A handbook
for researching online. Thousand Oaks, CA: Sage
Publication.
Matsuhashi, A. (Ed.). (1987). Writing in real time:
Modelling production processes. Norwood, NJ:
Ablex Publishing Corporation.
McHenry, R. (2004). The faith-based encyclopedia. Retrieved February 10, 2005, from http://tcsdaily.com/article.aspx?id=111504A
McKierman, G. (2005). WikimediaWorlds Part I:
Wikipedia. Library Hi Tech News, 22(8), 46-54.
Neuwirth, C., Kaufer, D., Chimera, R., & Gillespie,
T. (1987). The notes program: A hypertext application for writing from source texts. Paper presented
at Hypertext 87, Chapel Hill, NC.
Orlowski, A. (2005). Wikipedia founder admits
to serious quality problems [Electronic Version].
The Register. Retrieved February 10, 2006 from
http://www.theregister.co.uk/2005/10/18/wikipedia_quality_problem/
Palloff, R., & Pratt, K. (2005). Collaborating
online: Learning together in community. San
Francisco, CA: Jossey-Bass.
Pea, R. D. (1996). Practices of distributed intelligence and designs for education. In G. Salomon
(Ed.), Distributed cognitions: Psychological and
educational considerations (pp. 47-87). New
York: Cambridge University Press.
Raymond, E. S. (2000). The cathedral and the
bazaar. Retrieved March 10, 2007, from http://
www.catb.org/~esr/writings/cathedral-bazaar/
cathedral-bazaar/

Raymond, E. S. (2001). The cathedral and the

bazaar: Musings on Linux and open source by
an accidental revolutionary. Cambridge, MA:
OReilly.
Rheingold, H. (2000). The virtual community:
Homesteading on the electronic frontier. Cambridge, MA: MIT Press.
Rogoff, B. (1990). Apprenticeship in thinking:
Cognitive development in social context. New
York, NY: Oxford University Press.
Salomon (Ed.). (1993). Distributed cognitions:
Psychological and educational considerations
(pp. 4787). New York: Cambridge University
Press.
Salomon, G. (1988). AI in reverse: Computer tools
that turn cognitive. Journal of Educational Computing Research, 4(2), 123-139.
Sanger, L. (2004). Why Wikipedia must jettison its
anti-elitism. Retrieved February 10, 2006, from
http://kuro5hin.org/story/2004/12/30/142458/25
Scardamalia, M., & Bereiter, C. (1986). Research
on written composition. In M. C. Wittrock (Ed.),
Handbook of research on teaching. (3rd ed.), (pp.
778-803). New York: Macmillan Education Ltd.
Schrage, M. (1990). Shared minds: The technologies of collaboration. New York: Random
House.
Souzis, A. (2005). Building a semantic Wiki.
IEEE Intelligent Systems, 20(5), 87-91.
Utne, L. (2006). To wikifinity and beyond!
[Electronic Version]. Utne Magazine. Retrieved
April 29, 2006 from http://www.utne.com/issues/2006_133/view/11894-1.html
Wagner, C., & Bolloju, N. (2005). Supporting
knowledge management in organization with
conversational technologies: Discussion forums,
Weblogs, and wikis. Journal of Database Management, 16(2), i-viii.

1281

Twin Wiki Wonders?

Wenger, E. (1998). Communities of practice:

Learning, meaning, and identity. Cambridge
University Press.

Wertsch, J. V. (1991). Voices of the mind: A sociocultural approach to mediated action. Cambridge,
MA: Harvard University Press.

Wenger, E. (2002). Cultivating communities of

practice: A field guide to managing. Harvard.

YouTube. (2007). Web 2.0The machine is using us.

YouTube. Retrieved February 9, 2007, from http://
www.youtube.com/watch?v=6gmP4nk0EOE

This work was previously published in Social Information Technology: Connecting Society and Cultural Issues, edited by T.
Kidd and I.L. Chen, pp. 253-272, copyright 2008 by Information Science Reference, formerly known as Idea Group Reference
(an imprint of IGI Global).

1282

1283

Chapter 3.18

Wikis as Tools for Collaboration

Jane Klobas
Bocconi University, Italy, & University of Western Australia, Australia

Introduction
Tim Berners-Lee, the inventor of the World Wide
Web, envisioned it as a place where people can
communicate by sharing their knowledge in
a pool putting their ideas in, as well as taking
them out (Berners-Lee, 1999). For much of its first
decade, the Web was, however, primarily a place
where the majority of people took ideas out rather
than putting them in. This has changed. Many
social software services now exist on the Web
to facilitate social interaction, collaboration and
information exchange. This article introduces wikis, jointly edited Web sites and Intranet resources
that are accessed through web browsers. After a
brief overview of wiki history, we explain wiki
technology and philosophy, provide an overview
of how wikis are being used for collaboration,
and consider some of the issues associated with
management of wikis before considering the
future of wikis.
In 1995, an American computer programmer,
Ward Cunningham, developed some software to
help colleagues quickly and easily share computer
programming patterns across the Web. He called
the software WikiWikiWeb, after the Wiki Wiki

shuttle bus service at Honolulu International

Airport (Cunningham, 2003). As interest in wikis
increased, other programmers developed wiki
software, most of it (like WikiWikiWeb) open
source. Although wiki software was relatively
simple by industry standards, some technical
knowledge was required to install, maintain and
extend the wiki engines. Contributors needed
to learn and use a markup language to edit pages,
and even if the markup languages were often
simpler than HTML, non-technical users did not
find these early wikis compelling.
In the early years of the twenty-first century,
a number of developments led to more widespread use of wikis. Wiki technology became
simpler to install and use, open source software
was improved, and commercial enterprise-grade
wiki software was released. The not insignificant
issues associated with attracting and managing
a community of people who use a wiki to share
their knowledge were discussed in forums such
as MeatballWiki (http://www.usemod.com/cgibin/mb.pl?action=browse&id=MeatballWiki&ol
did=FrontPage). The publics attention was drawn
to wikis following the launch, in January 2001,
of the publicly written Web-based encyclopedia,

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Wikis as Tools for Collaboration

Wikipedia (www.wikipedia.org). And wiki hosting services and application service providers
(ASPs) were established to enable individuals and
organizations to develop wikis without the need to
install and maintain wiki software themselves.
By July 2006, nearly 3,000 wikis were indexed at the wiki indexing site www.wikiindex.
org, popular wiki hosting services such as Wikia
(www.wikia.org) and seedwiki (www.seedwiki.
org) hosted thousands of wikis between them, and
Wikipedia had more than four and a half million
pages in over 100 languages. Moreover, wikis
were increasingly being used in less public ways,
to support and enable collaboration in institutions
ranging from businesses to the public service and
not-for-profit organizations.

THE NATURE OF WIKIS

Wiki software allows users to collaboratively edit
pages for the Web or intranets. The pages created
with wiki software are called wiki pages and
sites that contain wiki pages are called wiki sites,
or simply wikis.
Technically, wikis consist of four basic elements:

Content
A template which defines the layout of the
wiki pages
Wiki engine, the software that handles all
the business logic of the wiki

Figure 1. How wikis work (Adapted from Klobas & Marlia, 2006)

Template

Wiki
engine

Wiki page

Wiki engine draws on template to display content in browser

Edits made by the user are stored in the content database

1284

Content

Wikis as Tools for Collaboration

Wiki page, the page that is created by the

wiki engine as it displays the content in a
browser

Figure 1 shows how these elements work together.

Wikis consist of pages accessible from a Web
browser. They are edited by opening an editing
screen in the browser and using either a simple
markup language or, increasingly, a rich text
editor to edit the text. Links to pages internal or
external to the wiki site can be added using simple
conventions. These conventions allow a link to be
created to a page that does not yet exist; the wiki
engine flags such links for future editing. Unless
the wiki managers decide otherwise, the content
is updated in real time, and once an author saves
and closes their changes on the editing screen, the
changes are immediately visible online.
Almost all wikis keep track of changes. Older
versions of a page can be viewed and, if necessary,
restored. Most wikis include a page where recent
changes are listed. This feature helps members
of the wiki community to keep up to date with
changes in content, and can help newcomers get
a quick feel for the current concerns of the wiki
community. Increasingly, wiki software is integrated with news aggregators like RSS or e-mail
notification to alert users to changes without their
having to enter the wiki itself.
Another important feature of wikis is the
simple permissions structure. Typically, there
are three levels of permission: reader, editor, and
administrator. Reading permissions may be open
to anyone on the World Wide Web or limited to
specific, registered individuals. When reading
permissions are limited, the wiki is known as a
private wiki.
Various wikis offer support for non-Latin
character sets, different media and file types,
mathematical notation, style sheets, conflict handling, spam handling, and facilities for merging,
exporting and backing up. The different features
available in different wiki engines can be seen

at the wiki engine comparison site, WikiMatrix

(www.wikimatrix.org).
Wiki features are based on design principles
established by Ward Cunningham. These principles address human as well as technical goals,
for example (in the terms used by Wagner, 2004),
wikis are:

Open: If any page is found to be incomplete

or poorly organized, any reader can edit it
as he/she sees fit
Organic: The structure and content of the
site evolve over time
Universal: Any writer is automatically an
editor and organizer
Observable: Activity within the site can be
watched and revised by any visitor to the
site
Tolerant: Interpretable (even if undesirable)
behavior is preferred to error messages

Wikis usually adopt soft security, social conventions that assume that most people behave in
good faith, establish that users (rather than the
software or a system administrator) operate as
peer reviewers of content and behavior, allow that
people might make mistakes but that mistakes
can be corrected, and emphasize the importance
of transparency in their management (Meatball,
2006). Together, these technical features and social
principles provide a supportive environment for
human collaboration.

how wikis are being used for

collaboration
Wikis are used by groups of people who collaborate to produce information resources that
range from meeting agendas to Web sites. They
are used in business, government, research, and
education.
Public wikis are often built by communities
of practice, hobbyists and other interest groups.

1285

Wikis as Tools for Collaboration

Most of these wikis are concerned with quite

specific topics such as a specific sport, book,
author, religion, or philosophy. Often, they are
maintained and read by small groups of friends
and colleagues. An example is ukcider, a wiki that
provides information about real cider (http://ukcider.co.uk/wiki) and supports advocacy for real
cider enthusiasts and small producers.
Wikis can be used to create and maintain
knowledge repositories for communities of
practice (Godwin-Jones, 2003; Roberts, 2005).
Members of the community update the repository
with information and solutions to problems. As
with other types of wiki, the communities served
might be public or private. Potential uses of private
wikis to increase autonomy among US intelligence
workers were described by Andrus (2005).
Committees and working groups, particularly
those that work across institutional and/or geographical boundaries, use wikis to develop and
maintain agendas and documentation. The DCMI
Education Working Group Wiki (http://dublincore.org/educationwiki), for example, contains
work plans, FAQs, agendas, notes, documentation,
and links to key references for the group.
A common organizational use of wikis is
project support (Bean & Hott, 2005). Wikis are
used to support the sharing of agendas, ideas,
resources, plans and schedules. Documents can be
jointly produced within the wiki or made available
as attachments. Angeles (2004) describes how
wikis have been used by Lucent Technologies for
project documentation including preparation of
meeting notes, product specification notes, product
requirements documents, project deliverables,
content audits, technical documentation, and
style guides. Other companies that use wikis for
project support include Michelin China and the
investment bank, DrKW (Paquet, 2006).
Wikis are also being used to write and maintain system documentation. A public wiki has
been developed by Mozilla, the provider of the
Firefox Web browser, to maintain Mozilla product

1286

documentation (http://kb.mozillazine.org). Any

member of the public can contribute. Some companies have established private wikis to maintain
documentation for their internal information
systems (IS). Users contribute to documentation
based on their experience with the system.
Organizations and groups can use wikis to
produce reports of news or events. Schools, in
particular, appear to be experimenting with this
type of wiki, in which parents, teachers, administrators, and students can all contribute reports,
photographs, audio and video of a school event.
Some groups use wikis to plan conferences or
meetings and to continue discussions that were
begun at a meeting. Examples include the American Library Association (ALA) 2006 Conference
wiki (http://meredith.wolfwater.com/ala2006/)
and the Yale Access to Knowledge wiki (http://research.yale.edu/isp/a2k/wiki/). Some experiments
in use of wikis during conferences have found
them valuable (Boyd, 2004; Suter, Alexander, &
Kaplan, 2005). Boyd describes how workshop
participants quickly learnt to use wikis to capture
their own streams of consciousness or the comments of others.
Wikis are used in a variety of ways for collaboration in classrooms ranging from primary
school to university (Bold, 2006; Ferris & Wilder,
2006; Lamb, 2004; Skiba, 2005). Mitchell (2006)
notes that wikis are consistent with modern
theories of learning such as connectivism
(Siemens, 2004), the idea that learning occurs in
a complex and changing environment that cannot
be controlled. In the classroom, wikis are used in
collaborative writing, student projects and group
assignments. Collaborative writing tasks can be
given to develop writing and collaboration skills
or to develop information resources for use by
other students.
Wikis are particularly well adapted to shared
creation of directories and lists such as bibliographies, staff lists and job vacancies. In schools
and universities, teachers who use wikis in this

Wikis as Tools for Collaboration

way assign groups of students or whole classes

the task of jointly preparing annotated lists of
information resources. Although most educational
wikis are private, some become public resources.
An example is the Wiki for IS Scholarship maintained by the Information Systems Department
in the Weatherhead School of Management at
Case Western Reserve University (http://isworld.
student.cwru.edu/tiki/tiki-index.php) which
contains summaries of major books and articles
in IS as well as reviews of the contributions of
major IS scholars. Often, wiki-based directories
also provide access to resources. Participants
in the Government Open Code Collaborative
(http://www.gocc.gov) use a public wiki to list and
describe source code that can be shared among
its members.
Wiki software is also being used to create Web
sites that draw on the contributions of a number of
people. One example is the site developed by the
Bach-Academie of Montreal to publicize a series of
concerts. The musicians themselves added details
to the wiki, which was then published to the Web
(http://www.bach-academie-de-montreal.com/).

issues in wiki management

The ease with which individuals can contribute
to wikis, the open nature of contribution, and the
philosophy that the guardians of the content are
the people who create the wiki can produce outstanding knowledge resources. On the other hand,
these very characteristics can also be associated
with problems such as reluctance to contribute or
chaos through unstructured contribution. Some
management is therefore required by the person
or group that wishes to develop and maintain
the wiki. The most significant management issues tend to be those of encouraging people to
contribute, and managing intellectual property,
problem users and spam.

As with any knowledge resource, people need

both a reason to contribute to a wiki and a sense
that they have the ability to do so. Wikis created
for small groups or communities with a specific
purpose should quite readily attract users if those
users also know how to access and use the wiki.
For example, a wiki established for the purpose of
jointly preparing an agenda or document, where
all members of the group responsible for preparing the document are motivated to contribute and
know how to access the wiki and use the editor,
is likely to be used by the intended users for the
intended purpose. The more diffuse the purpose
and the user group, the more uncertainty there
will be about the development and continuation
of a wiki. In practice, however we usually find a
core group of regular contributors and a peripheral
group of occasional contributors and people who
read the wiki but do not contribute to it. (This pattern is common in online communities; see, for
example, Wenger, McDermott, & Snyder, 2002).
The core users can act as role models for other
users. Other patterns that have been observed
include the champion pattern (single wiki-nut,
encourages coworkers to add, view, improve) and
the trellis pattern (egregiously boring content
calls for fixing My [contribution can be] far
more interesting than that! Confluence, 2006).
Techniques for ensuring that potential contributors feel able to contribute include providing
simple, easy to follow documentation for editing
pages and assurance that mistakes are permitted
and can quickly be remedied. Anti-patterns that
discourage use include the gate pattern, too
many procedural barriers to adding content
(Confluence, 2006).
Where the content of a wiki may include information, images and files drawn from other sources,
the intellectual property of the original sources
must be respected. Standard policies for dealing
with material that is the intellectual property of
others include: (1) only material that does not

1287

Wikis as Tools for Collaboration

violate the copyright of the original creator can be

included in the wiki and (2) the sources of all included material should be acknowledged. Equally,
a decision has to be made about ownership and
conditions for re-use of original material included
in a wiki. The fact that the content of any single
wiki page is usually produced by multiple authors
needs to be acknowledged. A typical statement of
the intellectual property in a wiki acknowledges
its joint authorship and permits re-use, provided
the wiki page is acknowledged as the source of
the material. The GNU Free Documentation
License used by Wikipedia provides an example
(Free Software Foundation, 2002).
The more open a wiki to the public, and the
larger the wiki user community, the greater the
potential of encountering a problem user. Problem
users may post unacknowledged copyrighted
material, post material designed to offend or
anger readers or contributors, remove material
or otherwise vandalize the wiki. Some wikis
use registration to limit the possibility that a
problem user will continue to damage the wiki,
but many wiki communities prefer to adopt the
soft security principles described earlier in this
article. It may be sufficient simply to let someone
know (privately, by e-mail) the first time they post
unacceptable content, that this type of content
is not appropriate for the wiki (Turnbull, 2004).
A wiki page that contains problem content can
easily be replaced by an earlier, clean version.
The active members of a wiki community often
share the task of watching pages and dealing
with problems.
Another potential problem, for public wikis in
particular, is spam. Pages can be watched for spam
in the same way that they are watched for other
problem content, but this can be a time-consuming
task. Some wikis use registration processes that
require human intervention, such as CAPTCHA
(Carnegie Mellon University School of Computer
Science, 2000-2005), to prevent mass spam attacks. Spam capturing services are being added

1288

to wiki engines to reduce the amount of spam

that reaches a wiki page.

Future trends AND

CONCLUSION
At the time of writing, wikis were still the domain of early adopters, rather than a part of the
mainstream. Despite widespread use of Wikipedia, knowledge of what wikis are and how they
can be used for collaboration, particularly for
private use, is not widespread. As of the end of
2006, only two non-technical books about wikis
had been published (Klobas, 2006; Tapscott &
Williams, 2006). Nonetheless, in October 2006,
Google Inc. bought the wiki software and Web
hosting service, JotSpot, an indication that consumer use of wikis is on the rise. The Gartner
Group envisages wikis reaching maturity for
business use some time before 2010 (Fenn &
Linden, 2005). As wiki software matures, we
can expect improvements in technical qualities
such as ease of implementation and stability, as
well as improvements in editing interfaces and
graphical quality. These improvements, along
with the positive experiences of early adopters,
should help gain the interest and confidence of
potential users and result in further diffusion of
wikis as tools for collaboration.

References
Andrus, D. C. (2005). The wiki and the blog: Toward a complex adaptive intelligence community.
Studies in Intelligence, 49(3). Retrieved October
26, 2006, from http://ssrn.com/abstract=755904
Angeles, M. (2004). Using a wiki for documentation and collaborative authoring. Retrieved
November 1, 2006, from http://www.llrx.com/
features/librarywikis.htm

Wikis as Tools for Collaboration

Bean, L., & Hott, D. D. (2005). Wiki: A speedy

new tool to manage projects. Journal of Corporate
Accounting and Finance, 16(5), 3-8.

Godwin-Jones. (2003). Blogs and wikis: Environments for on-line collaboration. Language,
Learning and Technology, 7(2), 12-16.

Berners-Lee, T. (1999). Transcript of Tim BernersLees talk to the LCS 35th anniversary celebrations, Cambridge, Massachusetts, 14 April 1999.
Retrieved October 26, 2006, from http://www.
w3.org/1999/04/13-tbl.html

Klobas, J. (2006). Wikis: Tools for information

use and collaboration. Oxford: Chandos Publishing.

Bold, M. (2006). Use of wikis in graduate course

work. Journal of Interactive Learning Research,
17(1), 5-14.
Boyd, S. (2004, February). Wicked (good) wikis.
Darwin Magazine. Retrieved November 1, 2006,
from http://www.darwinmag.com/read/020104/
boyd.html
Carnegie Mellon University School of Computer
Science. (2000-2005). The CAPTCHA Project.
Retrieved November 1, 2006, from http://www.
captcha.net/
Confluence. (2006). Patterns of wiki adoption.
Retrieved November 1, 2006, from http://confluence.atlassian.com/display/PAT/Patterns+of+Wi
ki+Adoption
Cunningham, W. (2003). Correspondence on the
etymology of wiki. Retrieved October 25, 2006,
from http://c2.com/doc/etymology.html
Fenn, J., & Linden, A. (2005). Gartners Hype
Cycle special report for 2005. Stamford, CT:
Gartner Group.
Ferris, S. P., & Wilder, H. (2006). Uses and
potentials of wikis in the classroom. Innovate, 2(5). Retrieved November 1, 2006,
from http://www.innovateonline.info/index.
php?view=article&id=2258
Free Software Foundation. (2002, November).
GNU free documentation license. Retrieved
November 1, 2006, from http://www.gnu.org/
copyleft/fdl.html

Klobas, J., & Marlia, M. (2006). Creating a wiki:

The technology options. In J. Klobas (Ed.), Wikis:
Tools for information work and collaboration (pp.
149-182). Oxford: Chandos Publishing.
Lamb, B. (2004). Wide open spaces: Wikis, ready
or not. EDUCAUSE Review, 39(5). Retrieved
October 26, 2006, from http://www.educause.
edu/ir/library/pdf/erm0452.pdf
Meatball. (2006, September). Soft Security. Retrieved October 26, 2006, from http://usemod.
com/cgi-bin/mb.pl?SoftSecurity
Mitchell, P. (2006). Wikis in education. In J.
Klobas (Ed.), Wikis: Tools for information work
and collaboration (pp. 119-147). Oxford: Chandos
Publishing.
Paquet, S. (2006). Wikis in business. In J. Klobas
(Ed.), Wikis: Tools for information work and
collaboration (pp. 99-117). Oxford: Chandos
Publishing.
Roberts, A. (2005). Introducing a Wiki to a community of practice. Retrieved January 9, 2006,
from http://www.frankieroberto.com/dad/ultrastudents/andyroberts/year2/AEreport/AEtool.
html
Siemens, G. (2004). Connectivism: A learning
theory for the digital age. Retrieved October 26,
2006, from http://www.elearnspace.org/Articles/
connectivism.htm
Skiba, D. J. (2005). Do your students wiki? Nursing
Education Perspectives, 26(2), 120-121.
Suter, V., Alexander, B., & Kaplan, P. (2005).
The future of FTF. EDUCAUSE Review, 40(1).

1289

Wikis as Tools for Collaboration

Retrieved November 1, 2006, from http://

www.educause.edu/apps/er/erm2005/erm0514.
asp?bhcp=2001

Soft Security: Social conventions for trust,

peer review and correction of errors adopted by
contributors to wikis.

Tapscott, D., & Williams, A. D. (2006). Wikinomics: How mass collaboration changes everything.
Portfolio Hardcover.

Wiki: An information resource that is created

by multiple authors who use Web browsers that
interacts with wiki software.

Turnbull, G. (2004). Talking to Ward Cunningham about wikis. Luvly, Retrieved November 1,
2006, from http://gorjuss.com/luvly/20040406wardcunningham.html

Wiki Engine: The software that handles the

business logic of a wiki.

Wagner, C. (2004). Wiki: A technology for conversational knowledge management and group
collaboration. Communications of the Association
for Information Systems, 13, 265-289.
Wenger, E., McDermott, R., & Snyder, W. M.
(2002). Cultivating communities of practice. Boston: Harvard Business School University Press.

Key Terms

Wiki Page: A page of wiki content displayed

in a Web browser.
Wiki Site: A set of related wiki pages. When a
wiki site can be viewed on the World Wide Web,
it is also a Web site.
Wiki Software: The suite of software used to
produce and manage a wiki. This software may
include, in addition to the wiki engine, add-ons
and extensions that extend the feature set and
functionality of the wiki.
WikiWikiWeb: The first wiki engine, written
by Ward Cunningham.

Editor: The authors of wiki pages may be

called editors because they have editing permissions.

This work was previously published in the Encyclopedia of E-Collaboration, edited by N. Kock, pp. 712-717, copyright 2008
by Information Science Reference, formerly known as Idea Group Reference (an imprint of IGI Global).

1290

1291

Chapter 3.19

Academic Weblogs as
Tools for E-Collaboration
Among Researchers
Mara Jos Luzn
University of Zaragoza, Spain

Introduction
Although scientific research has always been a
social activity, in recent years the adoption of Internet-based communication tools by researchers
(e.g., e-mail, electronic discussion boards, electronic mailing lists, videoconferencing, weblogs)
has led to profound changes in social interaction
and collaboration among them. Research suggests that Internet technologies can improve and
increase communication among noncollocated
researchers, increase the size of work groups,
increase equality of access to information by
helping to integrate disadvantaged and less established researchers, help to coordinate work
more efficiently, help to exchange documents
and information quickly (
Carley
&
Wendt, 1991;
Nentwich, 2003). There is abundant research on
new forms of group work originated from the use
of computer technologies. Carley and Wendt (1991)
use the term extended research group to refer
to very large, cohesive, and highly cooperative

research groups that, even being geographically

dispersed, are coordinated under the supervision of a single director.
The term collaboratory
is also used to refer to similar groups (Finholt,
2002). Although there is much research on how
Internet technologies are used by unified and
cohesive work groups to collaborate (e.g., Moon
& Sproull, 2002; Walsh & Maloney, 2002), less
attention has been paid to how the Internet facilitates collaboration among researchers outside
these highly cohesive groups. Weblogs (blogs) can
become a useful tool for this type of collaboration
and for the creation of virtual groups. Weblogs
are frequently updated Web pages, consisting of
many relatively short postings, organized in reverse chronological order, which tend to include
the date, and a comment button so that readers
can answer (Herring, Scheidt, Bonus, & Wright,
2004). They enable users to communicate with
a worldwide nonrestricted community of people
in similar fields, which leads to several forms of
collaboration. The purpose of this article is to

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Academic Weblogs as Tools for E-Collaboration Among Researchers

present a brief overview of the uses of weblogs

as tools for research e-collaboration.
Defining the concept of research e-collaboration precisely is extremely difficult. Here we
assume that members of a virtual community
engage in research e-collaboration when they
use e-collaborating technologies in order to share
information and discuss issues which contribute
to advancing knowledge in a specific area.

Background
The term Weblog was coined by Jorn Barger in
1997 to refer to personal Web sites that offer frequently updated information, with commentary
and links. Blood (2002) classifies blogs into two
styles: the filter type, which includes links pointing to other sites and comment on the information
on those sites, and the personal-journal type, with
more emphasis on personal self-expressive writing. There are many other types of blogs described
in the literature, defined on the basis of different
criteria; for example, knowledge blogs (k-blog),
community blogs, meta-blogs.
The capabilities of blogs make them helpful tools for communication between members
of a community or organisation. Some types
of weblogs have originated as an answer to the
communicative needs of specific communities; for example, knowledge blogs, weblogs for
personal knowledge publishing. Kelleher and
Miller (2006) describe knowledge blogs as the
online equivalent of professional journals used
by authors to document new knowledge in their
disciplines. A related concept is that of personal
knowledge publishing, defined by Paquet (2002)
as an activity where a knowledge worker or researcher makes his observations, ideas, insights,
interrogations, and reactions to others writing
publicly in the form of a weblog. Many corporate and academic blogs make use of capabilities
that afford collaboration: they enable scholars to
communicate with a wide community, fostering

1292

peer review and public discussion with researchers from different disciplines. These weblogs
have a precedent in what Harnard (1990) terms
scholarly skywriting: using multiple e-mail
and topic threaded Web archives (e.g., electronic
discussion) to post information that anybody can
see and add their own comments to.
There are many types of academic blogs (blogs
from journal editors, individual scholars blogs,
research groups blogs, PhD blogs), each of them
used for different purposes. For instance, while
the main purpose of the weblogs implemented
by universities is discussion, weblogs by PhD
students are mainly used to comment on the days
progress and on the process of PhD writing, and
blogs from journal editors are usually filter blogs,
which provide links to articles or which comment
on news related to the journal topic.
The uses of weblogs in research have been
discussed in several papers and blog posts
(Ameur, Brassard,, & Paquet, 2003; Efimova,
2004; Efimova & de Moor, 2005; Farmer, 2003;
Mortensen & Walker, 2002; Paquet, 2002). These
researchers depict blogs as facilitating scientific
enquiry in two ways: (1) they help to access and
manage content, through features such as archives,
RSS (an automated system that enables bloggers
to syndicate their content to other blogs), searchable databases, post categories; and (2) they are
tools for collaboration, through communication
and network features. These features include
hyperlinks, comments, trackbacks (records of
the Web address of the blogs that have linked to
a blog posting), RSS, or blogrolls (a list of blogs
the author usually reads, and that, therefore,
deal with similar topics). But the most important
ingredient for collaboration is the bloggers
perception of blogs as a means to point to, comment on, and circulate information and material
from other blogs.
As a tool for collaborative research, blogs have
several uses: (1) supporting community forming; (2) helping to find other competent people
with relevant work; (3) facilitating connections

Academic Weblogs as Tools for E-Collaboration Among Researchers

between researchers: blogs make it easier for

researchers to develop, maintain and activate
connections with others; (4) making it possible
to obtain speedy feedback on ideas; (5) fostering
diversity and allowing for radical new ideas to be
acknowledged, circulated and discussed; (6) fostering communication and collaboration between
researchers from different fields; (7) supporting
the initiation and development of conversations.
In weblogs, ideas can be generated and developed
through discussion with others.

E-Collaborating Through
Academic Weblogs
We used a corpus of 100 academic English-language weblogs, collected between January and
March, 2006, in order to analyse how they are
used as tools for e-collaboration and how the elements of weblogs contribute to this purpose. Many
academic weblogs are not intended as collaborative tools and thus do not include social software.
In other cases, although comments are allowed,
interactivity is very limited, and the weblogs are
not really used as conversation tools. Therefore, for
the purpose of this research, we have selected only
active weblogs that include and make use of the
comment feature. Drawing on previous research
on blog analysis (e.g., Herring et al., 2004) and on
our initial inspection of the blogs of the corpus,
we identified the functional properties that make
interaction and collaboration possible (e.g., comments, links). Then, we analysed the function of
the 15 most recent entries in each blog (e.g., ask
for feedback), the function of the comments to
these entries (e.g., disagree with a post), and the
types of links included in the postings and comments (e.g., to other blogs, to news sites, to Web
sites by others, to the bloggers own site).
The analysis showed that academic weblogs
can be used to support the following forms of
collaboration (or collaboration stages):

1.

Finding like-minded peers and establishing

contact and collaboration with them, thus
helping to create a virtual community of
people interested in the same topic.

Weblogs share features of personal and public genres. As personal genres, they are used by
researchers to keep records of their thoughts,
ideas or impressions while carrying out research
activities. They also have an open nature, since
anybody can access and comment on the authors
notes, and they are in fact intended to be read. By
revealing the research a researcher is currently
involved in and, at the same time, providing tools
for immediate feedback on that research, weblogs
facilitate collaboration. In addition, since weblogs
are open to the public, not restricted to peers
working in the same field, they facilitate contact
among scholars from different disciplines and
invite interdisciplinary knowledge construction
(Ameur et al., 2003).
Blogrolling lists enable readers of a weblog to
find peers with relevant work. These lists do not
only point to related work, but also function as
signs of personal recommendation, which help
others to find relevant contacts faster (Efimova,
2004). Links and comments are also useful to expand a researchers community. Other researchers
read the postings in the research blog and leave
comments, usually including links to their own
research. Authors of weblogs can also include
links to articles on similar work carried out by
other researchers, thus enabling the readers to get
into contact with them.
Many researchers use the comment feature to
explain that they are working on areas similar to
the ones dealt with in the original posting and to
suggest the possibility of collaboration or contact,
in a more or less direct way (e.g., I would like to
keep in touch about your findings). Community
blogs are sometimes used by bloggers to summarize briefly their research and invite collaboration
and contact with others.

1293

Academic Weblogs as Tools for E-Collaboration Among Researchers

a. I am in the middle of a research project that

looks at (...). I have a research log in which I make
occasional observations, related and sometimes
unrelated, to the ongoing work. Interested? Let
me know and I will give you the URL.
The entry above got several comments of
people eager to exchange ideas; for example:
b. Id love to exchange ideas (). How can we
establish contact?
2. Asking for and receiving advice/feedback
from others.
Weblogs enable researchers to post an idea or
theory and have it circulated and discussed, thus
helping the author revise and refine it. Authors
who want to use weblogs for this purpose embed
elements that facilitate hyperlinked conversations
into them, such as links, comment software, and
trackback features.
Very frequently, authors use weblogs to explicitly request feedback, comments, suggestions, and
so forth on a paper, presentation, or idea. Comment software facilitates the discussion between
the researchers asking for feedback and anybody
wanting to contribute: Other bloggers state their
agreement and disagreement with the author and
with other commentators, ask questions, offer
new perspectives, and so forth.
a. I was asked by the editors of the New Palgrave
Dictionary of Economics to contribute a short
article on the analysis of variance (...) Heres
the article. Any comments () would be appreciated.
The entry above got eight comments suggesting ways to improve the paper; for example:
b. * Missing from the draft is a discussion of
ANOVA versus regression (...)

1294

* Here is one question you might address: When

is ANOVA preferred to linear regression? (...)
Comments may also include links to other
blogs with similar topics and thus help the original author with other peoples perspectives. The
weblog becomes a forum for discussion of the
paper prior to publication, similar to that of a
conference. When a blogger requests and gets
feedback on a paper, he/she usually acknowledges
the help from the others and shows the result of
the revision in an update to the original entry. The
result is a more consensual paper, which makes
use of knowledge from other researchers.
Posting articles for peers to comment on in
a weblog has several advantages. In the first
place, the author can get immediate and public
feedback, thus accelerating the development of
new knowledge and involving many people in the
review process: Anybody can contribute, even
people unknown by the author, nonestablished
researchers who are not usually involved in peerreviewed publishing, and researchers from different cultural backgrounds. Besides, since bloggers
usually comment on others ideas on their own
blogs, feedback may come from others than just
immediate readers. Authors can post any idea,
even very innovative or radical ideas which would
not be accepted in traditional publishing, and, as
Paquet (2002) points out, they can use weblogs to
let others know not only about positive results,
but also about negative results, blind alleys and
problems. Peers can help by providing their ideas,
suggestions, and so forth, and this way knowledge
construction becomes a communal activity.
In many cases, researchers do not ask for
feedback, but for references, information or
resources on a topic, or help to solve a problem,
hence using blogs as a tool to get information on
a topic from competent peers they may not know
personally.
3. Discussing topics or theories in which
several researchers are interested, thus

Academic Weblogs as Tools for E-Collaboration Among Researchers

contributing to developing or articulating

an idea and to reinforcing the links between
the members of the virtual community.
Efimova and de Moor (2005) show how
weblogs can serve as conversation tools which
support the exchange of multiple perspectives
and fast and meaningful reactions and facilitate
the joint development of ideas. They define a new
form of conversation enabled through the use of
weblogs, distributed weblog conversation: conversations across many weblogs, where several
authors link their ideas to construct a collective
piece of knowledge.
The joint articulation of ideas by several researchers relies on elements such as links, comments and trackback features. Links may be to text
but also to audio or video documents presenting
or illustrating an idea, to simulations, to other
Web pages or to other posts. Scholars use links to
present and discuss an idea by someone else, or to
articulate and summarize a previous discussion,
before contributing their own idea and presenting it for others to comment on. Very frequently,
when presenting others arguments on a topic,
scholars include quotes accompanied by links to
the original post, that way allowing the reader to
follow the conversation in the original blog.
The comment feature facilitates discussion
based on the posting. Bloggers write their comments, get answers back from the author of the
original post, comment on other bloggers comments, and so forth, thus engaging in a conversation that may contribute to the development of
innovative ideas. Comments are used for many
different purposes: to show agreement or disagreement with the blogger, to ask for clarification on
a specific point that the commentator does not
understand or does not agree with, to request
further information, to strengthen a point, to
provide links to ones own blog, and so forth.
This back-and-forth may lead to collaborative
revision and refinement, and even permuting, of
an idea or argument, e.g.:

a. * In general, I really side with your post. Over

at Anthropology.net, there have been a couple
posts and discussions on (...). However, this
post of yours, much more eloquently advocates
breaking the boundaries between the West and
non-West (...)
* Thanks, X. Ill have a look at the posts you link
to (...)
Since discussion of an idea may be distributed
across several weblogs, it is necessary to have
mechanisms to keep track of the discussion that
ones posting may have sparked. One such mechanism is trackbacks. Trackback is a special feature
of the program Movable Type, which notifies the
blogger when another weblog has linked to his/her
weblog. Other users can view the trackbacks and
thus follow the discussion in other weblogs.

Future Trends
Weblogs are a genre whose capabilities offer a
great potential for collaboration among researchers. However, there are still few academics who
engage in blogging, and in many cases academic
blogs are used for purposes other than collaboration. Several reasons have been suggested: the risk
of sharing information and having ideas stolen
or attacked before the research is published, the
fear of damaging credibility, the time that blogging takes away from more traditional research
activities (Lawley, 2002; Mortensen & Walker,
2002).
Since the weblog is quite a recent genre,
there is little research on how weblogs are used
for enquiry and knowledge creation. Their open
nature makes weblogs appropriate for nonestablished researchers, who can that way collaborate
with peers they dont know, and researchers who
seek collaboration with a worldwide non-discipline-restricted community of researchers. There
are many different types of weblogs written by

1295

Academic Weblogs as Tools for E-Collaboration Among Researchers

academics and researchers. Further research is

needed to determine which of these weblogs are
more involved in collaboration, which form of
collaboration takes place in the different academic
weblogs, and how bloggers use language to engage
in collaboration.

interactive generation and articulation of ideas

and theories and for the constant revision and
update of theories. The informal and open nature of weblogs facilitates collaboration among
anybody interested in a topic and, hence, fosters
interdisciplinary collaboration.

Conclusion

References

In this article, we start from the assumption that,

since scientific knowledge is socially constructed,
scientific knowledge creation is collaborative in
nature. Thus, rather than considering collaboration
as restricted to the work by unified and cohesive
groups involved in a project under the supervision of a coordinator, we see it as any sharing of
information and exchange of perspectives which
contribute to advancing knowledge in a specific
area.
Weblogs are a genre whose capabilities allow
the exchange of any type of information and support conversations with a rhizomatic structure,
thus being an appropriate tool for collaboration and
for the sharing of ideas among several researchers
in different places around the world. Weblogs can
represent and distribute knowledge in different
forms; that is, text, multimedia, simulations. They
have a modularized format, relying highly on
hypertext: weblogs consist of different postings,
linked together to postings in other weblogs, on
which they comment or which comment on them,
making it possible to construct a web of knowledge
on a topic. Links and comments allow connecting
different pieces of related research into a coherent whole, thus facilitating the development of
cumulative knowledge.
The use of CMC tools (and specifically of
weblogs) does not only lead to an increase in collaboration, but also to a change in collaboration
patterns. Weblogs enable almost instantaneous
communication and feedback among a worldwide community, thus allowing for a much more

Ameur, E., Brassard, G., & Paquet, S. (2003).

Using personal knowledge publishing to facilitate
sharing across communities. In M. Gurstein (Ed.),
Proceedings of the Third International Workshop
on (Virtual) Community Informatics: Electronic
Support for Communities. Retrieved February 13,
2006, from: http://is.njit.edu/vci-www2003/pkpsharing-across-comms.pdf

1296

Blood, R. (2002). The Web log handbook: Practical advice on creating and maintaining your blog.
Cambridge, MA: Perseus Publishing.
Carley, K., & Wendt, K. (1991). Electronic mail and
scientific communication: A study of the SOAR
extended research group. Knowledge: Creation,
Diffusion, Utilisation, 12(4), 406-440.
Efimova, L. (2004). Discovering the iceberg of
knowledge work. In Proceedings of the Fifth
European Conf. on Organisational Knowledge,
Learning, and Capabilities (OKLC 2004). Retrieved February 20, 2006, from https://doc.telin.
nl/dscgi/ds.py/Get/File-34786
Efimova, L., & de Moor, A. (2005). Beyond
personal webpublishing: An exploratory study
of conversational blogging practices. Proceedings of the 38th Hawaii International Conference
on System Sciences (HICSS-38) (pp.107a). Los
Alamitos, CA: IEEE Press.
Farmer, J. (2003). Personal and collaborative
publishing (PCP): Facilitating the advancement
of online communication and expression within

Academic Weblogs as Tools for E-Collaboration Among Researchers

a university. Retrieved March 13, 2006, from

http://radio.weblogs.com/0120501/myimages/
030829PCP.pdf
Finholt, T. A. (2002). Collaboratories. Annual
Review of Information Science and Technology,
36, 73-108.
Harnard, S. (1990). Scholarly skywriting and the
prepublication continuum of scientific inquiry.
Psychological Science, 1, 342-343.
Herring, S. C., Scheidt, L. A., Bonus, S., & Wright,
E. (2004). Bridging the gap: A genre analysis of
weblogs. Proceedings of the 37th Hawaii International Conference on System Sciences (HICSS-37)
(pp. 101-111). Los Alamitos, CA: IEEE Computer
Society Press.
Kelleher, T., & Miller, B. M. (2006). Organizational blogs and the human voice: Relational
strategies and relational outcomes. Journal of
Computer-Mediated Communication,11(2). Retrieved March 20, 2006, from http://jcmc.indiana.
edu/vol11/issue2/kelleher.html
Lawley, L. (2002) Thoughts on academic blogging.
Retrieved March 20, 2006, from http://many.corante.com/archives/2004/04/01/thoughts_on_academic_blogging_msr_breakout_session_notes.
php
Mortensen, T., & Walker, J. (2002). Blogging
thoughts: Personal publication as an online research tool. In A. Morrison (Ed.), Researching
ICTs in context (249-279). Oslo, Norway: InterMedia Report.
Moon, Y., & Sproull, L. (2002). Essence of distributed work: The case of the Linux Kernel. In
P. Hinds & S. Kiesle (Eds.), Distributed work (pp.
381-404). Cambridge, MA: MIT Press.
Nentwich, M., (2003). Cyberscience: Research in
the age of the Internet. Vienna: Austrian Academy
of Sciences Press.

Paquet, S. (2002). Personal knowledge publishing and its uses in research. Retrieved March 12,
2006, from http://radio.weblogs.com/0110772/
stories/2002/10/03/personalKnowledgePublishingAndItsUsesInResearch.html
Walsh, J. P., & Maloney, N. G. (2002). Computer network use, collaboration structures and
productivity. In P. Hinds & S. Kiesler (Eds.),
Distributed work (pp. 433-458). Cambridge, MA:
MIT Press.

KEY Terms
Blogroll: A list of links to Web pages the
author of a blog finds interesting.
Comment: A link to a window where readers
can leave their comments or read others comments and responses from the author.
Personal Knowledge Publishing: The use
of Weblogs by knowledge workers or researchers
to make their observations, ideas, insights and
reactions to others writing public.
Research E-Collaboration: The use of
e-collaborating technologies in order to share
information and discuss issues which contribute
to advancing knowledge in a specific area.
Scholarly Skywriting: Using multiple e-mail
and topic threaded Web archives (e.g., electronic
discussion) to post information that anybody can
see and add their own comments to.
Trackback: A link to notify the blogger
that his/her post has been referred to in another
blog.
Weblog: A frequently updated Web page,
consisting of many relatively short postings,
organized in reverse chronological order, which
tend to include the date and a comment button
so that readers can answer.

This work was previously published in the Encyclopedia of E-Collaboration, edited by N. Kock, pp. 1-6, copyright 2008 by
Information Science Reference, formerly known as Idea Group Reference (an imprint of IGI Global).
1297

1298

Chapter 3.20

Assessing Weblogs as
Education Portals
Ian Weber
Texas A&M University, USA

Introduction
Education is one of the key sectors that has benefited from the continuous developments and
innovations in information and communication
technology (ICT). Web-based facilities now
provide a medium for learning and a vehicle for
information dissemination and knowledge creation (Khine, 2003). Accordingly, developments
in ICTs provide opportunities for educators to
expand and refine frameworks for delivering
courses in innovative and interactive ways that
assist students achieve learning outcomes (Kamel
& Wahba, 2003). However, the adoption of ICTs
has also created tensions between traditional
control and directiveness in teaching and student-centred learning, which relies on flexibility,
connectivity, and interactivity of technology-rich
environments.
This chapter examines the introduction of
Web-based technologies within a media studies
course. The objective was to establish a community of learning, which provides students with a

portal or entranceway into a common work area

and out to networks of media related organizations.
So doing, a pilot study was conducted within the
Department of Communication at Texas A&M
University to blend Weblog facilities with a classroom setting to enhance students interpersonal
and content interaction, and build citizenship
through participation and collaborative processes.
Four key aims frame this study:
1.

2.

3.

provide an accessible, interactive online

environment in which students can participate with peers and engage with new media
technologies within a learning community
setting;
develop an instructional technology framework that enhances the learning experience
and outcomes within online educative environments;
establish a portal or gateway for students to
access media advocacy and special interest
groups and enhance and diversify perspectives on global media; and

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Assessing Weblogs as Education Portals

4.

evaluate student-learning experiences facilitated through innovative online instructional

technologies.

Background
Early approaches to integrating ICTs into education environments emerged from conventional
learning models, originating from the objectivist
approach in which a reality exists and experts
instruct individuals of that reality (Belanger &
Slyke, 2000). However, such teacher-centric,
information-based approaches failed to adequately prepare students to become independent learners. Responding to these limitations,
educators embraced learner-centric approaches
such as constructivism, which leant weight to
the empowerment of individuals to take charge
of their own learning environments. As Wilson
(1996) suggests, the constructivist movement in
instructional design emphasized the importance
of providing meaningful, authentic activities that
can help the learner to construct understandings
and develop skills relevant to solving problems and
not overloading them with too much information.
Solis (1997) supports this position, suggesting that
student-centred learning relies on groups of
students being engaged in active exploration, construction, and learning through problem solving,
rather than in passive consumption of textbook
materials (p. 393).
In spite of these favorable positions, Khine
(2003) warns that creating such learning environments supported by ICTs can be intrinsically problematic. Accordingly, it is critically
important that careful planning and design is
employed at the early stages of instructional
design to provide proper support and guidance,
as well as rich resources and tools compatible
to each context. When adequate consideration
is given to new learning and teaching strategies
that incorporate ICTs, real opportunities exist
for educators to provide students with a dynamic

environment to learn, to think critically, and to

undertake productive discussions with their peers
in supportive, constructive environments. Given
the potential of such technology-rich learning
environments, educators have the opportunity
to make student learning more interesting and
enriching, preparing them for the demands of
the future workplace. Accordingly, instructional
designers must consider matching the strengths
of new technology (flexibility, connectivity, and
interactivity) with traditional forms of education
(control and directiveness) to inspire, motivate,
and excite students in ways that maximize the
individuals learning potential.
Achieving these goals requires the development of individual competencies in problem
solving, participation, and collaboration, and
communities of learning (Kernery, 2000; Khine,
2003; Wilson & Lowry, 2000). Problem solving
provides ways for students to engage with authentic episodes, providing opportunities for students
and educators to examine events and reflect on
solutions. One way of maximizing the benefits of
problem solving is to support these through collaborative processes, which can be built around
these episodes by focusing on the use of instructional methods to encourage students to work
together as active participants on such tasks. Such
efforts can be facilitated through structuring and
organizing online interactions using computermediated communication, which provides the
means to overcome limitations of time and place
(Harasim, Calvert, & Groeneboer, 1997). Based
on the principles of the transformative paradigm,
multiple perspectives, and flexible methods, it is
possible for students to adapt, to process and to
filter content into their own logical frameworks,
resulting in outcomes that may not be thoroughly
predictable (Bento & Schuster, 2003). As Morrison and Guenther (2000) note, such collaborative
environments provide a forum for students to discuss issues, engage in dialogue, and share results.
However, Bento et al. (2003) also warn that one
of the main challenges in Web-based education

1299

Assessing Weblogs as Education Portals

is to achieve adequate participation. They offer

a four-quadrant taxonomy of learner behaviours
when participating in interpersonal and content
interaction--missing-in-action, witness learners,
social participants, and active learnersas a way
of understanding this dynamic.
When building components of problem solving, participation, and collaboration around small
group processes, or learning communities, it is
critical that these dynamics have beneficial effects on student achievements and psychological
well-being. As Khine (2003) suggests, building
community imparts a common sense of purpose
that assists members grow through meaningful
relationships. Accordingly, learning communities
can be characterized by associated groups of
learners, sharing common values and a common
understanding of purpose, acting within a context
of curricular and co-curricular structures and
functions that link traditional disciplines and
these structures (p. 23). Rickard (2000) equates
the notion of common values to that of campus
citizen, in which students not only engage in
educative endeavour but also learn networking
and become life-long members of our communities (p. 13).
Such communities though should not be limited to just participants within narrowly defined
student groups or the domain of institutional
environments. ICTs also provide opportunities
for students to connect to other informative communities related to their area of discipline or study
via the Web. So doing, the educator can build into
the instructional design gateways or portals to
direct participants of small learning communities
to other relevant organizations, groups, and individuals to extend campus citizenry and enhance
knowledge creation. Tatnall (2005) suggests that
such portals can be seen:
as a special Internet (or intranet) site designed
to act as a gateway to give access to other sites.
A portal aggregates information from multiple
sources and makes that information available to

1300

various users . In other words, a portal offers

centralised access to all relevant content and
applications. (pp. 3-4)
In accessing Tatnalls (2005) notion of portals,
it is possible to think of these starting points in
diverse ways. While no definitive categorization of the types of portals exists, Davison et al.
(2003) offer a list of possible alternatives: general
portals, community portals, vertical industry
portals, horizontal industry portals, enterprise
information portals, e-marketplace portals, personal/mobile portals, information portals, and
niche portals. However, it is important to point
out that these categories are not mutually exclusive, highlighting the malleable nature of these
gateways for educators to blend the strengths of
one with advantages of others to achieve a more
effective portal design.
Even though portals are conceptually difficult
to categorize and define, there exist a number of
important characteristics that assist in facilitating the objectives of gateways as access points
to, and aggregators of, information from multiple sources. For example, Tatnall (2005) draws
from a number of scholars to present a general
guideline of beneficial characteristics employed
to facilitate community collaboration amongst
users and the rapid sharing of relevant content
and information. For example, the characteristics
of access, usability, and functionality, as well as
sticky web features like chat rooms, e-mail, and
calendar functions, have been used to assist in
maintaining user interest, attention, and participation within a site.
Within the education sector, portals offer great
potential to achieve the kinds of goals laid out so
far. However, portal development in these learning environments present a number of challenges
for universities as they continue to grapple with
a variety of business, organizational, technical,
and policy questions within the framework of a
larger technology architecture. For example, Katz
(2002) highlights the following challenges:

Assessing Weblogs as Education Portals

build standards to create compelling and

sticky Web environments that establish
communities rather than attract surfers;
create portal sites that remain compelling
to different members of the community
within the short-term and throughout student
lives;
create the technical and organizational infrastructure to foster cradle-to-endowment
relationships via virtual environments;
and
integrate physical and virtual sites to foster
social and intellectual interactions worthy
of the universitys mission.

Within this framework of challenges, the

emphasis continues to return to the realization
that bringing education and ICTs together strategically relies less on technology per se than on
the educators ability to design portals that are
instructionally sound (to meet learner needs),
customizable and personalized (for each member
of the community), meets the institutional vision
and mission, and fits within the larger technology
infrastructure. However, Katz (2002) argues that
while the challenges are great, opportunities can be
realized. With universities engaged in relationship
management enterprises, the situation requires a
proactive role in developing a belief system, a
worldview, a set of approaches and technologies
organized to arrange and rearrange both our front
door and our rich holdings (p. 13).

Enhancing the
Undergraduate Experience
Texas A&M Universitys (TAMU) focus on Enhancing the Undergraduate Experience provides
a series of strategies to increase and expand opportunities for students to be actively involved in
learning communities. Establishing such an experience is guided by the recognition that students

are involved in transitions (freshman to career)

and connections (building networks).
For most students, these connections do not happen automatically. Rather, they happen because
a professor or advisor intentionally creates opportunities and challenges that serve as a catalyst
for such connection building. (TAMU, 2005)
To achieve and maintain such connections,
TAMU implemented a three-tier model for development and implementation of foundation,
cornerstone, and capstone learning communities
within the university setting (see Figure 1). In
developing this model for establishing learning
communities, a number of recommendations
were made:
1.

2.

3.

Draw from faculty and staff knowledge to

create initiatives that make existing educational opportunities even better.
Provide resources and incentives to develop,
implement, and access learning communities; and encourage innovative pedagogies
to support and reward the implementation of
technology-mediated instruction, use of peer
instructions and collaborative learning.
Provide a clearinghouse so the benefits from
experience and expertise can be utilized in
future design and implementation of learning communities. (TAMU, 2005)

With this framework in mind, a Weblog was

integrated into the COMM 458 Global Media
course using a generic blogger.com template as
part of pilot study funded by a $3,000 TAMU
Summer Institute for Instructional Technology Innovation Grant.

The template included an archive

(of posts and comments), calendar function, a
database of user information including e-mail
links to registered members, e-mail notification
facility, and links to other relevant websites. In
providing 22 students access to Weblog facilities,

1301

Assessing Weblogs as Education Portals

Figure 1. Learning community model (TAMU, 2005)

Time Line

Courses

Content

Freshman

Foundation

Introduce
competencies

Sophomore,
Junior, or both

Senior
Career or
Graduate School

Core Curriculum Courses

Reinforce
Critically analyse,
and integrate
Cornerstone
Master depth of knowledge,
competencies in
Communication
the discipline
Emphasize,
Critically analyse,
Master depth of knowledge,
synthesize
and apply
Personal integrity,
Capstone
Communication,
competencies in
Contribute to society
the discipline

the study set out to establish a model of content

production and platform interoperability for
inspiring innovative and creative appropriation
of Web-based facilities by TAMU students and
faculty. Accordingly, the Weblog was strategically
integrated into the course to achieve the following
outcomes:

1302

Outcomes
Critically analyze,
Personal integrity,
Contribute to society,
Communication

facilitate community building and understandings of social networking and citizenship at local, national and international
levels;
provide a platform for students to evaluate,
communicate and critique current issues
and problems within informal and engaging
online environments;
improve content, information and knowledge
management within the course structure;

improve teaching and learning through

understanding the dynamic between participation and collaborative processes within a
blended educative environment; and
develop alternative pedagogical and technological strategies to support and encourage
each student to become an independent
learner.

All students were invited to be members of the

Weblog community by e-mail, which contained
a hyperlink that directed them to blogger.coms
registration site. There, students selected a user
name and password, which brought them into the
course Weblog site (see Figure 2). Once in the site,
members could interact with each other by posting summaries of readings (including discussion
points and questions), comment on posts in which

Assessing Weblogs as Education Portals

Figure 2. COMM458 Global Media Weblog site

students raised issues and problems (moderated

by the course instructor), and source stories and
information from national and international news
services (e.g., CNN, BBC, or Aljazeera). Users
were also given the task of building a database of
relevant links to outside media organizations and
advocacy groups to assist each other in answering problems and issues posed within online and
classroom discussions. All students were assigned
administration rights to the Weblog to instil a
sense of ownership among members.

Future Trends
Participation within the Global Media course
(online and classroom) was measured in several
ways, including the number of entries (posts and
comments), user log-ins, survey, interviews, and

observations. Overall, students and instructor

made a total of 303 entries (72 posts and 231
comments) over the 10-week period from September to November of the 2005 Fall Semester.
On average, there were 3.25 comments per post.
However, included in this number were 108 comments (81% response rate) by students to a series
of six quizzes (which accounted for 15% of the
courses total grade) implemented throughout the
semester to test students general knowledge of key
theories and concepts. When the quiz comments
were excluded and the figure re-calculated, the
average number of comments per post significantly
decreased to 1.70. These figures would suggest
a relatively poor level of participation in online
interactions. However, these figures do not reveal
the complexity of the online-classroom dynamic
in catering for different learner preferences.
Another way of assessing student participation

1303

Assessing Weblogs as Education Portals

Figure 3. Taxonomy of COMM 458 course student behavior relating to content and interpersonal interaction

is to categorize their behavior into types of

interactions through observing participation in
both settings.
Drawing from Bento et al. (2003) four-quadrant taxonomy, it is possible to map the types of
behaviors these learners engaged in as part of the
participatory and interactivity components of the
Global Media course (see Figure 3). Quadrants I
and II share the characteristics of low interpersonal interaction. However, whereas Quadrant I

1304

students do not care about course content and their

peers, witness learners are actively engaged
with the course materials and discussion (high
content interaction), log in frequently and do
all the readings but do not actively contribute to
course discourse (low interpersonal interaction).
On the other hand, Quadrants III and IV share
the characteristic of high interpersonal interaction. Students in Quadrant III (high interpersonal
interaction, low content interaction) thrive on

Assessing Weblogs as Education Portals

the social aspects of education, demonstrating

characteristics of great conversationalists, with
high communication and interpersonal skills.
Such students, though, engage in social interaction most often at the expense of reflection and
thoughtful consideration of course content. Quadrant IV represents what educators define as good
participation, with high content interaction and
interpersonal interaction. Their contributions in
discussions are substantive and frequent. Accordingly, Bento et al. (2003) argue that these students
contribute not only to the problem-solving task
but also to building and sustaining relationships
in the learning community.
Given the findings of the study, the Weblog
added value to the learning community by providing students with access into a common area
to connect and interact with one another and as
a gateway out to sources of information beyond
the confines of the institutional environment.
Through student-established links to media organizations and advocacy groups, members connected to important and divergent understandings
and knowledge of the role of media in a global
context. One student comment encapsulates the
general feeling of course participants toward the
Weblog facilities:
Even if all students had equal enthusiasm to
participate in class discussion, class time limits
everyone from making significant contributions
and feeling part of learning. Blogs provide a
place for sharing related material [discussion
and provision of links to outside sources] that the
classroom setting does not, and a way to question
or comment on material we have already covered.
In general, blogging lets students engage with
each other in a way different from class. By allowing discussions to continue and expand, it helps
students who struggle with the material or who
are afraid to ask questions in class. It certainly
helped me to learn much more than just reading
textbook information.

In spite of these positive responses, the

chronologically organized Weblog structure
offered insufficient flexibility and functionality to
increase online participation. Given the findings,
the study recommends re-designing the Webbased learning environment using software,

innovative interfaces, and workflows. So doing,

there is a need to:
1.

2.

3.

4.

Develop a technology interface that acts as

an entranceway into a series of linked, embedded pages so (a) information storage and
retrieval; (b) discussion and commentary; (c)
communication; and (d) project facilitation
can be purposefully compartmentalize to
improve technology, content, interpersonal
and intellectual interaction.
Integrate chat room facilities into the portal
design to create more compelling, sticky
Web-environments for participants to interact in real-time with peers and instructor.
Introduce more comprehensive monitoring facilities to track technology, content,
interpersonal and intellectual interaction,
providing instructors with timely diagnostic
metrics to enhance decision making during
course implementation.
Develop more strategic collaborative, problem-solving tasks and exercises to increase
participation and improve content, interpersonal, and intellectual interaction.

Conclusion
This article presented pilot study findings on
using weblog facilities to increase participation
and improve interaction in Web-based learning
communities. The study revealed a number of
aspects to assist educators in providing more
structured and comprehensive online learning
environments. The most important of these are
people-related functions such as creating strategically-focused collaborative learning environ-

1305

Assessing Weblogs as Education Portals

ments and improving the quality and quantity

of, and accessibility to, course metrics to help
educators make informed, timely decisions in
online instructional environments. Accordingly,
it is anticipated that the recommended strategies
will improve participation and interactivity and
contribute to more productive, supportive learning communities.

References
Belanger, F., & Slyke, C. V. (2000, Spring). Enduser learning through application play. Information Technology, Learning, and Performance
Journal, 18(1), 61-70.
Bento, R., & Schuster, C. (2003). Participation:
The online challenge. In A. Aggarwal (Ed.),
Web-based education: Learning from experience
(pp. 156-164). Hershey, PA: Information Science
Publishing.
Davison, A., Burgess, S., & Tatnall, A. (2003).
Internet technologies and business. Melbourne,
Australia: Data Publishing.
Harasim, L., Calvert, T., & Groeneboer, C. (1997).
Virtual-U: A Web-based system to support collaborative learning. In B. H. Khan (Ed.), Web-based
instruction (pp. 149-158). Englewood Cliffs, NJ:
Educational Technology Publications.
Kamel, S., & Wahba, K. (2003). The use of a
hybrid model in Web-based education: The
global campus project. In A. Aggarwal (Ed.),
Web-based education: Learning from experience
(pp. 331-346). Hershey, PA: Information Science
Publishing.
Katz, R. N. (2002). Its a bird! Its a plane! Its
a portal? In R. N. Katz & Associates (Eds.),
Web portals & higher education: Technologies
to make IT personal (pp. 1-14). San Francisco:
Jossey-Bass.

1306

Kernery, E. R. (2000). Developing online collaboration. In A. Aggarwal (Ed.), Web-based

learning and teaching technologies: Opportunities and challenges (pp. 227-245). Hershey, PA:
Idea Group Publishing.
Khine, M. S. (2003). Creating a technology-rich
constructivist learning environment in a classroom management module. In M. S. Khine & D.
Fisher (Eds.), Technology-rich learning environments: A future perspective (pp. 21-40). River
Edge, NJ: World Scientific.
Morrison, G., & Guenther, P. (2000). Designing
instruction for learning in electronic classrooms.
In R. Weiss, D. Knowlton, & B. Speck (Eds.), Principles of effective teaching in the online classroom
(pp. 15-21). San Francisco: Jossey-Bass.
Rickard, J. (2000, October). Portals: Creating lifelong campus citizens. Converge Magazine, 13.
Solis, C. R. (1997). Virtual worlds as constructivist learning tools in a middle school education
environment. In B. H. Khan (Ed.), Web-based
instruction (pp. 393-398). Englewood Cliffs, NJ:
Educational Technology Publications.
TAMU. (2005, May). Enhancing the undergraduate experience: Final report. College Station, TX:
Texas A&M University.
Tatnall, A. (2005). Portals, portals everywhere. In
A. Tatnall (Ed.), Web portals: The new gateways
to Internet information and services (pp. 1-14).
Hershey, PA: Idea Group Publishing.
Wilson, B. (1996). Constructivist learning environments: Case studies in instructional design.
Hershey, PA: Educational Technology Publications.
Wilson, B., & Lowry, M. (2000). Constructivist learning on the Web. In E. Burge (Ed.), The
strategic use of learning technologies (pp. 79-88).
San Francisco: Jossey-Bass.

Assessing Weblogs as Education Portals

Key Terms
Constructivism: Learning as interpretive,
recursive, building processes by active learners
interacting with physical and social worlds.
Information and Communication Technology (ICT): Refers to an emerging class of technologiestelephony, cable, satellite, and digital
technologies such as computers, information
networks, and softwarethat act as the building
blocks of the networked world.
Learning Communities: Characterized
as associated groups of learners, sharing common values, and a common understanding of
purpose.

Portal: Considered an all-in-one Web site that

acts as a centralized entranceway to all relevant
content and applications on the Web.
Sticky: Web Features include chat rooms,
e-mail links, and calendar functions.
Student-Centred Learning: Focuses on the
needs of the students rather than on teachers and
administrators, resulting in implications for the
design of curriculum, course content, and interactivity of courses.
Weblog: A personal dated log format that is
updated on a frequent basis with new information
written by the site owner, gleaned from Web site
sources, or contributed to by approved users.

This work was previously published in the Encyclopedia of Portal Technologies and Applications, edited by A. Tatnall, pp. 5864, copyright 2007 by Information Science Reference, formerly known as Idea Group Reference (an imprint of IGI Global).

1307

1308

Chapter 3.21

Innovative Technologies for

Education and Learning:
Education and Knowledge-Oriented
Applications of Blogs, Wikis,
Podcasts, and More
Jeffrey Hsu
Fairleigh Dickinson University, USA

Abstract

INTRODUCTION

A number of new communications technologies

have emerged in recent years that have been
largely regarded and intended for personal and
recreational use. However, these conversational
technologies and constructivist learning tools,
coupled with the power and reach of the Internet,
have made them viable choices for both educational learning and knowledge-oriented applications. The technologies given attention in this
article include instant messaging (IM), Weblogs
(blogs), wikis, and podcasts. A discussion of the
technologies and uses, underlying educational
and cognitive psychology theories, and also applications for education and the management of
knowledge, are examined in detail. The implications for education, as well as areas for future
research are also explored.

For many years, the mediums employed for education have remained fairly constant and traditional:
tried and true methods such as the blackboard
and chalk, whiteboards, flipcharts, and overhead
projectors. The employment of computing technologies has resulted in the use of PowerPoint,
e-mail, and Web-based course portals/enhancements such as Blackboard and WebCT.
There have been numerous studies done, and
papers written, about the use of technology in the
classroom, together with work on the related areas
of e-learning, Web-based learning, and online
learning. The usage of computing technologies
in education has been examined in numerous
studies, and there is a sizable body of work on
Web and online learning, including the studies
by Ahn, Han, and Han (2005), Liu and Chen

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Innovative Technologies for Education and Learning

(2005), Beck, Kung, Park, and Yang (2004), and

numerous others.
In particular, some of these technologies
have been recognized as useful in the classroom,
and have been engaged in innovative ways. The
technologies of particular interest are those that
are referred to as conversational technologies,
which allow for the creation and sharing of information (KPMG, 2003; Wagner, 2004). Another
term often used to describe these technologies
is the concept of constructivist learning tools,
which encourage, and are focused on, users creating, or constructing, their own content (Seitzinger,
2006).
The interest in employing these kinds of technologies stems not only from the unique pedagogical benefits gained, but also from the basic
need to stay in tune with the focus and strengths
of todays students. Prensky (2001) suggests that
the students being taught today are no longer the
people our educational system was designed to
teach and that while the students of today can be
termed digital natives, many educators could be
better termed digital immigrants. Yet another
way to look at this is to view earlier educational
approaches as print-based, while those of the
current environment can be called digitallybased, secondly-oral (Ferris & Wilder, 2006).
The purpose of this article is to examine these
technologies and explore both the evolution of
their use from personal applications to that of
educational tools, and also to examine the key
educational applications for which these are
being used. Relevant research and applications
are examined and analyzed. The future of these
technologies for educational and professional use,
together with viable research areas, is examined
as well.

CONVERSATIONAL
TECHNOLOGIES AND
CONSTRUCTIVIST LEARNING
TOOLS
The notion of conversational technologies is
not a new one, as it encompasses many types
of systems that have been widely used for some
time, including e-mail, video conferencing, and
discussion forums.
The term conversational technology is derived from the work of Locke et al. (2000) relating
to conversational exchanges and his Cluetrain
Manifesto. One of the key concepts here is that
markets are conversations and that knowledge
is created and shared using question and answer
dialog. Specific theses that relate to this form of
conversational knowledge management suggest
that aggregation and abstraction of information
helps to create information. Other characteristics
of conversational knowledge management include
the fact that it is fast, stored in different locations,
and does not require sophisticated technologies in
order to be accomplished (Wagner, 2004).
Conversational technologies encompass a wide
range of systems and software, many of which
are familiar, including e-mail, instant messaging,
Web pages, discussion forums, video and audio
content/streaming, wikis, and Weblogs. While
there are specific aspects that are of interest in
terms of the more mature technologies, the ones
that will be given attention in this article are the
issues, impacts, and applications relating to IM,
blogs, wikis, and podcasts. These are technologies
that are newer, have a growing base of users, and
are starting to become recognized as viable tools
for education.
The term constructivist learning tool has also
become associated with these, particularly blogs
and wikis, in that they have a key characteristic
of allowing users to develop and maintain their
own content. Some of the characteristics of con-

1309

Innovative Technologies for Education and Learning

structivist learning include engagement, active

learning, collaboration, real world based, and
the usage of reflection as a part of the learning
process (Seitzinger, 2006).
It should be noted that these technologies and
tools are best suited to course structures where
class collaboration and communication are encouraged, rather than those with an emphasis on
lectures and a presentation of factual information.
In addition, in courses where there is substantial
group work, or projects where a collaborative
document is created, the use of these would be
especially helpful and useful. Both hybrid and
full distance learning courses would be situations
where these could also be used effectively.

TEACHING AND LEARNING: NEW

TRENDS
Conversational and constructivist technologies
are certainly here to stay, as evidenced by their
extensive role in our society. It would therefore
be useful to examine their applicability in the
educational realm. While usage based on popularity or student preference seems to be one factor,
there are also theoretical and conceptual bases
for employing these kinds of technologies in the
classroom.
Earlier paradigms of teaching emphasized
print-based materials for instruction, which
included printed textbooks, paper-based instructional materials, and written tutorials, all of
which are grounded in the notion that the teacher,
lecture, and instructional materials form not only
the basis, but also the authority in the educational
process. The transmission of material from the
teacher (lecture) and/or textbook to the student
(called the print model) is still the central basis
of most teaching, even if they are supplemented
with other methods including discussion and other
forms of student interaction/participation (Ferris
& Wilder, 2006).

1310

However, the advent of digital and conversational technologies has brought forth the new
concept of secondary orality (Ong, 1982). This
concept emphasizes that teaching and learning
should go beyond printed materials toward a
greater emphasis on group work, fostering student
communities, and encouraging student participation. The concept encourages a greater sense of
interaction with and ownership of knowledge,
emphasizing self-awareness and expression, and
effectively using electronic tools (Gronbeck, Farrell, & Soukup, 1991).
The use of conversational technologies can
have a positive impact, because they attempt to
not only improve upon the print approach, but also
use secondary-oral techniques. In other words,
while a student can still be presented with material (in different formats) using the print model,
the introduction of secondary-oral methods can
be used to improve the overall learning experience. Using the latter, there is the opportunity to
work and learn collaboratively, explore, analyze,
engage in discussion, and otherwise learn in
new and innovative ways (Ferris & Wilder, 2006;
Wallace, 2005).

INSTANT MESSAGING (IM)

It is unlikely that there would be many college
students who are unfamiliar with the use of IM.
Allowing for interactive and real-time synchronous communications with instant response,
instant messenger is truly conversational in that
it allows for chat and communications between
both individuals and groups. The major instant
messaging systems in use include AOL (AIM),
MSN Messenger, Yahoo! Messenger, and ICQ.
IM is a means for users to chat and communicate in real-time. While originally the domain
of personal users, over time the unique benefits
and effectiveness of this medium were realized,
and IM started to become accepted as a form

Innovative Technologies for Education and Learning

of communication in businesses (particularly

high-tech firms), and now has been studied and
tested as an educational tool (Kinzie, Whitaker,
& Hofer, 2005).
The important features of IM include both its
synchronous nature and its ability to support both
chat and phone-like interaction. While real-time

interaction allows for rapid communications to

occur, there is also no need to enter an interaction
space as with chat rooms. Instead, the main usage of IM is in one-on-one communications, which
can be more formally termed as a dyadic call
model, which more closely resembles phone call
interaction. It should be noted that even though

Table 1.

Description

Instant Messaging
Real-time communications that allow for informal communications to
be conducted easily and quickly
Availability and acceptance by students
Social presence (know the status of other users online)

Advantages

Real-time (synchronous) communications

Encourages collaboration
Reduces formality in communications
Distracted attention, especially in a classroom setting
Time waster that is not directed toward course content, but on personal discussions

Disadvantages

Expectations of 24-7 instructor access

Can be time consuming for instructors
Benefits are uncertain in classroom settings

Virtual office hours (instructor-student)

Educational Collaboration on group projects
applications Real-time class discussions
Mentoring
Courses with group projects and assignments
Course/
subject suitDistance learning support
ability
Theoretical
foundations

Active learning
Dual (verbal and visual) processing

1311

Innovative Technologies for Education and Learning

much of the communication is done between two

individuals, there are some systems that support
multiparty instant messaging.
Some of the salient features of IM include the
ability for users to see user details as to current
status (online, idle, away, out to lunch), and also
on a users changes in status (active, logged out,
etc.). Lists of users can be displayed on the screen,
so that contact can be made when desired. If a
chat is initiated, a special window comes up,
and the interaction can commence, provided that
both parties are online and willing to proceed.
The real-time nature of IM has resulted in the
technology being used for reasons aside from personal chat. In business, IM has become in some
industries an accepted form of communication.
A number of studies have concluded that instant
messaging is ideal for informal interaction. In
particular, the use of IM has been shown to be
helpful in cases where collaborative coordination
and problem solving is involved. Social bonding
and interaction, which is a component contributing to the success of more complex collaboration
situations, is also enhanced by using instant messenger technology (Nardi & Bradner, 2000).
An important difference between IM and email is the tendency for instant messenger interaction to be more casual and informal than e-mails,
which helps to bring about a more friendly
communication atmosphere. This may in part be
due to a reduction in the formalities that are typically involved when using e-mail or the phone. In
particular, IM has been considered more suitable
for such tasks as scheduling meetings, asking or
answering quick questions, and for other kinds of
tasks that are brief, require a prompt response, or
are less formal. It is perceived to be far simpler to
IM someone to ask a quick question, for example,
or to confirm a meeting or lunch, rather than to
e-mail or call (Nardi & Bradner, 2000).
It is also of interest that IM communications tend to be more flexible in terms of their
uses (everything from task-related questions
to a new joke), and can allow for greater ex-

1312

pressiveness in terms of emotion, humor,

and personality (Nardi & Bradner, 2000).
Another interesting aspect is what Nardi and
Bradner (2000) refer to as outeraction, which
focuses on the processes associated with IM.
These include conversational availability, communications zones, intermittent conversations,
awareness, and conversational progress/media
switching. IM is useful in certain communications
situations, since it tends to be less disruptive and
interrupting, while at the same time a users availability is more clearly known (scanning buddy list
status, for example). It is also a convenient means
for setting up more formal interactions, such as
arranging a conference call (media switching).
Intermittent, dispersed communications can be
conducted over a longer period of time, which
includes interruptions. Another benefit includes
the knowledge that others are there and available, even if not currently in chat mode; however
there is always the opportunity to make contact,
whether through IM or a different form of communications.
While some educators may scoff at and even
express criticism at the thought of instant messaging as a viable educational tool, others believe
there is potential in the medium.
In terms of educational uses for IM, they are
being explored and tested. Clearly, IM not only
allows students to collaborate more effectively
on homework assignments and projects, but also
helps to maintain a closer social network between
students, which could have a positive impact on
learning. In addition, if IM is carefully targeted
and focused toward the material or lecture topic
in hand, the use of IM may actually help and
stimulate deeper and more active learning.
On the other hand, it has been hypothesized
that the distraction of working on various other
tasks in parallel with IM, known as distracted
attention, may have a negative impact on learning (Hembrooke & Gay, 2003).
Active learning (Grabinger, 1996) and dual
(verbal and visual) processing (Clark & Paivio,

Innovative Technologies for Education and Learning

1991) are at work here. It could be said that using

IM to encourage greater discussion and reflection
on the course contents would be likened to the
use of discussion boards; however, since IM is a
real-time technology, the interaction is conducted
during the lecture or class, not afterward. Some
studies have reported positive effects and student
satisfaction from IM being used to discuss course
subjects in real-time (Guernsey, 2003).
A study by Kinzie et al. (2005) examined the
use of IM during classroom lectures and found
that while the general idea of using IM online
discussions was positively received, the actual
process and experience of using IM to conduct
discussions during class lecture sessions was
not found to be less than a positive experience
by both teachers and students. It was suggested
that the difficulties of multitasking and dividing
ones attention between the lecture and instructor,
doing the IM discussion, contributed to the lack
of satisfaction with the process.
Burke (2004) used instant messaging as a medium for creating course diaries in three different
mathematics courses. IM was chosen since it was
thought to be popular, widely used by students,
and considered more fun, so there was some
hope that this popularity would transfer over to
greater and more enthusiastic usage by students.
In fact, the choice was made to use IM over a
seemingly more suitable choice, blogs. A bot was
created that would retrieve IM diary entries from
students and store them in a PostgreSQL database,
and there was also a program set up to display
diary entries from each student, viewable by both
the student and the instructor. The main finding of
the study was that the IM media was not ideally
suited for all kinds of courses, especially those
that involved creating longer portions of text,
or involved diagramming. Error detection and
recovery was also not that well developed, and
also there was a need for better editing tools.
In summary, while instant messenger can be
appropriate for various applications, in particular
for information communications in a business

setting, the results from educational studies appear to be mixed, with both positive and negative
effects noted. While there seem to be advantages
to real-time communications between students,
between students and instructors, and also between groups working on a project, it appears
that there are problems and limitations if the
technology is used in a classroom setting. The
challenges of focusing on a class lecture, together
with maintaining a conversation online, seem to
be a problem that has not yet been resolved. In addition, while instructors can often establish closer
relationships with students using IM, there is also
the problem of unreasonable student expectations
of continuous teacher access, which may not be
present if IM was not available as an option. In
connection with this, using IM for student help
can result in a greater time commitment, since
sessions can become lengthy with many questions and responses being sent back and forth.

BLOGS (WEBLOGS)
Blogs started as a means for expressive individuals to post online diaries of themselves. Complete
with text and photos, these logs were essentially an
individuals online narrative or diary, with events,
stories, and opinions. While its original use was
for personal expression, recently its effectiveness
as a tool for education has been discovered, including its use as an extension of learning logs,
which are created online (Barger, 1997). One of
the earliest blogs, as we know and use them today,
was Dave Winers Scripting News, which was put
online in 1997. While the use of Weblogs can be
considered generally new, the concepts of keeping
a log or learning log is not.
The concept of learning logs has been in use
since before the advent of the Weblog. The concept
of this is to enable someone to document his or her
learning, and also to do some critical reflection
(Fulwiler, 1987) and self-analysis. The use of a
learning log or journal is related to action research

1313

Innovative Technologies for Education and Learning

learning strategies (Cherry, 1998) and attempts

to link previous knowledge and new information
learned. Blogs are a natural extension of learning logs/journals in that they are electronic and
can be made available (published) more easily
(Armstrong, Berry, & Lamshed, 2004).
The use of electronic Weblogs as educational
tools offers the benefits of increased information

sharing, simplified publication of information,

and improved instructor monitoring and review
(Flatley, 2005; Wagner, 2003). The use of blogs has
been expanding, as Perseus Development reported
that there were some 10 million blogs in 2004, and
the number is ever increasing (Nussbaum, 2004).
The growth in this area is expected to increase
in the years to come.

Table 2.

Description

WEBLOGS (BLOGS)
A technology that allows a sequence of entries (online diary, journal) to be posted and published online
Reflection and critical thinking are encouraged

Advantages

Disadvantages
Educational applications
Course/
subject
suitability

Theoretical foundations

1314

Authenticity through publication

Social presence
Development of a learning community
Active learning encouraged
Ability to receive and respond to feedback
Controlled primarily by blog author
Editing/modifications not open as in a wiki
Online learning journal
Problem solving/manipulation space
Online gallery space (writings, portfolio, other work)
Peer review exercises
Writing courses
Foreign language courses
Research seminars
Activity theory
Guided discovery
Cognitive scaffolding
Receptive learning
Social cognition
Community practice
Communities of inquiry

Innovative Technologies for Education and Learning

Blogs can be defined more formally as being

frequently updated Web sites consisting of dated
entries arranged in reverse chronological order
(Walker, 2005) and can take several forms, including the personal diary/journal, knowledge-based
logs, and filter blogs. The first, an electronic, online
diary of ones life events and opinions, is probably
the most common. The online diary/journal blog
is one that, being on the Internet, is public, as opposed to the traditional (typically paper) diaries,
which are generally kept private. It should come
as no surprise that there are many different online
diary/journal blogs that are currently online, where
one can find out details, often much more than
one might want to know, about someones life
and thoughts. Personal blogs form the majority of
the blogs that are currently online and available,
which make up roughly 70% of all the blogs in
existence (Herring et al., 2003).
The second type (knowledge-based) captures
knowledge and places it online in various formats.
The third type (filter) attempts to select, rate, or
comment on information contained in other sites
(Herring et al., 2004).
There are software packages that are designed
to help users create blogs, including Blogger,
Xanga, Blurty, and MovableType. While the basic
features of most blog software emphasize the creation of blog pages, some of the more sophisticated
ones offer the capability to track readership, see
who followed what links, add photos, and set up
more advanced structures. When online, blogs
can range from being totally public (listed in the
blog service directory), to being unlisted but
still open to being highly restricted (passwordprotected).
Blogs are also interesting and unique in that
they are not merely online versions of paper diaries and journals. Rather, as a communications
medium under the control of the main writer
(author), it is reflective of the fact that an audience is watching and listening. What is put on a
blog is not merely a one-sided set of thoughts and

reporting of events; there can also be responses

to feedback and reactions from the viewing
audience. Therefore, blogging is considered a
highly social activity, rather than a personal one.
In fact, recent work has indicated that the momentum for creating, and also updating a blog, came
about as a result of encouragement from friends
and the viewing audience (Nardi et al., 2004). In
addition, blogging can have negative repercussions when posted information is perceived to be
confidential, proprietary, or improper. In some
cases, employees posting what was considered
by their employers as confidential information
can cause problems.
Blogs do not, in general, exhibit a free-flow of
information between the blogger and the outside
audience. While feedback is often requested,
received, and desired by the blogger, the level
and quantity of feedback from readers is generally limited compared with the output from the
blog writer. In addition, while blogs may have
sections where hyperlinks are mentioned, the
number of hyperlinks in blog pages is frequently
not very large (Schiano, Nardi, Gumbrecht, &
Swartz, 2004).
More formally, Weblogs can be considered to
be objects motivating human behavior, which is
related to activity theory. Activity theory states
that there are objects that have motives that respond
to a human need or desire, and that they manifest
a persons desire to accomplish that motive (Leontiev, 1978; Vygotsky, 1986). The objects that
connect bloggers to their own social networks
include letting people know what is happening in
their lives, voicing opinions, asking for feedback,
and letting off steam about certain challenges or
difficulties currently being experienced, to name
a few (Nardi et al., 2004).
Blogs have been categorized by Krisnamurthy
(2002) as being categorized into four different
types, along the dimensions of individual versus
community, and personal versus topical. A blog
can therefore range from being very individual

1315

Innovative Technologies for Education and Learning

and personal, all the way to being open to the

community, however very focused on a particular
topic.
The acceptance of blogs for educational purposes is gaining interest, with one university, the
University of Maryland, attempting to implement
blogging software campus-wide (Higgins, Reeves,
& Byrd, 2004).
In addition, the educational uses of blogs take
advantage of their ability to encourage expression
and the development of online relationships. Blogs
allow for learning and interaction to be more
knowledge-centered, especially if the assignments are structured in the format of encouraging
feedback and input from the instructor and outside
experts. Blogs also allow students to gain a better
understanding of a subjects knowledge domain
(Glogoff, 2005). As an example of this type of
blog-enhanced class structure, students might be
provided with a Weblog from which to complete
certain course assignments. After researching the
indicated subject, the student would present the
new information by publishing it to the Weblog.
The Weblog would constitute the students assignment, which would then be subject to review
and critique by not only the instructor, but also
by other students in the class. Supplementing
this could be discussion boards, where threads
would be devoted to discussions on the Weblogs
created by the students. This kind of assignment
and interaction would be especially useful for
both hybrid and fully online distance learning
courses (Glogoff, 2005).
There are other benefits of Weblogs. These
could be expressed using the learning theories and
concepts of guided discovery, directive learning,
receptive learning, and social/community-centered instruction.
Guided discovery allows for the exploration
and study of a certain topic, which is then followed
by assignments that emphasize the synthesis of
information. In effect, a student can be asked to
research an area and construct knowledge using

1316

the Weblog as a medium. Part of the assignment

goes beyond merely explaining or presenting the
material, and asks for the application of the concept using a real-world situation. The ability for
students to post and make comments about other
students blogs provides an atmosphere of interactivity and collaboration. One of the advantages
of using blogs together with guided discovery is
that it encourages the use of cognitive scaffolding,
where students would approach learning (together
with the use of blogs and interaction) by repeatedly
seeking information, reflecting and thinking about
what has been learned, and then going back and
obtaining more information, so as to build upon
and dig deeper into the subject area. This can result
in a more active and productive form of learning
(Betts & Glogoff, 2004; Glogoff, 2005).
Directive learning, where responses from
students are followed by prompt feedback from
instructors, can also be supported using blogs. In
this case, students would not only use a blog to
submit assignments, but also to review instructor
feedback. In additional to feedback, there would
be opportunities for the instructor to ask additional questions, in effect, to encourage further
exploration and drilling down into the subject
(Betts & Glogoff, 2004; Glogoff, 2005).
Receptive learning is where instructional modules are presented that focus on certain broader
areas, from which certain sub-areas within these
are highlighted for a student to research and report
on. Generally, the areas are contained within a
designated theoretical context (Betts & Glogoff,
2004; Glogoff, 2005).
Social/community-centered instruction is
a logical component of educational work using
blogs, and in particular the use of peer and social
interaction as a part of the learning process. The
use of blogs functions as an easily accessible medium for students to present their findings (and to
be read by others) and also to integrate not only
the information presented, but also related links
and references to other resources. This form of

Innovative Technologies for Education and Learning

interaction helps to encourage further exploration

by students. A blog-based discussion can then be
continued by conducting peer reviews of other
students blogs, which may include commentary,
critique, the posing of questions, and opening up
the way for further inquiry. The ability to share
and benefit from the findings of other students (and
to explore further) is another important outcome.
The theories of community practice (Snyder,
2002), social cognition (Vygotsky, 1978), and
communities of inquiry (Lipman, 1991) provide
support for the blog-related techniques mentioned
previously.
Ducate and Lomicka (2005) discuss their
experiences in using Weblogs to support foreign language classes. Weblogs help the foreign
language student to learn during the process of
reading, and then creating blog entries. Students
can learn by reading blogs that are written in
the new, target language, including learning
new vocabulary, checking out links and further
information on words, and learning associated
cultural information. The reading and absorption
of blogs on the culture associated with the target
language, including literature and lifestyles, all
would contribute to the learning process.
Another approach would be to have students
maintain blogs written in their new, target language, and then the goal would be to seek commentary and critique on these blogs by others in
the class. Yet another innovative method might
be to share blogs with other classes studying the
same language, and for students to read and comment on each others postings. In the case where
students travel to a country where the target language is spoken, the compilation of travel blogs
would be a useful learning tool as well (Ducate
& Lomicka, 2005).
Wang and Fang (2005) looked at whether the
use of blogs encouraged or enhanced cooperative
learning in an English rhetoric/writing class taught
in Taiwan. The main premise was that blogs can
encourage students to spend more time working

within a class community and can benefit from

a greater sharing of contributions and inputs. In
general, cooperative learning benefits can be divided into three different types: formal, informal,
and base groups. Formal cooperative learning is
where the instructor explicitly provides course
materials and assignments to a group and then
observes the students learning processes. When
the instructor provides information more generally
(such as detailing how to use a blog for course
assignments) and then lets the group work out
their own methods for handling an assignment,
that is known as informal cooperative learning.
When a learning-oriented group is maintained for
an extended period of time, such as throughout a
semester, then this form of cooperative learning
is known as a cooperative base group (Johnson
& Johnson, 1998; Johnson, Johnson, & Holubec,
1991). The study, run over the course of a semester,
found that the use of blogs contributed not only
to cooperative learning in general, but also to
autonomous learning. Autonomous learning is
focused on how much students take responsibility for their own learning, and also develop selfconfidence in the task or skill (Wenden, 1991).
The use of blog technologies was also found to
help improve information processing, learning
self-evaluation, and effective time management
(Wang & Fang, 2005).
Martindale and Wiley (2005) also used blogs
in their courses and looked at two cases of the
impact of this technology on teaching and learning. Martindale taught a doctoral-level course on
instructional design and technology. In it, students
were introduced to blogs and used them throughout
the course, which overall tended to promote higher
levels of quality in their course work. Blogs were
used to post ideas and abstracts of their projects,
and also to place links for relevant research papers
and Web-based resources. The end result was a
course knowledge base that represented the
cumulative output of the students in the course.
The blogs were also used for article critiques,

1317

Innovative Technologies for Education and Learning

which were an integral part of each weekly class.

Students were given access to the blogs of other
students and were able to offer feedback.
Wiley taught two different courses, one on
online learning research, and the other on online
interaction culture. Both included the use of blogs
as a supporting technology, the first employing a
general course blog where information about the
course, student assignments, and class updates
and student/instructor interaction exchanges were
posted on an ongoing basis. In the second, blogs
were used to discuss experiences using different
online communications technologies, causing
students to become highly engaged, resulting in
passionate discussions and detailed commentaries
posted to the blogs, far exceeding the level and
depth of feedback that was expected (Martindale
& Wiley, 2005).
In summary, blogs can be useful for educational purposes, particularly where there is the
need to encourage and stimulate critical thinking
and reflection on a work, concept, or idea. The
submission or publication of a document or text
as a blog can then lead others in a class to review
and comment, setting the stage for greater analysis
and study. In particular, blogs are suited to writing courses, where a text can be analyzed and
critiqued, or for a situation where a peer review
is desired. The use of blogs is also ideal for the
situation where someone keeps an online journal
of ones learning, or wants to display her or her
work to an audience. The blog approach is also
considered useful for group study of a certain
focused problem or case, such as those used in
graduate courses and seminars.

WIKIS
Yet another technology, known as the wiki, has
emerged, which allows for improved collaboration
compared with Weblogs. While the major emphasis of Weblogs is the creation of a set of pages

1318

and documents primarily by a single individual,

the strength of a wiki is the ability for numerous
interested readers and users to express ideas online, edit someone elses work, send and receive
ideas, and post links to related resources and sites.
As a result, wikis go a step further and allow for
greater collaboration and interactivity (Chawner
& Gorman, 2002).Wikis have been found to have
value for educational purposes, and their use has
begun to be integrated into a number of university
courses, in particular (Kinzie, 2005).
The term wiki comes from the Hawaiian
word wikiwiki, which means fast. The technology is computer-based and can be generally
described as a knowledge sharing and creation
system that has as its basis a set of Web pages,
which can be created and updated on an iterative
and collaborative basis, and is in many ways a
form of groupware. A wiki is designed to run
on the Internet and World Wide Web, uses the
HTTP protocol, and resembles traditional Web
sites in terms of its underlying structure. Some of
the benefits of wikis include the ability to easily
create pages (using a simplified form of HTML
or basic HTML) and the ability for a document
to be authored collaboratively and collectively. In
particular, simplicity is the key to wikis, and wiki
pages have been designed to be easy to create,
(simpler than the process of creating standard
Web pages). One of the better-known examples
of a wiki is www.wikipedia.org, which is an
online encyclopedia with entries authored and
edited by different persons worldwide, and in
several different languages as well. In essence, it
is an online information resource that is authored
by interested and knowledgeable persons from
around the world.
Wagner (2004) developed a set of design
principles that relate to wikis. These are the
principles of open, incremental, organic, mundane, universal, overt, unified, precise, tolerant,
observable, and convergent wikis. Open means
that anyone can edit a wiki, creating an open

Innovative Technologies for Education and Learning

source environment for the sharing of knowledge.

Incremental means that new pages can be added,
even if they do not yet exist. Organic means that
the information can be continuously evolving, as
changes and edits are made. Wikis are mundane
because they involve the use of a simplified set
of essential commands. The design of wikis is
also universal, meaning that writing and editing
is a combined activity, formatting is related to
input (overt), page names are not context specific
(unified), and pages are generally named with
some precision (precise). Wikis should be tolerant of error; activity should be observable by all,
and duplications are undesirable and should be
deleted (convergent).
There are a number of software programs that
enable the effective creation of wiki pages, includ-

ing TikiWiki, TWiki, and Pmwiki. These allow for

the effective creation, modification/editing, and
management of wikis, including creating pages,
creating links, formatting, and feature modules
(discussion forums, photo pages, download areas,
etc.) (Chawner & Lewis, 2006)
Wikis are set up to allow for easy collaboration, and more specifically, editing. Rather than
passively reading a passage of text or related
information (which may include graphics, multimedia, hyperlinks, etc.), a reader of a wiki can
also take on the role of a writer, making changes
to the text (re-organizing, editing, re-writing, and
marking up) at will. In essence, the document is
open to changes by a collaborative community,
which allows for the secondary-oral model in
education to be applied.

Table 3.

Description
Advantages

Disadvantages

Educational
applications
Course/subject suitability
Theoretical
foundations

WIKI
A technology that allows for material to be easily published online,
and also allows open editing and inputs by a group
Contributions and editing by a group
Open access to all users
Collaborative
Lack of organization and structure may result in an unmanageable
wiki
Tracking of contributions and modifications can be difficult
Quality control
Collaborative writing/authoring
Group project management
Brainstorming activities
Knowledge base creation (knowledge management)
Knowledge management
Writing
Group work in courses
Conversational technology
Constructivist learning tool

1319

Innovative Technologies for Education and Learning

One reservation on the part of educators to

embrace wikis is the fact that wikis are designed
to allow for open and free access and editing by
all members of a community. As a result, if
improperly managed, a wikis content can become an unreliable, inaccurate, or biased source
of information due to its largely unmonitored
format. There is also the issue of having little or
no quality control, resulting in a wiki not being
trusted by its readers and users. An example of
this was the controversy over the accuracy and
reliability of Wikipedia, in the case of John Seigenthaler, in which the subject alleged that false and
incorrect statements were posted in his biography
(Seigenthaler, 2005). However, other studies have
attempted to prove that Wikipedia was overall an
accurate information source. One article reported
that after analyzing a set of Wikipedias science
articles, they were judged to be, by the British
journal Nature, as reliable as the Encyclopedia
Britannica (BBC News, 2005).
Wikis are useful for education in that they
help to promote student participation and also a
sense of group community and purpose in learning. Indeed, an important element of this is the
relaxed sense of control over the content, allowing
students to have a greater role in managing its
focus and direction.
Wikis are not all the same, and there is significant diversity between various forms and
implementations of wiki systems. In fact, it
could be debated what features truly characterize a true wiki. The features inherent in most
include the ability for users to both read and edit
information, without the need for security or access restrictions. The emphasis is on simplicity,
and the informal, never finished nature of wikis,
which may constitute the contributions of multiple
authors, is another key characteristic. While the
emphasis of many wikis is on simplicity and a
lack of access restrictions, that does not mean that
all wikis work this way. In reality, there can be a
continuum of features from simple to complex. At

1320

the complex end of the scale can be capabilities

for security/access restrictions, complex organizational structures, and for integrated with content
management systems (Lamb, 2004).
Now that the strengths and weaknesses of
wikis have been established, it would be useful to examine the educational applications of
wikis. In general, the most suitable applications
are those that take advantage of the wikis free,
open structure. As such, the use of wikis as
discussion/bulletin boards, brainstorming tools,
and online sketchpads is appropriate. Meeting
planning is another viable application area, in
that the organizer can start with a preliminary
agenda, from which the other participants can
then add their own additions or make modifications or comments.
An important application area for wikis has
been identified in knowledge management (KM).
The use of wikis for knowledge management may
allow for an improvement over existing systems
and technologies. Currently, with existing KM
systems, there does exist a number of bottlenecks
relating to knowledge acquisition, namely acquisition latency, narrow bandwidths, knowledge
inaccuracy, and maintenance traps. Basically,
these knowledge acquisition bottlenecks result
from a time lag between when the knowledge is
created, and then distributed. In addition, there
are the problems of limited channels of knowledge
input, possibilities of erroneous information being
received, and also the difficulties of maintaining
the knowledge base as it grows larger (Land, 2002;
Wagner, 2006; Waterman, 1986).
The use of wikis to elicit a bazaar approach to
knowledge management, rather than a cathedral
approach, is proposed as a better alternative. These
terms are derived from software development,
whether the bazaar allows for more continuous
and open access to code (or information), as opposed to the cathedral approach where access is
only provided on certain (release) dates to certain
persons. The difference between the cathedral

Innovative Technologies for Education and Learning

(closed), sources of knowledge acquisition management and bazaar (open) could be illustrated
by the difference between encyclopedias that are
created by a single firm, such as Encarta or the
Encyclopedia Brittanica, and those that obtain
information from readers and users, such as the
well-known Wikipedia.
The emphasis therefore is on teamwork,
continuous review and testing, and the development of conversational sharing (Wagner, 2006).
Inherent in the workings of wikis is support
for an open, collaborative environment, where
many people can contribute to the development
of knowledge instead of being limited to a set of
experts. It appears that conversational knowledge acquisition and management are appropriate
for wikis (Cheung, Lee, Ip, & Wagner, 2005).
As for educational applications and KM, a study
by Raman, Ryan, and Olfman (2005) examined
the use of a wiki to help encourage and support
collaborative activities in a knowledge management course. More specifically, using wikis in the
course helped to encourage openness and better
sharing and updating of knowledge bases. Manyto-many communication is supported, and the
persistence of the created pages formed the basis
of a knowledge repository. In short, the impact
of easy page creation and improved updating and
editing, together with effective maintenance of
knowledge histories, were seen as positives (Raman et al., 2005; Bergin, 2002).
Activities in the KM course activities included
group article review assignments, answering questions about sharing knowledge and uses of the
wiki technology, and also creating a wiki-based
knowledge management system. Students were
asked to create, update, refine, and then maintain
a class knowledge management system. In terms
of these experiences, while the use of the wiki
technology was generally viewed positively, feedback received indicated that, since the goals of
using the wiki were not made clear, using one was
perceived to be counter-productive. More specific
guidance on goals and objectives, a clearer system

structure, and advanced training were suggested

as ways to make the wiki a more effective educational tool. The availability of too many features
made the task of doing the course activities more
difficult, since much time was spent learning the
various features rather than focusing on the task
at hand. A simpler, less feature-rich version was
therefore preferred (Raman et al., 2005).
Another popular application of wikis in the
classroom is in supporting writing courses. The
use of this technology can help to foster the
impression that writing is fun, and that there
can be a shared and collaborative side to writing, revising, and commenting on written work.
In other words, the technology can benefit not
only the writing and editing process, but also in
bringing about an awareness that writing is being
done for a specific audience.
An example of the use of wikis in English
is the Romantic Audience Program at Bowdoin
College, where students used a wiki to discuss
and examine Romantic literature, focusing on
poetry, poets, and related topics. The technology
was used to elicit discussion and analyses by the
group, encourage elaboration on areas where further study or insight was sought, and to seek out
linkages to additional sources and commentary.
Another project is Auburn Universitys Wikifish,
which was created by one school within the university, where questions are posed, and opinions
and comments by all are encouraged.
Difficulties encountered in using wikis for
education include the difficulty of tracking the new
pages and contributions made by students, since
modification can be made to another students
text without specifically identifying the author or
editor. As a result, it can be difficult to monitor,
and properly attribute, what contributions were
made by whom, on a particular page. A proposed
solution to this was an instructors use of categories and topics, requiring that students link to and
add, rather than simply modify, the contributions
of other students. Another issue was how much
of a balance in terms of the tradeoff between

1321

Innovative Technologies for Education and Learning

total freedom, and total control, was ideal. Since

a clear benefit of a wiki is the emphasis on free
expression and on spontaneous inputs, reducing
this may have a negative effect on open interaction
and student contributions (Lamb, 2004).
An interesting application of the use of wikis
in the classroom was the work by Bruns and
Humphreys (2005), where wikis were used in a
New Media Technologies course to collaboratively
develop entries for an online wiki encyclopedia
called the M/Cyclopedia of New Media, an actual live wiki resource made available online.
The effort to develop entries involved over 150
students spanning six classes in the program.
Feedback indicated that while students had little
difficulty with using the wiki system, obstacles
came about more with the writing and format of
the encyclopedia entries.
Another study examined the use of wiki
technology in German-speaking communities
in Switzerland (Honegger, 2005), while Desilets, Paquet, and Vinson (2005) looked at how
usable wikis are and found that the main problems encountered related to the management of
hyperlinks. Wikis were also examined from the
perspective of structure: how did the use of certain templates affect the appearance and layout
of wiki pages? The results suggested that they
are useful in certain circumstances and could be
helpful overall to end users (Haake, Lukosch, &
Schummer, 2005).
In summary, wikis are best suited to course
and activities where there is a document, text, or
other project to be worked on jointly by a class
or group. In a sense, it is a tool for collaboration,
and a form of groupware. The compilation of a
class or group report or project, the creation of a
knowledge base, or brainstorming sessions appear to be viable applications. The free and open
structure, however, can fall prey to disorganization
and degradation in quality, and so it is important
to have safeguards and procedures in place to
ensure an effective result.

1322

Podcasts
While the terms pod and podcast at first
mention might evoke visions of Invasion of the
Body Snatchers, for most tech people in the
know, the reference to Pod is almost certainly a
reference to Apples popular and ubiquitous iPod.
However, podcasts are in actuality not what
their name might imply them to be. A podcast,
a combination of iPod and broadcast, neither
refers to a technology specifically requiring an
iPod, nor broadcasts information to users. Instead,
podcasts are multimedia files (typically audio or
video) that are downloaded to users on a subscription basis. Because of the potential confusion due
to the use of the word pod, some have called for
the letters to mean personal option digital or
personal on demand, rather than iPod.
Podcasts can be played back on any device or
system that can play digital audio (typically MP3)
or video files, and are not broadcast to a large audience, in the way that television, radio, or spam
e-mails are sent. Instead, they are sent to users who
have specifically subscribed to a podcast service,
and as such, files are automatically downloaded
to the users computer when they are ready and
available. In addition, podcast files are generally
not streamed (as video is streamed), but rather
are downloaded for later playback (Lim, 2005;
Lum, 2006). Podcasts are delivered to subscribers through the use of RSS or RFD XML format
media feeds, rather than more traditional forms
of downloading (Descy, 2005).
Podcasts are considered to be a viable educational tool for several reasons. First, because of
the popularity and wide use of devices such as
the iPod and similar units, it would seem like a
good medium from which to distribute educational
materials. Secondly, the ease with which information can be retrieved and accessed makes this
a good choice for students, who are using these
devices on a regular basis for music and should
have few technical difficulties or learning curves
(Lum, 2006).

Innovative Technologies for Education and Learning

There are multiple facets to podcasts. First,

there is the consumption perspective, where
someone downloads the podcast and then listens
or views it. This involves subscribing to a service
(or enrolling in a course), finding the relevant
file, and then downloading and playing it. Alternately, someone can actually create podcasts; an
instructor can produce lessons for students, or
students can produce an assignment in the form
of a podcast file (Lum, 2006).
Education is one area where the power of the
podcast has been used in various ways. At Duke
University, orientation material was distributed
as podcasts, loaded onto iPod units, and given

to students in its 2004 incoming freshman class.

The units were provided not only for orientation
purposes, but also for use in playing podcast
lectures when the students take certain courses at
the university. At Mansfield University, students
were sent podcasts about various student life issues, and at Arizona State University, President
Michael Crow used podcasts to deliver messages
to the university community (Lum, 2006).
While there appear to be sound reasons for
using podcasts, there is also a theoretical basis
behind the use of podcasting. This is based on
cultural-historical activity theory (Engestrom,
2002) and is based on the fact that podcasting

Table 4.

Description

Advantages

PODCASTS
The ability to create audio (and other media) based files to be distributed
on a regular/subscription basis to users; these can be easily retrieved and
played back on handheld devices, computers, and through other means
Allows for information to be retrieved and played back on widely available, ubiquitous devices
More suitable to auditory and visual learners

Disadvantages

In consumption (playback) mode, does not directly support collaboration

Educational
applications

Recorded class lectures

Case-based instruction
Guest lectures in the form of podcasts
Supplemental course materials
Support for distance learning courses
Subject matter lends itself to auditory format

Course/subject suitability
Theoretical
Foundations

Is not inherently interactive in nature

Cultural-historical activity theory

1323

Innovative Technologies for Education and Learning

can be considered a tool that can be used to help

learners to better interact with or understand a task
and its environment. Vygotsky (1978) argues that
the effectiveness of podcasts rests in its linkage
between social/cultural influences present in the
environment and the cognitive development of
the learner. Expressed another way, the concept
is that since so many student have access to iPods
and MP3 players, it would make sense to explore
the viability of using such a device for learning
and educational purposes.
Lim (2005) discussed experiences in using
podcasts for teaching geography. Because of
the nature of the subject, it was found that video
would have a greater impact and result than audio.
Students were asked to submit assignments to be
created and submitted as podcasts. Overall, this
helped to bring about satisfaction and interest in
terms of the subject. Ractham and Zhang (2006)
looked at the potential for podcasts in a classroom
setting. While there are the obvious benefits in
terms of being able to distributing class materials,
there also are benefits in terms of contributing
to social networking in the class and continuing
a flow of academic knowledge. Class discussions, announcements of research activities and
conferences, and also campus activities could be
distributed as podcasts. Review materials could
be distributed effectively on an automatic basis
to interested students. In addition, the ability for
students to create podcasts to be distributed to
others would also be a new means of submitting
assignments or expressing creativity.
Podcasts, unlike IM, blogs, and wikis, offer
a greater emphasis on providing engaging auditory and visual course materials to students,
rather than on collaboration and group work.
While not generally a substitute for traditional
lectures and knowledge presentation, they offer
the benefits of easily accessible and digestible
course material. Whether it is excerpts from class
lectures, highlights from a guest speaker, or an
oral test review, the use of podcasts provides a
means by which students can obtain and easily

1324

receive course-related information. In addition,

it also provides a means by which students can
express and present their work, which can then be
published and distributed in podcast format.

Discussion and Conclusion

The face of education, whether online, hybrid,
or classroom, is constantly changing, and it is
important for educators to stay abreast of the
many opportunities and possibilities that are
available.
In this article, several technologies, generally
termed as conversational technologies due to
their interactive and collaborative nature, were
discussed in detail, together with their capabilities, benefits, and educational applications.
Relevant research and case studies as they relate
to classroom and educational applications were
discussed.
In general, the tools discussed here fall into
the class known as conversational technologies
or constructivist learning tools. As such, they
emphasize student interaction, group learning,
and collaboration, rather than the more traditional
classroom mode. In light of this, they are more
suited to educational courses or environments
where the emphasis is on student communication,
where students have access to technology, and
where creative output and thinking is encouraged.
In the situation where a course is primarily
lecture-based, or is mainly concerned with the
delivery of factual or conceptual information,
these tools may have limited applicability. The one
application that may be helpful in this case may
be for interaction to be extended outside of the
classroom, through the use of instant messenger,
or for supplemental materials to be distributed
as podcasts.
Since each of the tools has its own characteristics and suitable applications, it would be up to the

Innovative Technologies for Education and Learning

educator to select those that are most appropriate

to ones course and activity.
Instant messenger, which is commonly used
by students for personal use, has found its way not
only into the business community, but also into
the classroom, because of its strengths in terms
of informal communications that are conducted
real-time. There are some mixed results regarding the use of IM for use for education; benefits
are claimed by some, but there are limitations
as well. The use of IM would best be employed
in situations where group work, student communication, and real-time discussion would be
helpful. However, it should be used cautiously,
since it can be distracting, and students may end
up carrying on personal, rather than course-related
discussions.
Both blogs and wikis have been hailed as
useful tools for education, and the specific advantages and disadvantages of each are noted
and discussed. Blogs tend to be more one-sided,
with an author presenting his of her information,
with generally limited input from the readers and
public. However, the use of the technology has
been used effectively to promote information
sharing and to support writing courses. The use
of blogs to support online learning journals, class
article/peer reviews, creating online portfolios
and galleries, and for solving a specific problem
or case would be advantageous. It would also appear to be a good medium for graduate research
seminars, where papers and studies are analyzed
and critiqued.
Wikis, which allow freedom in creation and in
editing and enhancement by others, are especially
useful in collaborative situations where the input
of an entire group is desired instead of a single
persons contribution. In the classroom, wiki support for collaborative writing and group activities,
where students contribute to the creation of a common result, would be useful. In general, any kind
of learning activity that involves the collection and
maintenance of knowledge or information may
benefit from the use of wiki technology.

The use of podcasts, which may include audio

or video, is growing in popularity and is being
used for delivery of information to subscribers on
an automatic basis. Educational podcasts, both for
the delivery of audio and video-based knowledge
to students, and also as a new medium for the
creation and presentation of assignments, appear
to have potential. While podcasts are useful as
a means for publishing and distributing files of
multimedia-based class materials, there also exists
the potential to create new podcast content, both
for educators and as a course activity.
Clearly, the use of these new conversational
technologies is allowing for the enhancement and
continued evolution of new and innovative forms
of support for teaching and learning.

FUTURE RESEARCH AREAS

Certainly, there are many benefits to the use of
conversational technologies and constructivist
tools for educational use. However, more research
needs to be done, not only in terms of identifying
additional types of applications and uses, but also
in terms of how to more effectively identify and
apply new approaches to learning with the aid of
these kinds of technologies.
Some of the broader research issues that can
be examined include measuring both learning,
and the perceived quality of education, depending on the specific technology or tool employed.
Are there measurable benefits to using a certain
technology in terms of the material learned, better
class performance, or more subjective factors? It
would also be useful to determine, particularly
when using blogs and wikis, the neutrality or bias
in the entries, and how much these contribute
to (or detract from) the material submitted for a
course assignment.
Other research areas are more technology
specific. It was mentioned earlier that wikis can
be a useful tool in knowledge management. The
application of wikis to effective knowledge man-

1325

Innovative Technologies for Education and Learning

agement deserves further attention, both in terms

of developing or adapting the wiki structure and
features to knowledge management uses, and also
for identifying various kinds of user interfaces
and functionality that would improve usability.
The establishment of wiki-supported communities of practice is one area where the tool could
possibly be useful.
Podcasting also has many areas that are ripe for
further investigation. There are issues that can be
explored in the areas of knowledge management,
collaboration, and the adoption of podcasts. Some
of the specific topics of interest include the management and sharing of knowledge using podcasts,
examining whether their use actually improves
learning, studying their effects on collaboration
and networking, and what the factors (or features)
are that would help to promote its use.
There also has been work on the psychological
aspects of distance learning and online courses
(Dickey, 2004), and a study of learners reactions
to using IM, blogs, and wikis, for example, would
yield insights into its appropriateness for its further
use in education. Does the use of these technologies contribute to the satisfaction of students, or
is more classroom face-to-face still better?
The realm of these new technologies is certainly ripe with a host of opportunities for both
interesting and meaningful research studies.

References
Ahn, J., Han, K., & Han, B. (2005). Web-based
education: Characteristics, problems, and some
solutions. International Journal of Innovation
and Learning 2005, 2(3), 274-282.
Armstrong, L., Berry, M., & Lamshed, R. (2004).
Blogs as electronic learning journals. E-Journal
of Instructional Science and Technology, 7(1).
Retrieved from http://www.usq.edu.au/electpub/ejist/docs/Vol7_No1/CurrentPractice/Blogs.htm

1326

Barger, J. (1997). Weblog resources FAQ, Robot

Wisdom Werblog. Retrieved from http://www.
robotwisdom.com/weblogs
BBC News. (2005). Wikipedia survives research
test. BBC News, December 15, 2005.
Beck, P., Kung, M., Park, Y., & Yang, S. (2004).
E-learning architecture: Challenges and mapping
of individuals in an Internet-based pedagogical
interface. International Journal of Innovation
and Learning 2004, 1(3) 279-292.
Bergin, J. (2002). Teaching on the Wiki Web.
ACM SIGCSE Bulletin, 1.
Betts, J. D., & Glogoff, S. (2004). Instruction
models for using Weblogs in e-learning. In Proceedings of Syllabus 2004. San Francisco.
Blood, R. (2004). How blogging software reshapes
the online community. Communications of the
ACM, 47(12), 53-133.
Bruns, A., & Humphreys, S. (2005). Wikis in
teaching and assessment: The M/Cyclopedia
Project. In Proceedings WikiSim05. San Diego,
CA.
Burke, M. (2004). Instant messaging and course
diaries. CCSC Northeastern Conference Proceedings, 19(5), 207-274.
Chawner, B., & Gorman, G. E. (2002), Wikis
are wicked. New Library World, 103(1182/1183),
483.
Chawner, B., & Lewis, P. (2006). Wiki wiki webs:
New ways to communicate in a Web environment. Information Technology and Libraries,
25(1), 33-43.
Cherry, N. (1998). Action research: A pathway
to action, knowledge, and learning. RMIT Publishing.
Cheung, K., Lee, F., Ip, R., & Wagner, C. (2005).
The development of successful online communi-

Innovative Technologies for Education and Learning

ties. International Journal of the Computer, the

Internet, and Management, 13(1).
Clark, J. M., & Paivio, A. (1991). Dual coding
theory and education. Educational Psychology
Review, 3, 149-210.
Cold, S. (2005). Using RSS to enhance student
research. ACM SIGITE Newsletter, 3(1).
Dearstyne, B. (2005). Blogs: The new information
revolution. Information Management Journal,
(September/October), 38-44.
Descy, D. (2005). Podcasting: Online media deliverywith a twist. TechTrends, 49(5), 4-6.
Desilets, A., Paquet, S., & Vinson, N. (2005).
Are wikis usable? In Proceedings WikiSim05.
San Diego, CA
Dickey, M. (2004). The impact of Weblogs on
student perceptions of isolation and alienation
in a Web-based distance learning environment.
Open Learning, 19(3), 279-291.
Ducate, L., & Lomicka, L. (2005). Exploring
the blogosphere: Use of Weblogs in the foreign
language classroom. Foreign Language Annals,
38(3), 410-421.
Efimova, L., & Fiedler, S. (2004). Learning webs:
Learning in Weblog networks. In Kommers,
Isaias, & Nunes (Eds.), Proceedings of the IADIS
Intl. Conference on Web Based Communities 2004.
Lisbon: IADIS Press.

Fisher, M., & Baird, D. (2005). Online learning

design that fosters support, self-regulation, and
retention. Campus-Wide Information Systems,
22(2), 88-107.
Flatley, M. (2005). Blogging for enhanced teaching
and learning. Business Communication Quarterly,
68(2005).
Fuchs-Kittowski, F., & Kohler, A. (2002).
Knowledge creating communities in the context of work
processes. SIGGROUP Bulletin, 23(3).
Fuchs-Kittowski, F., & Kohler, A. (2005). Wiki
communities in the context of work processes. In
Proceedings WikiSim05. San Diego, CA.
Fulwiler, T. (1987). The journal book. Portsmouth,
NH: Boynton/Cook.
Glogoff, S. (2005). Instructional blogging: Promoting interactivity, student-centered learning,
and peer input. Innovate, 1(5).
Grabinger, R. S. (1996). Rich environments for
active learning. In D. H. Jonassen (Ed.), Handbook of research for educational communications and technology (pp. 665-692). New York:
Macmillan.
Gronbeck, B., Farrell, T., & Soukup, P. (1991).
Media consciousness and culture. Newbury
Mark, CA: Sage.
Guernsey, L. (2003, July 24). In the lecture hall,
a geek chorus. The New York Times.

Engestrom, Y. (2002). Cultural historical activity theory. Retrieved from http://www.helsinki.

fi/~jengestr/activity/6a.htm

Haake, A., Lukosch, S., & Schummer, T. (2005).

Wiki templates. In Proceedings WikiSym05 (pp.
41-51). San Diego, CA.

Engstrom, M. E., & Jewitt, D. (2005). Collaborative learning the wiki way. TechTrends, 49(6),
12-15.

Hargis, J., & Wilson, B. (2005). Fishing for learning with a podcast net. Instruction and Research
Technology, Univ. of North Florida.

Ferris, S. P., & Wilder, H. (2006). Uses and potentials of wikis in the classroom. Innovate, 1(5).

Hembrooke, H., & Gay, G. (2003). The laptop

and the lecture. Journal of Computing in Higher
Education, 15(1).

1327

Innovative Technologies for Education and Learning

Herring, S., Scheidt, L., Bonus, S., & Wright,

E. (2004). Bridging the gap: A genre analysis
of Weblogs. In Proceedings of HICSS 2004. Big
Island, HI.
Higgins, C., Reeves, L., & Byrd, E. (2004).
Interactive online journaling. In Proceedings
SIGUCCS04. Baltimore, MD.
Honegger, B. (2005). WikisA rapidly growing
phenomenon in the German-speaking school
community. In Proceedings WikiSim05. San
Diego CA.
Hsu, K., & Tanner, J. (2006). An assessment of
classroom technology usage in business. International Journal of Innovation and Learning,
3(5), 488-498.
Johnson, D. W., & Johnson, R. T. (1998). Cooperation and competition. Edina, MN: Interaction.
Johnson, D. W., Johnson, R. T., & Holubec, E.
(1991). Cooperation in the classroom (3rd ed.).
Edina, MN: Interaction.
Kinzie, S. (2005, February 12). Colleges land lines
nearing an end. The Washington Post.
Kinzie, M., Whitaker, S., & Hofer, M. (2005).
Instructional uses of instant messaging during
classroom lectures. Educational Technology and
Society, 8(2), 150-160.
KPMG. (2003). Insights from KPMGs European
Knowledge Management Survey 2002/2003.
KPMG Knowledge Advisory Services, Amsterdam.
Krisnamurthy, S. (2002). The multidimensionality of blog conversations. Internet research 3.0.
Maastricht, The Netherlands.
Lamb, B. (2004). Wide open spaces Wikis ready
or not. Educause, September/October, 36-48.
Leontiev, A. (1978). Activity, consciousness,
and personality. Englewood Cliffs, NJ: Prentice
Hall.

1328

Leuf, B., & Cunningham, W. (1999). The WIKI

way: Quick collaboration on the Web. Reading,
MA: Addison Wesley.
Lim, K. (2005). Now hear thisExploring podcasting as a tool in geography education. Nanyang
Technological University.
Lipman, M. (1991). Thinking in education. New
York: Cambridge University Press.
Liu, C., & Chen, S. (2005). Determinants of
knowledge sharing of e-learners. International
Journal of Innovation and Learning 2005, 2(4),
434-445.
Locke, C. et al. (2000). The Cluetrain Mainfesto:
The end of business as usual. Cambridge, MA:
Perseus.
Louridas, P. (2006). Using wikis in software development. IEEE Software, March/April, 88-91.
Lum, L. (2006). The power of podcasting. Diverse
Issues in Higher Education, 23(2), 32.
Martindale, T., & Wiley, D. (2005). Using Weblogs
in scholarship and teaching. TechTrends, 49(2).
Mock, K. (2002). The use of Internet tools to
supplement communication in the classroom.
CCSC Northeastern Conference Proceedings,
17(2), 4-21.
Nardi, B., & Bradner, E. (2000). Interaction and
outeraction. In Proceedings of CSCW00. Philadelphia, PA.
Nardi, B., Schiano, D. J., & Gumbrecht, M. (2004).
Blogging as social activity. In Proceedings of
CSCW 2004 (pp. 222-231). Chicago.
Nardi, B., Schiano, D. J., Gumbrecht, M., &
Swartz, L. (2004). Why we blog. Communications
of the ACM, 47(12), 41-46.
Nussbaum, E. (2004, January 11). My so-called
blog. The New York Times.

Innovative Technologies for Education and Learning

ONeill, M. (2005). Automated use of a wiki

for collaborative lecture notes. In Proceedings
SIGCSE05 (pp. 267-271).
Ong, W. (1982). Orality and literacy. New York:
Routledge.
Prensky, M. (2001). Digital natives, digital immigrants. On the Horizon, 9(5).
Ractham, P., & Zhang, X. (2006). Podcasting in
Academia. In Proceedings SIGMIS-CPR06.
Raman, M., Ryan, T., & Olfman, L. (2005).
Designing knowledge management systems for
teaching and learning with wiki technology.
Journal of Information Systems Education, 16(3),
311-320.
Schiano, D., Nardi, B., Gumbrecht, M., & Swartz,
L. (2004). Blogging by the rest of us. In Proceedings CHI 2004 (pp. 1143-1146).
Seigenthaler, J. (2005, November 29). A false
Wikipedia biography. USA Today.
Seitzinger, J. (2006). Be constructive: Blogs,
podcasts, and wikis as constructivist learning
tools. Learning Solutions, July.
Seng, J., & Lin, S. (2004). A mobility and
knowledge-centric e-learning application design
method. International Journal of Innovation and
Learning 2004, 1(3), 293-311.

Wagner, C. (2004). Wiki: A technology for conversational knowledge management and group
collaboration. Communications of the AIS,
13(2004), 265-289.
Wagner, C. (2006). Breaking the knowledge
acquisition bottleneck through conversational
knowledge management. Information Management Resources Journal, 19(1), 70-83.
Walker, J. (2005). Weblog. Definition from the
Routledge Encyclopedia of Narrative Theory.
Wallace, M. (2005). Notes towards a literacy for
the digital age. Retrieved from http://uclaccc.ucla.
edu/articles/article-digitalage.htm
Wang, J., & Fang, Y. (2005). Benefits of cooperative learning in Weblog networks. In Proceedings
APAMALL2005 and ROCMELIA2005, Kun Shan
University.
Wenden, A. (1991). Learning strategies for learner
autonomy. New York: Prentice-Hall.
Wickramasinghe, N., & Lichtenstein, S. (2006).
Supporting knowledge creation with e-mail. International Journal of Innovation and Learning
2006, 3(4), 416-426.
Wikipedia. (2006). Mobile computing, definition. Retrieved August 25, 2006, from http://
en.wikipedia.org/wiki/Mobile_computing

Vygotsky, L. (1986). Thought and cognition.

Cambridge, MA: MIT Press.

This work was previously published in the International Journal of Information and Communication Technology Education,
edited by L. A. Tomei, Volume 3, Issue 3, pp. 70-89, copyright 2007 by IGI Publishing, formerly known as Idea Group Publishing (an imprint of IGI Global).

1329

1330

Chapter 3.22

Ubiquitous Computing
Technologies in Education
Gwo-Jen Hwang
National University of Tainan, Taiwan
Ting-Ting Wu
National University of Tainan, Taiwan
Yen-Jung Chen
National University of Tainan, Taiwan

Abstract
The prosperous development of wireless communication and sensor technologies has attracted
the attention of researchers from both computer
and education fields. Various investigations have
been made for applying the new technologies to
education purposes, such that more active and
adaptive learning activities can be conducted in
the real world. Nowadays, ubiquitous learning
(u-learning) has become a popular trend of education all over the world, and hence it is worth
reviewing the potential issues concerning the
use of u-computing technologies in education,
which could be helpful to the researchers who
are interested in the investigation of mobile and
ubiquitous learning.

Ubiquitous Learning: The New

Age for Education
In recent years, digitalization around the globe
has been proceeding toward wireless communication and sensor technologies, which are able to
detect the contexts of our daily lives, and provide
personal supports accordingly. Such technology
has been called ubiquitous computing (u-computing). The innovation and advance of those new
technologies have led to a new research issue
in education; that is, to develop an innovative
learning environment so that the students can
learn in any place at any time. Moreover, with
the help of context-aware (sensor) technology,
the learning system is able to detect the student
learning behaviors in the real world, and hence

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Ubiquitous Computing Technologies in Education

more active and adaptive learning activities can

be conducted. Such a learning scenario is called
context-aware ubiquitous learning (context-aware
u-learning), which has gradually become a popular
trend of education.
Researchers have demonstrated how a context-aware u-learning environment can be used
to help the learners in increasing their ability for
solving problems in the real world. For example,
a group of researchers in Japan has employed ucomputing technologies to conduct students to
learn Japanese under real-world situations. The
systems can provide learners with appropriate
expressions according to different contexts (e.g.,
occasions or locations) via mobile devices (e.g.,
PDA, Personal Digital Assistant).

Ubiquitous Computing Technologies

that Facilitate Education

U-computing technologies can be used to provide personalized services in the context-aware

u-learning environment. For instance, when
a student enters a lab or stands in front of an
instrument, the context-aware devices are able
to detect the location of the student and transfer
the information to the server. Based on the decision of tutoring program in the server, relevant
information, such as the operating procedure for
each device, the need-to-know rules for working
in the lab and emergency handling procedures,
will be timely displayed to the student based on
the personal and environmental contexts. Some
ubiquitous computing technologies that might
be useful in educational applications are given
as follows:
Sensors for Detecting Personal Contexts
Researchers have proposed several significant
characteristics of u-learning, which make it different from conventional e-learning, including
seamless services, context-aware services, and

adaptive services. In an ideal context-aware ulearning environment, the computing, communication and sensor equipment will be embedded
and integrated into the articles for daily use. In
addition, researchers also indicated that time
and location might be the most important parameters for describing a learners context.
There are several ways to detect the timely
location of a learner. GPS (Global Positioning
System) is one of the popular technologies for
continuously detecting an objects position by
satellites, which trace air waves shot from the
IC chips embedded in the objects. The objects
location is described with longitude, latitude and
elevation. Other sensors, such as RFID (Radio
Frequency Identification), which is an automatic
identification method relying on storing and remotely retrieving data using devices called RFID
tags or transponders, can also be used to detect
the location of a learner by reading the messages
from the tags, and then calculating the learners
position based on the intensity of the signals.
Advanced Technologies for Detecting Personal Contexts
Learners might feel distressed or confused
while encountering problems in the u-learning
environment. Under such circumstances, a ulearning system could actively provide timely
hints or assistance if the contexts concerning
human emotions or attitudes can be sensed.
Recent studies have depicted the possibilities
for detecting such advanced personal contexts.
Sensing devices with affective aware ability can
not only capture the expressions of human faces,
but also tell apart their emotional conditions. For
example, the Affective Computing Group in MIT
Media Lab of America have presented significant
progress in this field, which can be used to create
more friendly interaction between human and
computer by the detection of affective computing.
Other studies concerning facial expression detec-

1331

Ubiquitous Computing Technologies in Education

tion also demonstrated the ability of computers in

recognizing emotional conditions of people via
various sensors and algorithms.
Human voice is another context for describing
the learners status, which might be affected by
personal emotion, health condition, or surrounding noise. It is suggested that voice cognition
could be used to work with facial expression and
be physiological aware, to more precisely realized
the status of the learner.
Another advanced technology concerning
personal context detection is the development of
wearable computers, which can derive information from human actions as well as psychological/physiological conditions. Interactions between
human and environment could produce physiological changes directly or indirectly, including
human body temperature, pulse, blood pressure
and heartbeat, which can be automatically detected by those wearable sensors or context-aware
clothes for further analysis and explanation of the
learners behaviors.
Technologies for Detecting Timely Environmental Contexts
Context awareness is an important feature in
a u-learning environment. The learning system
can detect the contextual changes around learners,
and gather related parameters via various kinds of
sensors to provide students with adaptive learning
resources and assistance actively or passively.
Temperature is an important environmental
context for many applications, such as plant cultivation, physics and chemistry experiments. In
the industries of precision instruments, such as
semiconductor productions, it is very important
to control the temperature in the laboratory. Making humidity stable is also an important factor,
because some electrical instruments are easily
oxidized in the humid environment. On the other
hand, electrical instruments are easily affected
and out of order by static electricity in the dry
environment.

1332

In addition, for some labs with special purposes (e.g., precise instruments, biotechnology
and medical science), it is necessary to detect the
volume of particles in the air. Taking semiconductor as an example, these particles may cause
instruments a short circuit and even disable the
devices.

Research Issues and Target

Subjects
In Taiwan, the Conference on Mobile and Ubiquitous Learning was initiated in 2006 (which
is called Ubilearn2006). Based on the issues
discussed in Ubilearn2006 and Ubilearn2007,
several potential issues have been taken into
account:
1.

Architectures and infrastructures for ubiquitous learning systems

2. Adaptive and adaptable learning environments using mobile and ubiquitous devices
3. Agent support for mobile and ubiquitous
learning
4. Architectures and implementations of context-aware learning technology systems
5. Designs for wireless, mobile and ubiquitous
technologies in education
6. Design of learner-supportive interfaces of
ubiquitous learning applications
7. Evaluation and evaluation methodologies
for ubiquitous learning
8. Entertainment computing for ubiquitous
learning
9. Mobile and ubiquitous computing support
for collaborative learning
10. New pedagogical theories for ubiquitous
learning environments and society
11. Innovative and practical use of wireless,
mobile and ubiquitous technologies for
education, learning and training

Ubiquitous Computing Technologies in Education

12. Psychological or theoretical foundations for

mobile and ubiquitous learning and training
Moreover, a special interest group on Mobile
and Ubiquitous Learning Environment (MULE)
was established in 2007. Based on the previous
experiences from these researchers, some potential applications of context-aware ubiquitous
learning have been given, including art courses
(painting or drawing from life or nature), physical
education courses (motor skill training), language
courses (conversation training), natural science
courses (plant and animal classification) and

engineering courses (equipment operating and

work flow training).
Nowadays, u-learning is gradually becoming
an important learning style owing to the popularity of sensor and wireless network technologies.
As long as we strive toward the aim and develop
a feasible and effective learning model, the ideal
learning environment will become reality. In this
innovative learning environment, the system can
more actively provide more adaptive assistance to
the student based on their learning behaviors in
the real world, which have revealed the coming
of a new educational age.

This work was previously published in the International Journal of Distance Education Technologies, edited by S.-K. Chang
and T. K. Shih, Volume 5, Issue 4, pp. 1-4, copyright 2007 by IGI Publishing, formerly known as Idea Group Publishing (an
imprint of IGI Global).

1333

1334

Chapter 3.23

Utilizing Web Tools for

Computer-Mediated
Communication to Enhance
Team-Based Learning
Elizabeth Avery Gomez
New Jersey Institute of Technology, USA
Dezhi Wu
Southern Utah University, USA
Katia Passerini
New Jersey Institute of Technology, USA
Michael Bieber
New Jersey Institute of Technology, USA

Abstract
Team-based learning is an active learning instructional strategy used in the traditional face-to-face
classroom. Web-based computer-mediated communication (CMC) tools complement the face-toface classroom and enable active learning between
face-to-face class times. This article presents the
results from pilot assessments of computer-supported team-based learning. The authors utilized
pedagogical approaches grounded in collaborative

learning techniques, such as team-based learning,

and extended these techniques to a Web-based environment through the use of computer-mediated
communications tools (discussion Web-boards).
This approach was examined through field studies in the course of two semesters at a US public
technological university. The findings indicate
that the perceptions of team learning experience
such as perceived motivation, enjoyment, and
learning in such a Web-based CMC environment
are higher than in traditional face-to-face courses.

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Utilizing Web Tools for Computer-Mediated Communication to Enhance Team-Based Learning

In addition, our results show that perceived team

members contributions impact individual learning experiences. Overall, Web-based CMC tools
are found to effectively facilitate team interactions
and achieve higher-level learning.

INTRODUCTION
Instructors of both traditional face-to-face and
online classrooms seek active learning techniques
that engage the learners. The increased use of
Web-based computer-mediated communications (CMC) as support tools that supplement
the face-to-face classroom (blended learning)
and enable active learning between face-to-face
class times fit this quest. CMC is regarded as
an efficient computer support tool to facilitate
student participation (Phillips & Santoro, 1989).
Prior research (Wu & Hiltz, 2004) reports that
adding asynchronous online discussions through
CMC platforms enhances students learning
quality in a face-to-face class setting. Although
various Web-based computer-mediated communications learning strategies have been applied
in the field (e.g., online collaborative learning),
limited research focuses on computer-supported
team-based learning in a face-to-face classroom.
Team-based learning (TBL) is an instructional
strategy that promotes active learning in small
groups that form a team over time (Michaelsen,
Fink, & Knight, 2002).
Our goal is to assess the impact of teambased learning when introduced in a face-to-face
classroom that utilizes Web-based CMC as a
supplemental learning tool between classes, thus
increasing team interaction across the semester.
A Web-based computer-mediated communications tool called WebBoard was utilized in
our computer-supported team-based learning
research to facilitate team learning activities
and communication. This paper describes results
from this experience. The paper begins with
a literature review building on constructivist

learning, collaborative learning, small group

learning, and Blooms taxonomy theories. It then
provides examples of the Web-based interface
and pedagogical implementation, introducing a
model for assessing computer-supported teambased learning. Research questions, hypotheses,
and data analysis results are presented. Finally,
the limitations of the study and future research
efforts are discussed.
We believe that our contribution is two-fold.
First, we describe an approach for transferring a
grounded pedagogical approach to a Web-based
environment by supplementing the experiences
from a face-to-face classroom. Second, we document preliminary assessment results that support
the feasibility and effectiveness of the proposed
approach. This discussion should be of interest to
educators and researchers in expanding the use of
current Web-based learning management systems
with a structured modular approach through the
integrated use of discussion forums to achieve
higher-order team learning outcomes.

THEORETICAL BACKGROUND
Constructivist Learning Theory
Leidner and Jarvenpaa (1995) classify learning
models and discuss their relevance and impact
in information systems educational approaches.
The broadest categories of this classification
are objectivism and constructivism. Objectivism posits that learning occurs in response to
an external stimulus. Learners respond to the
stimulus by modifying their behaviors. This
model assumes that abstract representations of
reality and knowledge exist independently from
the learners. Teaching consists of transferring
knowledge from the expert to the learner. Opposite
to objectivism, constructivism posits that learning is not a process of knowledge assimilation,
but an active process of constructing individual
mental models, in which knowledge is created

1335

Utilizing Web Tools for Computer-Mediated Communication to Enhance Team-Based Learning

in the mind of the learner. In this model each

individual controls the pace and depth of his/her
instruction. The instructor is only a moderator
in the process of hypothesizing, questioning, and
discovering the conceptual relationships between
and among various objects.
Team-based learning uses a constructivist approach, which converts the learner from a passive
to active learner. This differs from the traditional
teacher-learner or objectivist approach. Students
in this constructivist learning environment play
a more active role as learners, since they need to
be well-prepared in order to effectively engage in
various class activities, such as to facilitate class
discussions, to be able to take challenges from
their peer learners and instructors, and so forth.
Therefore, the constructivist approach is aimed
to facilitate students critical thinking to achieve
higher-level learning.
This research explores how the constructivist
approach can be utilized to promote team-based
learning in a Web-based CMC tool environment. Building on experiences from face-to-face
team-based learning teaching (Michaelsen et
al., 2002), computer-mediated communication
tools are adopted to expand the collaboration
opportunities via online team activities between
weekly classes. Suitable technologies used in this
context are listservs (email distribution lists with
a Web-based interface) and asynchronous learning network systems (e.g., WebBoard, WebCT,
BlackBoard, or similar technologies provide both
a synchronous and an asynchronous discussion
environment).

Collaborative Learning
Built upon the constructivist theory, collaborativism suggests that learning quality enhances
through shared understanding of more than one
learner. Due to social interactions and intensive
sharing among collaborative learners, collaborative learning results in higher learning compared

1336

with individual-oriented learning (Leidner &

Fuller 1996). Schlechter (1990) found that learners perceived higher-order learning by generating
more creative ideas and more diverse reasoning
in collaborative small study groups.
Our computer-supported team-based learning
approach takes advantage of collaborative learning strategies, with more structured teams, which
exist for a relatively longer period (in our case,
in the entire semester) and with a stable team
setup (i.e. same team members), in comparison
to other approaches where most teams are built
much more casually and temporally.

Small Group Learning and

Team-Based Learning
The importance of small groups learning and
knowledge creation has been increasing in both
education and industry. Team-based learning
focuses on fixed small groups, which are established for semester long collaboration instead
of temporary purposes. Small groups promote
each others learning and success by holding
each member personally responsible for the fair
share of the work (Johnson, Johnson, & Smith,
1991). This turns the learning experience into a
process, which improves the quantity and quality
of the learning by leveraging long-term caring and
peer relationships (Johnson & Johnson, 1999).
Team-based learning simulates the similar peer
collaboration experience in a real world. Therefore, the team-based learning experience may
represent a useful training for students long-term
career success.
The main emphasis of team-based learning
is the organization around modules (work units)
across the semester, consisting of 5-7 three
phase sequences (Michaelsen et al., 2002). Each
sequence includes preparation, application and
assessment before moving to the next unit. Teams
should be five to seven members in size. They
evolve through four essential procedures: team
formation, student accountability, team activi-

Utilizing Web Tools for Computer-Mediated Communication to Enhance Team-Based Learning

ties, and high quality feedback (Michaelsen et al.,

2002). Our research uses these four procedures
as the basis for the introduction of computer-supported techniques.

Phased Cognitive Development

Perspectives
The approaches presented in this study are based
on pedagogical theories that recognize the role
of incrementally promoting higher-level learning through progressive team activities. Bloom,
Englehart, Furst, Hill, and Krathwohl (1956) organize learning based on different phased cognitive
development stages of information elaboration,
starting from knowledge and comprehension to
the higher levels of evaluation and synthesis.
Bloom et al.s taxonomy of learning goals and
objectives summarizes the way student learning
progresses through various stages of increasing
complexity.
Blooms taxonomy is particularly useful to
identify a progression of learning, which is applied
to the computer-supported team-based learning
process. In the lower levels of the taxonomy
(knowledge, comprehension, application) students
absorb knowledge that is presented to them, comprehend it, and learn how to rank order and make
inferences in other contexts. In the higher levels of
the taxonomy (analysis, synthesis, and evaluation),
students learn to generalize and transfer what they
learned to their personal interpretation of reality.
They learn to analyze and classify the models they
have created. While the synthesis and evaluation
levels of the taxonomy can also be accomplished
by individual learners, collaborative methods and
working in teams provide opportunities to excel
in integrating associations and making judgments
through communication with peers. In this collaboration mode, students can reach the higher
levels of the Blooms taxonomy.

WEB-SUPPORTED
TEAM-BASED LEARNING
EXPERIENCES
Computer-supported team-based learning conducted via a Web-based CMC tool was introduced
in a 15-week semester, face-to-face graduate
course at a US public technological university
during the Fall 2004 and Spring 2005 semesters.
The first class period introduced the computer-supported team-based learning instructional strategy.
The class was divided into teams of five to six
students, including at least one woman and at least
one student with a wireless laptop. The course
materials were divided into six modules, with
no midterm or final. Two out-of-class individual
article reviews were assigned to the students.
The graduate courses in which we conducted
this research was re-modeled, according to
Michaelsen et al. (2002). Several steps of the
TBL instructional strategy were modified to
incorporate the use of a Web-based CMC tool
called WebBoard (http://www.webboard.com/)
intended for active learning activities between
weekly face-to-face classes. A phased approach
to active learning assignments via WebBoard was
implemented across the two semesters of the study
(Gomez & Bieber, 2005). Particular emphasis
was placed on additional individual preparedness pre-module activities and team post-module
activities through the use of WebBoard; a learning management system that provides a flexible
interface for classroom discussion management.
Figure 1 shows a typical discussion interface in
WebBoard. The left frame provides an organized
list of class activities. The right frame shows the
details of each individual posting. For example,
in the first semester, students posted preparation
materials on WebBoard (http://www.webboard.
com/) and wrote their post-team activity results
on a poster in class. In the second semester, teams
posted results of their work on WebBoard instead
of handwriting on posters.

1337

Utilizing Web Tools for Computer-Mediated Communication to Enhance Team-Based Learning

Figure 1. WebBoard interface

The team-based learning activities are based

on an iterative process that divides the course
into five to seven modules (themes/work units),
as Michaelsen et al. (2002) suggests. A module
is defined as a work unit or theme. Each module
introduces increasingly complex materials that
students must master before a subsequent module
is introduced. A module can span more than one
weeks worth of course materials, but a single
activity should not span more than one class.
Web-based CMC tools enable spanning beyond
the single class time (when materials are hard to
understand) by enabling the team to work between
scheduled class times. Each of the modules must
then be subdivided into individual preparation
(reading materials and discussions), readiness
assessment tests (individual and team), multiple
team activities, and post module assessment.
The instructor acts as a facilitator across each
module. He/she is responsible for hypothesizing,

1338

questioning, and discovering the conceptual relationships between and among various concepts
(Passerini & Granger, 2000), and spending more
time on preparation of the readiness assessment
tests (RAT) and team activities, rather than course
lecturing. The instructor plays a pivotal role in
the TBL process, both for the course organization
and ongoing module activities. In lieu of class
lecturing, the instructor actively participates
in the classroom as a task coordinator and time
manager while observing each teams discussions
to readily intervene either as a subject matter
expert or to clarify points of the task. Timely
grading and feedback are essential aspects of the
TBL process and should not be overlooked when
planning. The instructor also provides feedback
when team activities come together for sharing
across the classroom.
Our emphasis with the introduction of Webbased tools to the TBL process is on the individual

Utilizing Web Tools for Computer-Mediated Communication to Enhance Team-Based Learning

Table 1. Team-based learning instructional strategy

TBL Elements
Course Structure
Instructor Lecturing

Recommended (Michaelsen,
2002)
Discouraged

Individual Preparedness

Assign all module reading

materials at start of module.

Readiness Assessment
Test (RAT)

At the start of each module:

5-7 modules per semester.

Team Activities

Multiple activities of
increasing complexity per
module.
Tests, assignments and
certain in-class team
activities should be graded.
Not mentioned.

Grading
Course Survey
Team Formation
Team Size
Team Arrangement
Team Roles
Team Composition
Team Diversity
Team Building

5-7
Same groups
No assigned roles.
Not specified.
Recommended that
diverse groups are formed
randomly.
Achieved through initial
tRAT, team exercise to
determine grade weights and
choose a team name.

preparedness of the student before taking the

individual readiness assessment test for a given
module and subsequently taking the team readiness assessment test. To accompany the reading
assignments, students were asked to post a short
reading summary and then reply to another
students reading summary. In the Fall 2004
semester, the reading summaries were discussed
online across the classroom. In the Spring 2005

Description and Implementation

The instructor acted as a facilitator and observed
discussions. No lecturing was provided.
All reading materials were assigned, reading
summaries, and test notes were posted on
WebBoard before the RATs.
Six modules, each team can ask one question about
module materials, individual test (iRAT) followed
by team test (tRAT).
Multiple team activities were given with
increasing complexity per module.
Class determined grade weightings of all exams/
tests, assignments and activities.
End of semester.
5-6 students for each team.
Same groups working together all semester
Roles not assigned
Balanced distribution among gender
Students were assigned to teams based on their
computer experience
As recommended in addition to a phased in
approach to within team reading summaries and
sharing of notes for tests.

semester, the reading summaries were discussed

within each team. Our observations across both
semesters indicate that discussions within the team
demonstrated two benefits: extended discussions
with the team and increased discussions prior to
the team readiness assessment tests, reinforcing
the preparation at the start of a new module. This
suggests that discussions between team members
are more effective and rewarding than discussions

1339

Utilizing Web Tools for Computer-Mediated Communication to Enhance Team-Based Learning

across the entire classroom. In reviewing the

discussion board postings, the length and depth
of the discussions were richer and more engaging
when individuals were discussing within their
team. This provides the additional benefit of
increased team cohesion; a quantitative measure
to be included in future research.
The student transition from their individual
preparedness activities to the readiness assessment tests. Individual preparedness activities
include all reading materials for the module,
reading summaries, and individual exam notes
for sharing with teammates, requiring students to
organize their preparation time. It is suggested the
questions for the readiness assessment tests should
span the modules reading materials and be geared
toward concepts; not detailed questions typically
categorized as by rote. The questions should be
broad enough to cover all of the modules reading materials. Michaelsen et al. (2002) suggests

approximately twenty multiple choice questions

at the start of each new module for the readiness
assessment test (RAT). Our research used a point
system for the team readiness assessment test
where teams were given a scratch card and were
allowed to select more than one answer. Points
were awarded in decreasing value (e.g. five points
first try, three points second try, one point third
try). Teams are permitted to appeal a question,
with five to ten supporting sentences to ensure a
concise and well thought out rebuttal. Students
build cases with convincing evidence to persuade
the teacher to award credit for an answer missed by
the group on the test (Michaelsen et al., 2002).
Multiple team activities per module are needed
with increased levels of complexity and are not
assigned until after the team readiness assessment
test. Typically team activities span two or three
weeks for a single module, but a single activity
does not carry over from one week to the next.

Figure 2. Computer-supported team-based learning events

1340

Utilizing Web Tools for Computer-Mediated Communication to Enhance Team-Based Learning

The ideal activity is one that is a case study or

has a problem-solving aspect to the activity. After
each team activity, deliverables were reviewed
and critiqued by the other teams, possibly for
extra credit. The review and critique occurred in
the FtF classroom for the first semester and using WebBoard for the subsequent semester. The
premise was to deepen the learning and reinforce
the objectives of the activity, while providing
the instructor an opportunity to comment. Upon
completion of a team activity, students should
review the other teams deliverable and provide
feedback to generate discussion.
Figure 2 summarizes the team-based learning
iterative process and where computer-supported
activities were introduced in this study.

ASSESSING
COMPUTER-SUPPORTED
TEAM-BASED LEARNING
The constructivist approach, cooperative, and
collaborative theories and models, provide the

background for the elaboration of the team-based

learning assessment framework reviewed in this
study (see Figure 3). Earlier research on Web-supported participatory examinations (Wu, Bieber,
Hiltz, & Han, 2004) and collaborative examinations (Shen et al., 2004), informs the proposed
team-based learning constructs and expected
relationships. In order to succeed in this TBL
process, students should prepare first, so that
they can contribute their insights to the whole
team (individual preparedness construct). We
assume that their individual preparedness might
bring more confidence for them on the value their
team members contributions to problem solving, and they can also integrate and gain more
insights from their team members. During this
TBL process, students can learn more by communicating efficiently, and by building trust among
their team members. Overall, we expect that the
whole TBL process will provide the students an
enjoyable learning experience, and will motivate
them think at the higher levels of Blooms learning
taxonomy, enhancing the quality of the learning
experience. Figure 3 lists the key constructs and
hypotheses tested in this study.

Figure 3. Team-based learning (TBL) framework

Perceived
Individual
Preparedness from
TBL
H1c

H1a

H2a

H3b
H2c

Perceived Team
Member s Value
or Contributions
from TBL

Perceived
Learning from
TBL

H3a
H1b
H2b

Perceived Trust
/Communication
Skills from TBL

Perceived
Motivation from
TBL

H4

Perceived
Enjoyment from
TBL

1341

Utilizing Web Tools for Computer-Mediated Communication to Enhance Team-Based Learning

Definitions: Perceived Individual Preparedness is a self-reported student assessment

of deep versus superficial study of the materials (Gomez, Wu, Passerini, & Bieber,
2006; Michaelsen et al., 2002); Perceived
Team Member Value or Contributions is the
positive perception of the value of teamwork
(Isabella & Waddock, 1994); Perceived
Trust/Communication Skills refers to the
openness of communication within the
teams (Coppola, Hiltz, & Rotter, 2004);
Perceived Motivation refers to the intrinsic
and extrinsic factors that influence students
interest and attitudes towards the learning
experience (Malhotra & Galletta, 2003);
Perceived Enjoyment is the extent to which
the learning activity (the team based learning
experience) is pleasant and satisfactory to the
learners (Davis, Begozzi, & Warshaw, 1992);
Perceived Learning focuses on perception of
the course quality; usefulness and extend of
individual learning experiences as reported
by the learners (Wu et al., 2004).

In exploring the relationships and impacts

of computer-supported team-based learning, we
focused on a number of factors that could impact
students perceptions of learning. We adopted and
modified a few validated constructsperceived
learning, perceived motivation and perceived
enjoymentfrom asynchronous online discussion research (Wu & Hiltz 2004). In addition, we
created two new constructs called perceived
individual preparedness and perceived team
members value/contributions. More specifically,
we expect that the collaboration experiences and
team-based activities completed by the teams
throughout the courses will have a positive impact
on perceived learning. Individual preparedness,
measured as a self-reported student assessment
of their deep versus superficial study of the materials, will positively impact the perceived value
and contribution of the team-learning experience.
This assumption follows Michaelsen et al.s

1342

(2002) findings that individual contributions to

team-output will promote team development and
reduce social loafing and, therefore, will have a
higher impact on the overall team-based learning
experience.
Prior asynchronous learning networks research (Wu & Hiltz, 2004; Wu et al., 2004)
shows a positive correlation between perceived
motivation, enjoyment, and learning, when students participate in designing, answering, and
grading exam questions on an asynchronous
learning networks tool. These findings occur in
the context of students learning experience from
online discussions through computer-mediated
communications platforms for blended classes,
which mixed traditional classroom lecturing with
asynchronous online discussions.
Individual preparedness and perceived member contribution to computer-supported teambased learning are expected to have a positive
impact on learning, as well as on motivation and
enjoyment of the team-based learning experience.
In particular, motivation and enjoyment will act
as intervening variables (see Figure 3). We also
expect motivation to positively impact the overall
enjoyment of the team-based learning exercises.
Following Thorndikes Law of Intensity and
Law of Readiness (Thorndike, 1932), we expect
that students who are engaged in the learning
process through the multiple collaboration experiences embedded in the course organization,
will be more motivated to learn, and will enjoy
the learning experience better.
Lastly, we expect that the independent variables of individual preparedness, perceived team
member value and contribution to computer-supported team-based learning will impact the level
of trust and openness of communication within
the teams, indirectly influencing communication
and enjoyment. Establishing trust early (swift
trust) in on-line communities has been found to
have a positive impact (Coppola et al., 2004) on
the learning experience. While we do not report
results on trust measures (further analysis and

Utilizing Web Tools for Computer-Mediated Communication to Enhance Team-Based Learning

reiteration are needed as trust measures were collected only in the Spring 2005 reiteration of the
field study), we include trust and communication
as an important variable to consider in a computersupported team-based learning model, which will
be influenced by the individual preparedness and
team contributions. As discussed in the future
research section of this paper, additional observations in the Summer and Fall semester 2005 will
supplement our preliminary observations.

Research Questions and

Hypotheses
Based on the above discussion, our research is
set out to investigate the following general questions:

Does students perception of team contributions impact their learning from the
computer-supported team-based learning
process?
Does individual preparedness affect perceptions of computer-supported team-based
learning experiences?

A number of research hypotheses are therefore

derived from the proposed research framework
(see Figure 3).
Hypothesis 1a&b: Higher individual preparedness will increase students perceived
motivation and enjoyment from computersupported team-based learning process.
Hypothesis 1c: Higher individual preparedness
will increase the perception of team members value and contribution to the computersupported team-based learning process.
Hypothesis 2a&b: Higher perceived team
members contributions to computer-supported team-based learning will increase
perceived motivation and enjoyment from
this process.

Hypothesis 2c: Higher perceived team members contributions to computer-supported

team-based learning will enhance perceived
learning in this process.
Hypothesis 3a&b: Higher perceived motivation
will lead to higher enjoyment and learning
from the computer-supported team-based
learning.
Hypothesis 4: Higher perceived enjoyment from
the computer-supported team-based learning will lead to higher learning.

Methods and Sampling

To test the hypotheses, survey questionnaires
were used in the same masters-level information
systems course called Information Systems
Principles during Fall 2004 and Spring 2005
semesters. A total of 73 students volunteered to
participate in our computer-supported team-based
learning study. Among 73 respondents, 61 students
disclosed their demographic information. Over
50% of them are full-time students, about 33%
are part-time, and the rest are non-matriculated
students, who are studying without officially being admitted to degree programs. About 60% are
males, 36% are females, and three people did not
provide their gender information. The majority of
the respondents are between 21 to 30 years old,
about 16% are between 31 to 40 years old, and
the rest are between 41 to 50 years old.

Data Analysis and Discussion

In this study, the majority of data was collected
from the surveys, which covered all variables
proposed in the computer-supported team-based
learning research framework (Figure 3). The
primary data analysis methods to analyze our
survey are descriptive analysis, factor analysis,
and correlation analysis. First, descriptive data
analysis was conducted to evaluate the validity
of the constructs and the reliability of the scale
used in the study. All the constructs returned a

1343

1344

1
2
3
4
5
6
7
8
9
10
11
[]
23

Component

.363

100.000

Cumulative %
37.442
48.232
55.716
62.319
67.197
71.694
75.535
79.126
82.216
84.947
87.225

Extraction Sums of Squared

Loadings
% of
CumulaTotal
Variance
tive %
8.612 37.442
37.442
2.482 10.790
48.232
1.721 7.484
55.716
1.518 6.602
62.319
1.122 4.879
67.197
1.034 4.497
71.694

Rotation Sums of Squared

Loadings
% of
CumulaTotal
Variance
tive %
4.753
20.665
20.665
3.660
15.912
36.577
2.454
10.669
47.246
2.278
9.905
57.151
1.775
7.719
64.870
1.570
6.824
71.694

Cronbachs Alpha higher than 0.70. The detailed

question items for each construct are listed in
Appendix section.
The factor analysis was then performed to confirm the loading of the questions to the respective

8.357E02

Initial Eigenvalues
% of
Total
Variance
8.612
37.442
2.482
10.790
1.721
7.484
1.518
6.602
1.122
4.879
1.034
4.497
.883
3.841
.826
3.591
.711
3.089
.628
2.732
.524
2.278

Table 3. Components extraction matrix

Utilizing Web Tools for Computer-Mediated Communication to Enhance Team-Based Learning

factors. In general, the question items show high

reliability of the constructs they represent. The
exploratory factor analysis technique was used in
this study to verify whether the grouping of the
survey questions maps to the model constructs.

1
1
-0.026
1
0.288*
-0.188
1
0.708**
0.518**
-0.174

(4)
(3)
(2)
(1)

1
0.635**
0.637**
0.437**
-0.171
** Correlation is significant at 0.01 level (two-tailed)
* Correlation is significant at 0.05 level (two-tailed)

Table 4 shows the bivariate correlations among

variables and provides intriguing results. In
particular, it shows that perceived enjoyment
and motivation are highly correlated with the
dependent variable of perceived learning (R>0.63
for both constructs). It also shows that perceived
team members contribution positively correlates to learning, enjoyment and the motivation
constructs. However, the bivariate correlation
analysis shows that individual preparedness is
not significantly related to the other variables,
and the relationship, if any, is negative.
Based on the results of the above bivariate
correlations, we identify that hypothesis 1 (a, b,
and c) are not supported. The other hypotheses
are supported at the p=0.01, with the correlation

Perceived Learning from TBL

Perceived Enjoyment from TBL
Perceived Motivation from TBL
Perceived Team Members Value/ Contribution from TBL
Individual Preparedness from TBL

Extraction Method:
Principal Component Analysis

Table 4. Bivariate correlation analysis

An exploratory factor analysis (EFA) was run

to identify the underlying model, the number of
factors in the model, and the variables (by type of
questions) associated with each factor. Within the
factor analysis technique, a principal components
analysis (PCA) was used to highlight the number
of factors to be extracted from the model. The
confirmatory factor analysis (CFA), in its rotated
version (VARIMAX rotation) helps to identify
the underlying model. The Principal components
analysis yielded six factors with eigenvalues >1.
We focused on the first five components corresponding to the research model constructs
(Table 3) because of the limitation of one of the
research constructs (individual preparedness)
that yielded ambiguous results and because the
additional variance explained by the additional
factor did not substantially alter the model (as it
is the lowest incremental cumulative variance). In
addition, we did not test the trust/communication
construct, with a possible impact on the overall
model. Limitations of this approach as introduced
at the end of this section.

(5)

Utilizing Web Tools for Computer-Mediated Communication to Enhance Team-Based Learning

between team contributions and motivation significant at the p=0.05 level (see Figure 4).
From the above data analysis results, it is
suggested that how individuals value their team

1345

Utilizing Web Tools for Computer-Mediated Communication to Enhance Team-Based Learning

Figure 4. Computer-supported team-based learning bivariate correlation results

Perceived Individual
Preparedness from
TBL

Perceived Motivation
from TBL

(+) H3b supported

R=0.63**

(+)H2a supported
R=0.28*
Perceived Team
Member s Value or
Contributions from
TBL

(+)H2c supported
R=0.43**

(+) H2b supported

(-) H1a, b, c) not supported R=0.51**
* p < 0.05
** p < 0.01

(+) H3a supported

R=0.70**

(+) H4 supported
R=0.63**

Perceived Enjoyment
from TBL

members contribution has significant correlations

with their perception of enjoyment and learning
quality from computer-supported TBL (Gomez et
al., 2006). These results address the first research
question on how team interactions positively impact the whole computer-supported TBL learning
experiences. In addition, individual opinions on
team members contributions also have a positive
impact on their perceived motivation, although
the Pearsons R value is not high (R = 0.28). It
might be caused by other potential factors from
the computer-supported TBL experience, which
could decrease students motivation. For instance,
if the team leader is more dominant, his control
might impact other team members motivation.
Surprisingly, individual preparedness does
not impact perception of the team-based learning
experience. The correlation values among individual preparedness and other variables are not
significant. The results also show that their correlation values are negative. There might be a few
reasons. First, as indicated earlier, the ambiguous
two question items for the individual preparedness construct (i.e. ambiguous statement, negative factor loading) have a negative impact on the
research framework. The individual preparedness

1346

Perceived Learning
from TBL

tasks associated with studying differences may

not be self-evident to the student increasing the
ambiguity of the questions. Moreover, our observations indicate the importance of individual
preparedness on the entire iterative nature of the
module whereas our focus of the current individual preparedness construct was driven from
the importance of the readiness assessment test
process. The bivariate correlation analysis does
not show any significant relationships between
individual preparedness and other constructs
in the framework. This is evidently one of the
limitations for this research. For future research,
we plan to refine the individual preparedness
construct with question items targeted at the individual preparedness process more than studying
specifically for the tests. Second, there might be an
interaction effect of the experimental conditions:
the computer-supported TBL process design itself
might also impact the results. The team assessment tool (tRAT) is the same test as the individual
readiness assurance test (iRAT). Although the
overall team scores are better than individual test
scores, the test repetition may explain the decrease
in the students motivation and enjoyment. Also
they could not perceive more value from their

Utilizing Web Tools for Computer-Mediated Communication to Enhance Team-Based Learning

respective team members. We also speculate that

the test questions may not have lent themselves
well to interesting team discussions, leaving the
tRATs uninteresting. Alternatively, this simply
may indicate that many students found the TBL
process valuable even when they did not prepare
in the manner the instructor expected. These results show that our second research question on
the role of individual preparedness needs further
investigation and analysis to span the iterative
cycle of each module (preparation, test, activity)
and not simply focus on the readiness assessment
test aspect.
There are a number of limitations to this
study. First and foremost, we found that some
of the constructs could be better specified. Our
factor analysis displayed the possible existence of
another significant factor which should be further
analyzed in future research. This additional factor
was associated with an ambiguous question in the
individual preparedness construct. Looking at the
articulation of the questions in that construct, we
identified an instrumentation bias (one item in
the construct was weak), which may explain our
concerns with the specific construct. In addition,
because some of the constructs and the extension
of this grounded team-based learning approach to
a Web-environment is novel, we should supplement our conclusions with an in-depth analysis
of qualitative factors (observations, open-ended
questions, and content analysis of discussion
boards) that may help better understanding the
learning outcomes more objectively in a way that
complements the perceived learning measures.
Adjusting the activities layout in the Web-based
environment with our phased approach and the
nature of the individual preparedness activities
indicate the complexities of this construct. Our
qualitative analysis indicates that spanning beyond
the individual preparedness question items related
to the readiness assessment test to also include
question items that measure perceived preparation
of the materials related to team activities.

SUMMARY AND FUTURE

RESEARCH
Computer-supported team-based learning provides a powerful instructional experience and
reduces some of the disadvantages many instructors and students have found with traditional
small-group work. Blending the benefits of the
face-to-face classroom with computer-mediated
communication extends the learning process
between the weekly face-to-face sessions keeping
the team learning progress and group dynamics
growing.
Our research places emphasis on key variables
that affect learning in computer-supported teambased learning. Computer-supported team-based
learning is still a relatively new pedagogical approach, and to the best of our knowledge, this is
the first study blending computer-mediated communications with the iterative team-based learning
modular approach proposed by Michaelsen et
al. (2002). The use of Web-based CMC learning
techniques places emphasis on individual and
team learning outcomes. The surveys indicate
a high-perception of learning, motivation, and
enjoyment. These findings deemed computersupported team-based learning an approach for
further investigation both in the face-to-face
classroom and for online learning.
The emphasis of future research will be on
team-assessments and group cohesion in a purely
Web-based learning environment. The findings
around the team activities will allow for additional
adjustments in the team-based learning process
before it is introduced in a completely online
learning mode. Blending the face-to-face class
with computer-mediated communications provides a means to gauge the asynchronous learning
network process. Future studies will extend the
analysis of the computer-supported team-based
learning model and research framework using the
structural equations model (SEM), trust, com-

1347

Utilizing Web Tools for Computer-Mediated Communication to Enhance Team-Based Learning

munication, and team leadership factors. Further

review of individual and team preparedness is also
needed. The progressive nature of the readiness
exam process and team activities should ensure
individual preparation. Because of the novelty
of the preparedness and team contributions constructs, we will also implement content analysis
of team activities posted on WebBoard to support the evaluation of individual preparedness
for each module. Adding qualitative data and
observations will enhance our understanding of
the constructs. Actual grades and peer evaluation
results will also support the measurement of task
completion levels.
Team-based learning presents a promising
technique employing small teams that actively
engage students in learning. We look forward
to the day when instructors can effectively use
computer-supported team-based learning as a
standard approach in both face-to-face and online
classrooms.

education objectives: Handbook I. The cognitive

domain. New York: McKay.

ACKNOWLEDGMENTS

Gomez, E. A., Wu, D., Passerini, K., & Bieber,

M. (2006). Introducing computer-supported
team-based learning: preliminary outcomes and
learning impacts. In M. Khosrow-Pour (Ed.),
Emerging trends and challenges in information
technology management: Proceedings of the 2006
Information Resources Management Association
Conference, (pp. 603-606). Hershey, PA: Idea
Group Publishing.

We gratefully acknowledge partial funding support for this research by the United Parcel Service
Foundation, the New Jersey Center for Pervasive
Information Technology, the New Jersey Commission on Science and Technology, and the
National Science Foundation under grants IIS0135531, DUE-0226075 and DUE-0434581, and
the Institute for Museum and Library Services
under grant LG-02-04-0002-04. An earlier version of this paper was presented at the IRMA
2006 International Conference in Washington
DC, May 2006.

REFERENCES
Bloom, B.S., Englehart, M.D., Furst, E. J., Hill,
W.H., & Krathwohl, D.R. (1956). A taxonomy of

1348

Coppola, N., Hiltz, S.R., & Rotter, N. (2004).

Building trust in virtual teams, IEEE Transactions
on Professional Communication, 47(2) 95-104.
Davis, F. D., Bagozzi, R. P., & Warshaw, P. R.
(1992). Extrinsic and intrinsic motivation to use
computers in the workplace. Journal of Applied
Social Psychology, 22,1111-1132.
Gomez, E.A., & Bieber, M. (2005). Towards active
team-based learning: An instructional strategy. In
Proceedings of the Eleventh Americas Conference
on Information Systems (AMCIS)( pp. 728-734).
Omaha, .
Gomez, E.A., Wu, D., Passerini, K., & Bieber,
M. (2006, April 19-21). Computer-supported
learning strategies: An implementation and assessment framework for team-based learning.
Paper presented at ISOneWorld Conference, Las
Vegas, NV.

Isabella, L., & Waddock, S. (1994). Top management team certainty: environmental assessments,
teamwork, and performance implications. Journal
of Management, Winter.
Johnson, D., & Johnson, R. (1999). What makes
cooperative learning work. Japan Association for
Language Teaching, pp. 23-36.
Johnson, D.W., Johnson, R.T., & Smith, K.A.
(1991). Cooperative learning: increasing college
faculty instructional productivity. ASHE-ERIC
Higher Education Report No.

Utilizing Web Tools for Computer-Mediated Communication to Enhance Team-Based Learning

Leidner, D., & Jarvenpaa S.L. (1995). The use

of information technology to enhance management school education: A theoretical view. MIS
Quarterly, September, 265-291.

Phillips, G. M., & Santoro, G. M. (1989). Teaching group discussions via computer-mediated
communication. Communication Education, 39,
151161.

Leidner, D., & Fuller, M. (1996). Improving

student processing and assimilation of conceptual information: GSS-supported collaborative
learning vs. individual constructive learning.
In Proceedings of the 29th Hawaii International
Conference on System Sciences (HICSS-29), Big
Island, Hawaii, pp. 293-302.

Schlechter, T. M. (1990). The relative instructional

efficiency of small group computer-based training. Journal of Educational Computing Research,
6(3), 329-341.

Malhotra, Y., & Galletta, D. (2003, January 6-9).

Role of commitment and motivation in knowledge
management systems implementation: Theory,
conceptualization, and measurement of antecedents of success. In Proceedings of 36th Annual
Hawaii International Conference on Systems
Sciences, pp. 1-10. IEEE.
Michaelsen, L., Fink, D., & Knight, A. (2002).
Team-based learning: A transformative use of
small groups in college teaching. Sterling VA:
Stylus Publishing.
Passerini, K., & Granger, M.J. (2000). Information
technology-based instructional strategies. Journal
of Informatics Education & Research, 2(3).

Shen, J, Cheng, K., Bieber, M., & Hiltz, S. R.

(2004). Traditional in-class examination vs. collaborative online examination in asynchronous
learning networks: field evaluation results. In
Proceedings of AMCIS 2004.
Thorndike, E. L. (1932). Fundamentals of learning. New York: Teachers College Press.
Wu, D., & Hiltz, S. R. (2004). Predicting learning
from asynchronous online discussions. Journal
of Asynchronous Learning Networks (JALN),
8(2), 139-152.
Wu, D., Bieber, M., Hiltz, S. R., & Han, H. (2004,
January). Constructivist learning with participatory examinations, In Proceedings of the 37th
Hawaii International Conference on Systems
Sciences (HICSS-37), Big Island, CD-ROM.

This work was previously published in the International Journal of Web-Based Learning and Teaching Technologies, edited
by L. Esnault, Volume 2, Issue 2, pp. 21-37, copyright 2007 by IGI Publishing, formerly known as Idea Group Publishing (an
imprint of IGI Global).

1349

1350

Chapter 3.24

How Technology Can Support

Culture and Learning
David Luigi Fuschi
GIUNTI Labs S.r.l., Italy
Bee Ong
University of Leeds, UK
David Crombie
DEDICON S.r.l., The Netherlands

Abstract

Introduction

From the authors perspective, technology is both

a tool and a developing factor that can foster
culture and learning development. This chapter focuses on the interrelations that interleave
education, technology, content accessibility, and
intercultural issues. With an introduction and
related background, language learning is used as
an example further to explain these issues. This
chapter then discusses authoring and content development for e-learning applications (including
authoring tools, virtual communities, and forums),
and examines technology and accessibility issues
in this context. The current state of e-learning is
introduced along with a description of different
tools and approaches. The chapter concludes with
an overview of e-learning and the marketplace.

Major developments in culture and civilisation

are based on a combination of factors including personal abilities, societal, and economical
environment, and so on. According to Maslow
(Abraham, 1943), subsistence needs have to be
met before other and higher aims are reached
for (Figure 1). Nevertheless, the search for a solution to a problem (even a very practical one) may
provide the core of our evolution in technology,
science, culture, art, or education.
From the authors perspective, the issue regarding the relation interleaving technology, culture,
and education is twofold. On one side, technology has often been a major cause of disruptive
advances in the evolution of science, culture,
art, and education. On the other side, scientific

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

How Technology Can Support Culture and Learning

Figure 1. Maslows hierarchy of needs (Adapted from Abraham Maslow, 1943)

and cultural progress necessarily rests upon a

learning process, either formal or informal. For
this very reason, the authors have focused their
attention on how technology can support culture
and learning.
For example, the present development of
western culture has been reached thanks, in large
part, to the technical innovation introduced by
Gutenberg, which made possible the diffusion of
culture through print, making learning affordable
for everyone and therefore, also accessible. Yet
were it not for Braille, all the culture and knowledge available in printed form would be totally
inaccessible to visually impaired people unless
they could afford to have a personal reader. This
technical support is therefore a clear example of
how technology can support culture and learning
in the wider sense.

It is worth taking into account historical notes

related to the impact of technology on the learning process and therefore, also on the evolution
of society.

Some history notes

The art of teaching and learning has a long history. Initially it was a matter of subsistence and
therefore, the elders in a community would act
as tutors, trainers, and mentors for the younger
members. This process then evolved from mere
subsistence to something that became much valued
by the learner and rewarding for the teacher.
Due to the very nature of this process (based
on knowledge acquired through personal practical experience), teachers were numerically very

1351

How Technology Can Support Culture and Learning

limited in respect to potential learners; moreover,

not everyone could benefit from mentoring, as
most of the individuals would have to work hard
to support the groups living needs. Therefore,
only the most brilliant, or the more influential
members, would be granted such a precious gift
as formal learning and this had a strong social
impact.
For a long period of time, education and culture
has been accessible only to upper classes; religion
played a relevant role in this, as knowledge was
mainly held by priests or monks, who would
also act as tutors and mentors for nobles, ruling
classes, and their earls or illegitimate sons.
This phenomenon was so strong that during
the Middle Ages the term clerices1 was used
generically to indicate the people with an education. This is also the period of the birth of the first
universities like the Alma mater in Bologna
(1088), the Sorbonne in Paris, Montpellier,
the Scuola Medica Salernitana, Heidelberg,
Salamanca, and many other ancient and prestigious universities in Europe. The clerices vagantes, in their roaming through Europe, spread
culture, teaching in the various universities they
visited, rivalling each other to achieve eternal
fame thanks to the universal language of the time
(Latin). A famous example is Abelard (the wellknown character of Abelard and Eloise). Most
people consider this as a tale while it actually
was a real event in European history.
The main academic subjects at the time would
have been literature, philosophy, religion, and
law, while science was more limited in scope than
now. In this sense, it is interesting to find out how
some of the presently most reputed universities
started, for example Oxford and Cambridge.
Their origins date back to 1167 when Henry II
ordered all English students on the Continent to
return to England. Many of them decided to settle
in Oxford in an attempt to create a university as
they had seen in Europe. Unfortunately, disputes
between the students and residents led to riots and
a number of students fled to Cambridge, where they

1352

established a new university. In turn, the loss of

students hurt Oxfords local economy and in 1214,
traders and merchants invited them back again.
The competition between Oxford and Cambridge
colleges probably dates back to those times when
William of Durham founded University College in
1249. Balliol was established in 1263 and Merton,
the first residential college, followed a year later.
Merton set the collegiate pattern, which became
standard in both Oxford and Cambridge. These
colleges were self-governing institutions where
teachers and students lived together. Fourteen
other colleges were founded by the end of the
16th century2 (http://www.spartacus.schoolnet.
co.uk/EDoxford.htm).
In another part of the world, the clerices
role was covered by other scholars; it is remarkable that most of them would still be monks,
hermits, or anchorites like Buddhist and Confucian scholars in Asia, or Imams and Ayatollahs
in the Muslim world. Moreover, to most people,
the universities previously mentioned seem to
have had an extremely early start, but if we look
at the records of Nalanda, we find out that this
is actually mistaken.
Nalanda was a Buddhist monastic centre,
often referred to as a university, in the North
Indian state of Bihar. Though it is traditionally placed in the time of the Buddha (6th5th
century B.C.), archaeological excavations date
its current foundations to the 5th century A.C.; it
housed a population of several thousand teachers
and students. The topics covered spanned logic,
grammar, astronomy, and Ayurvedic Buddhist
medicine; the Chinese pilgrims Xuanzang and
Yijing provide vivid accounts of Nalanda as
early as the late 7th century. Nalanda continued
to flourish through the 12th century and became
a centre of Buddhist religious sculpture, but was
probably sacked during Muslim raids around
1200, and has never recovered since (http://www.
ayurveda-tcm.com/Buddhist_ayurveda_monastic_college_Nalanda.htm).

How Technology Can Support Culture and Learning

In China, the First Emperor (259 B.C.) established a process to formalise a set of hierarchies
and exams related to the acquisition of public roles
(becoming a Mandarin or an emperors officer). In
similar fashion measures, rules, laws, and even the
way of writing were standardised and formalised
so that the same characters are still in use. Since
then, in China, culture has always been highly
valued; there is even a common saying related to
this that translates as a word is worth a 1,000
pieces of gold.
The origin of this common Chinese saying can
be found in the famous book Shiji (or Historical
Records) by the historian Sima Qian (around
145-87 B.C.). He describes an episode from the
time when L Buwei (the natural father of the
first emperor) was prime minister. L Buwei
was hosting and protecting over 3,000 scholars,
so he compiled the best of their writings in a
book, claiming that it was the encyclopaedia of
his time covering science, literature, and all the

knowledge that was under heaven. Actually, the

prime minister did more; he had a collection of
texts engraved onto rock and an edict proclaimed
stating that: Whoever would have been able
to improve all that is collected and written by
simply adding or removing even a single word
would have been granted a 1,000 pieces of gold.
For fear of offending the prime minister, a very
powerful man, no one dared to attempt this, but
since then it was clear why even a single word
would be worth a fortune (http://www-chaos.umd.
edu/history/ancient2.html, http://www-chaos.
umd.edu/history/imperial.html#first, http://wwwchaos.umd.edu/history/toc.html)(Mah, 2003).
Coming back to Europe, culture and education
started to spread only with the rise of the middle
classes and the development of the publishing
industry, even though remaining the prerogative
of a minority. Only in the last century, well in
line with Maslows theory (Abraham, 1943), did
education become a patrimony for everyone.

Figure 2. Evolution in educational models and results in relation to technology innovation

1353

How Technology Can Support Culture and Learning

This is mainly due to the social and economic

evolution that occurred after the industrial revolution, bringing better living conditions.
In the process of providing access to education
and culture, technology had a crucial role. For
example, the evolution in educational models and
achievable results (in terms of educational objectives) in relation to technology innovation, along
with major driving forces, is presented in Figure
2. In the diagram, the role played by economical
forces as a push factor (companies demanding new
learning services in order to have a more highly
skilled work force) is shown. At the same time
there were also socio-political forces pushing for
more flexible and decentralised approaches, both
in work and in education, to keep on track with
societys evolution.
In all this, technology was playing a twofold
role. On one side, it was a driving force providing solutions to issues and needs arising from
market and production environments. On the
other side, many spin-offs of this evolution were
also fed back into everyday life. Education and
training could also benefit from technologies
by providing new tools and methods to convey
content and knowledge; thanks to the introduction of factors like simulation and interactivity at
first, and then collaboration and personalisation,
dramatic developments were achieved. At present, the new evolutionary step will likely be lead
by the effective combination of these factors and
the introduction of new technologies.
To better understand this, we will look at the
origin of open and distance learning (ODL). This
began in areas where the educational system was
facing huge challenges due to distances and the
spread of population over territory, combined
with a growing demand for access to education,
as in the USA, Australia, and Canada. To meet
the requests (coming from all levels of society)
for a higher level of education/training, a form
of distance learning needed to be developed.
This new learning process was initially supported
by regular mail, then by radio3 and television.

1354

Then tools initially developed to support totally

different scopes, like the Internet4, started to be
used for scientific and educational purposes. The
initial spread of data among researchers fostered
the exchange of knowledge; this brought on the
rapid evolution of the so called wired-world
that nowadays plays a crucial role in education,
yet all previously mentioned media are still in use
depending on needs and available budgets. As a
matter of fact, available solutions cover almost all
forms of training and teaching, from traditional
class learning to e-learning.
Early examples of e-learning were the computer-based training applications often referred to
as CBTs, mainly developed to foster professional
(re)training. Then low-cost computers were developed, with reasonably good performance, and
able to sustain multimedia applications and made
available to the mass market. In more detail, it was
the advent of the CD-ROM (mainly seen in this
case as a mass storage device), along with easy
access to the Internet that fostered the launch of the
first mass-market-oriented CBT (mainly devoted
to language teaching). Initially, products were
conceived in a stand-alone fashion. Then hybrid
versions (CD plus Internet) emerged to achieve a
higher level of modularity and flexibility (including periodic product update), and now there are
many online training courses.
According to IDC, e-learning means educational content, learning services, and delivery
solutions that support and enable Internet-based
learning (IDC, 2002). Thus, e-learning is a
subset of what is defined as technology-based
training (TBT) that includes CD-ROM and other
technology-delivered training. In turn, TBT is a
subset of what is called enhanced learning, which
includes support to disability for learners, and so
on. Finally, enhanced learning is a subset of the
overall training and learning including instructor-led training (ILT) and text-based training, as
shown in Figure 3, which elaborates on the basis
of concepts exposed from the already quoted IDC
report (IDC, 2002).

How Technology Can Support Culture and Learning

Figure 3. Relation among the various forms of

learning
eLearning
Technology-based training

Enhanced Learning
All Training & Learning

According to Tomita (1980) and Neuman

(1991), the scenario presented in Figure 4 shows
evidence of a media gap. Blogs, Wikis, virtual
communities, and many other forms of communication and social/educational experiences are
quickly arising and filling this media gap.
The availability of new and more pervasive
technology support, as well as the growth and
spread in access to culture and education, is ultimately addressing the issue of a lifelong learning
process to avoid knowledge obsolescence, especially in environments like science or technology.
We must bear in mind Novak (1998), when saying
that the rote-meaningful learning continuum is
distinct from the reception-discovery continuum
for instruction (see Figure 5). As a matter of fact,
it is not enough to have availability of information
and technology that can assist in accessing the
related knowledge.
A common mistaken notion is that inquiry
studies will assure meaningful learning. The
reality is that unless learners possess at least a
rudimentary conceptual understanding of the
phenomenon they are investigating, the activity may lead to little or no gain in their relevant
knowledge. In this respect, it is worth noting

that this aspect has been taken into account in

the e-learning sectors and therefore, cognitive
modelling and ergonomics have been closely
investigated (Sussman & Sedivy, http://www.
umass.edu/psychology/div2/eyelab/).
Studies on cognitive architecture have been
carried out by taking into account perception,
cognition, action (motor), and memory (Hornof &
Halverson) as well as studies on eye movements
and visual attention (Kowler). These results are
extremely interesting and promising, as the use
of these technologies could be very successful
if employed to foster the learning process and
support full accessibility, regardless of gender,
race, language, and impairment. The introduction and combination of wireless communication,
agents, adaptive user profiling, virtual reality,
and eye-tracking (http://www.vr.clemson.edu/
eyetracking/, http://www.nottingham.ac.uk/education/maps/eye.html) with artificial intelligence
and other technologies can move the frontier for
e-learning applications exactly in this direction.
The use of multimedia is probably one of the
most relevant approaches to properly support education in all environments. It is an excellent way
to support teaching activities, not only because
it enables one to enrich the quality level of the
service, but also because it allows one to attract
the learners attention, and stimulates the mind
from several perspectives. However, this is not
enough unless the effort involved in overcoming
accessibility problems and impairments in the
learning process is taken into account. To achieve
this result, the accessibility issue has to be addressed in the design phase and the available and
emerging technologies, regardless of their origin
and major field of application, have to be fully
exploited. In other words, to examine carefully
how accessibility-related needs, in general, and
impairment could be tackled and use whatsoever
technology that could fit to solve, or work-around,
the problem so as to grant full access to education
and culture and therefore, also greatly contributing to societal development.

1355

How Technology Can Support Culture and Learning

Figure 4. The media gap (Adapted from Neuman, 1991)

Figure 5. Relation between learning autonomy and learner satisfaction in relation to media richness
and bandwidth (Adapted from Novak, 1998)

1356

How Technology Can Support Culture and Learning

As stated, in theory (as well as in practice), it

is possible to devise usage of technologies presently applied in other sectors in the learning field
to empower better and more compete forms of
e-learning. As an example, just think how eye
tracking has been exploited in the military environment to control firing systems5, and how it is
presently used to support reading and comprehension exercises and other learning activities, but
still well below its huge potential.
Another example may be the adoption of tools
and methods used to teach diction, or opera singing, in the language learning environment (at least
as far as pronunciation is concerned), especially
when teaching languages like Chinese or Japanese
to westerners or English to easterners. In this case
there is a real need to learn things such as breath
control, tone usage, phonation, and so on, just as
it is done for opera singers or for actors.

Language learning
Most professional sectors have adopted technologies to support the learning process. Probably the
first one to do so was the professional training
sector (pilots, train conductors) and language
learning. Later came the informatics sector (programmers, EDP operators), then the finance and
banking sector (initially jointly with the informatics sector, then on its own, due to the increase in
complexity of the environment itself), and lastly
the medical training sector (initially in imaging
and diagnostic to progressively shift towards a full
coverage)6. This section focuses on the languagelearning environment as an example, given that
it represents a case where technology has already
been widely adopted to properly support learning
processes over a long period.
It is possible to find self-learning materials
based on a widespread set of technologies (from
vinyl audio support, to tapes, and now CDs or
DVDs) and furthermore, many different learning

methods have been adopted to support different

classes of learners (businesspeople, primary and
secondary students, home users, and so on). This
illustrates the relation between e-learning and
technology as a support for culture and education.
It is widely accepted that to master a language,
the four linguistic skills (speaking, listening,
reading, writing) have to be acquired , but the
social and cultural background also have to be
understood. Therefore, the study processes can be
enhanced through the guided use of contemporary
media such as newspapers and magazines, film
and television, video, CD-ROMs, and Internet.
Certainly, literature is an important part of the
study of language and culture, especially at the
more advanced levels, as it constitutes the most
versatile and complex of language usages. Literature essentially represents a compendium of
the past and present culture(s) of the language in
which it is written. Therefore, it also represents
the most difficult part to master when learning a
foreign language as a second language.
When using a foreign language, words will
turn out to be cages for concepts the learner
wants to express. Unless the learner is natively
bilingual, thinking directly in another language
may be achieved only partially. For some people
it is easier to learn the sound of the word/sentence
and to note it down in their own language. This
depends on many factors, starting from the degree of knowledge and mastery of the meanings
associated with a word in a specific language, to
end with the personal cultural background that
may bias the nuances given to a specific word or
expression (just think of all forms of professional
jargon and disambiguation needs encountered
when translating a word or a sentence out of its
context). Many problems, misunderstandings,
and issues have been caused by improper usage
of words among people. Therefore, there should
be a basic concern when teaching/learning a
foreign language:

1357

How Technology Can Support Culture and Learning

Concepts are expressed by words. If certain

English words are missing in Chinese, it follows
that the concept expressed by those words will be
absent in China, and vice versa. (Mah, 2003)
When taking all this into account, it is evident
that mastering a language is much more than
learning its vocabulary and grammar; it implies
also learning the culture behind the language;
the history, the traditions, the costumes, and the
related ways of thinking. In other words, all meanings, structures, metaphors, sounds, and tastes of
a language need to be learned, in the same way a
child from the country in question does this. This
lesson has been used by scholars working in the
language-learning domain, and is progressively
being widely adopted in regular language learning
in primary and secondary education. For example,
as far as the sound of a language is concerned,
exploiting songs, and their lyrics, has become a
common way to support teaching, as is evident
from the following quote by LeLup and Ponterio
(LeLoup & Ponterio, 2001):
Most foreign language teachers enjoy studying
song lyrics as authentic text in their classes. Songs
can be used at all levels and for a wide variety of
activities and purposes such as comprehension,
vocabulary introduction, illustration or recognition of grammar structures, and reinforcement
of topics. Traditional or new childrens songs,
musical classics, or the latest pop hits are all fair
game. The rhythm and melody of songs can make
the words and expressions easier to remember
and more enjoyable for students than other sorts
of texts.
Unfortunately, teaching the culture behind
a language is a long and difficult task, usually
unfeasible in the timeframe of regular classroom
teaching during primary and secondary education. However, this issue is being addressed more
frequently than in the past, and is also often ad-

1358

dressed by using e-learning in a lifelong learning

approach. Furthermore, there exists class-based
intercultural training devoted to specific audiences. Nowadays, learning words, sentences, and
polite phrases in a foreign language, mainly to
manage human/business relations when abroad,
is becoming a common habit among far-sighted
businesspeople, and therefore, it is essential to
also teach proverbs, phrasal verbs, idioms, and so
on. In the diplomatic environment, this has a long
tradition; for example, in China, the first emperor
studied Tibetan costumes and the language, to be
able to properly welcome the Dalai Lama7 (Mah,
2003). Similarly, the Jesuit missionaries in Japan
or China would spend long years in training before being sent to their final destination. During
their training, they would learn how to behave,
talk, act as if they were locals, and this training
was finalised to achieve a specific purpose: the
conversion of the locals to Christianity. In this,
they reached such a level of perfection that Matteo
Ricci is accounted among the classical Chinese
authors in maths, geography,8 and other topics,
despite the fact he was an Italian.
The power of multimedia-based e-learning
lies in the possibility to convey several media and
all the related content richness to the user (in a
single framework). For example, on a DVD, it is
possible to convey as much data as in an ordinary
encyclopaedia, with high-quality audio and good
quality video. The Internet allows the updating
of e-learning sources and tools in a combination
of self and assisted teaching, making it possible
to learn at ones own pace and having tutor support only when needed. Synchronous activities
can be performed along with asynchronous ones.
Reinforcement tools are available, and all this is
virtually usable by the learner.
There is a clear need to stress the word virtually, as disability support can be provided as well,
but if this is lacking, then all this power may be
lost for a relevant part of the population. From a
technical point of view, it is possible to take ac-

How Technology Can Support Culture and Learning

cessibility (in the wider sense) into consideration

from the very beginning, thus ensuring the widest
possible coverage of user needs (from impairment
to aging support). Nevertheless, this implies that
proper development tools and strategies have to
be used.
The outcomes from a 2001 European Commission open workshop to examine the future of
e-learning expected the research focus to be on
virtual and remote labs, emotional characters,
handwriting recognition, and eye tracking in elearning usability. All this should be combined
with situational-based training and mobile access
to resources so that: Research on eLearning will
support the development of innovative models for
the provision of learning services, fully exploiting
the potential of ambient intelligence technologies,
enabling ubiquitous, interactive, personalized
and tailored access to learning and knowledge
on demand to individuals at home, at work or on
the move. It will build advanced learning schemes
for schools, universities, the workplace and for
lifelong learning in general, reflecting the needs
of the knowledge economy. In other words, the
pillars of the knowledge society and economy
are ubiquitous access to personalized learning
throughout life for everyone.

Traditional and Innovative Language

Learning
Traditional class-based language courses are
structured according to a well-codified approach
that regards grammar and vocabulary to be the
core of the study. The main differences that are
worth noting in class-based learning are mainly
related to target audience and objective of the
class. Basically, we can distinguish the following
categories of class-based learning:

Basic institutional curricula (primary or

secondary education level).
Advanced institutional curricula (post secondary education level).
Specifically oriented curricula (business,
summer schools).

For each of these, there are specific programmes, schedules, and methods aimed to
maximise the training outcome.
Except for literature or business-oriented
courses, dialogues and drill downs are considered
as complementary or support activities, just like
exercises. Therefore, dialogues are usually linked
to curiosities and hints on culture and civilization

User proficiency

Figure 6. Language course levels according to several classifications

1359

How Technology Can Support Culture and Learning

related to the language being studied. Classes are

usually organized in terms of language knowledge
prerequisites that are assessed via admission texts
(this applies only to noninstitutional training).
Yet even when assessment is not performed in
advance, content and course structure reflect a
similar approach with the presentation of the most
difficult/uncommon concepts, authors, or issues
at the most advanced level. A gradual approach
to learning has always been adopted (see Figure
6), and is reflected in the courseware and content
structure. This is certainly a harsh generalization,
but still reflects a good part of the present classbased offers on the market.
The levels reported provide indications according to different classification approaches:
UCLA, Berlitz (http://en.wikipedia.org/wiki/
Berlitz_Language_Schools), common European
framework of reference for languages (http://culture2.coe.int/portfolio/inc.asp?L=E&M=$t/2081-0-1/documents_intro/common_framework.

html), and Interagency language roundtable

(http://www.utm.edu/staff/globeg/ilrhome.shtml,
http://en.wikipedia.org/wiki/ILR_scale). Usually
the passage from one level to the other is bound
by the accomplishment of a specific (and in most
cases well codified) exam. The schema also visualises the usual decline in participants from the
basic to the more specialized training levels that
are often not only more demanding in terms of
prerequisites, but also in terms of competences
and capacities of the trainees. Inside each course
level, the typical learning unit could then be
structured as follows (Figure 7).
This is a harsh schematisation, as the more
advanced the course is, the less such a grammarcentric schema is followed, as there is a natural
shift towards a more content-centric schema that
is best tailored to the form of advanced learning
provided. In fact, while the grammar-centric
schema perfectly suits certain pedagogical purposes (like teaching how to teach a language), it is

Figure 7. Typical structure of a grammar-centric learning unit

(1)

Grammar
Vocabulary
Examples

(2)

Dialogues
Exercises

(3)

Grammar
Listen & practice

Here level is referred to the kind of

exercises and not to the overall
course that may be beginners, basic,
intermediate, advanced, or
proficiency.
The student has to listen to a
sentence and then repeat. There is not
much interaction except repetition.
The sentence is first heard, then
repeated with blanks that the student
should fill, and then repeated again.

Beginner (1)
Dialogues

Listen & repeat

Drill down

Advanced (1)
Culture

Listen & repeat

Society

Listen & complete (3)

Costumes & traditions

1360

(2)

How Technology Can Support Culture and Learning

not suitable for others, such as teaching literature

or business/professional usage of a language.
Obviously, a full range of intermediate models
and combinations are also possible in order to
accommodate the various needs and objectives
of the training. A very similar approach has also
characterised the evolution of e-learning-based
language training content that was initially following the grammar-centric and is now shifting
towards the content-centric unit. One of the main
reasons for this shift also lays in the blended
approach to teaching that combines traditional
class-based to self-paced learning.
The difference in structure between the two
approaches is quite apparent. Only a few products
on the market are offering the user the chance
to listen and modify voice samples or dialogues
at playback speed. Some of the most effective
online language courses, for Latin-based speakers willing to learn foreign languages, focus on
reading/listening comprehension, grammar, and
vocabulary. Users are provided with feedback on

achieved performances, and exploit a pedagogical

philosophy based on the following principles:

The approach is student centred (user

chooses own learning path).
Each course level revolves around a central
theme.
Activities have a pedagogical rationale.
The content is interesting and related activities are challenging.

Usually language training is divided into at

least four different levels organized in about six
courses. On average, each course is composed
of more than 10 units plus a unit test. Units are,
in turn organized, in several stimuli alternated
with an average of five activities per stimulus plus
tests. The course becomes progressively harder
while progressing through units. For e-learning
solutions foreseeing the adoption of CD/Internetbased content delivery, it is worth limiting the
duration of each learning session to around 15-20

Figure 8. Typical structure of a content-centric learning unit

A sentence is heard and repeated. Speed
is partially adjustable.
(2)
A sentence is heard as is and afterwards
with blanks.
(3)
Focus is on the usage and understanding
of the language, including slang...
(1)

Dialogue
Understanding &
comprehension
Related grammar
Exercises

Beginner

Drill down
Culture

Listen & repeat (1)

Society

Intermediate
Listen (2)

Costumes & tradition

Slang & phrasal
(4)

(3)

A sentence is heard with blanks and

the student should complete it;
afterwards the sentence is heard as
it should have been.

Listen & complete

Advanced
Listen & complete
Listen(4)

1361

How Technology Can Support Culture and Learning

minutes for usability reasons, as schematically

described in Figure 9.
On average, each stimulus is accompanied by
a set of activities focused on different subskills,
related to the various abilities, plus grammar and
vocabulary. Listening is taught using slideshows,
videos, or pictures with captions, while reading
comprehension, and/or grammar are taught using texts.
Test units are presented at the end of a set
of units aiming at providing feedback on what
has been learned. Units may be repeated several
times and should last no more than 30 minutes
each. The structure will comprise a set of stimuli,
each followed by a set of questions. At the end
of the session, a score is provided as feedback.
The score provided is in the format X out of Y
correct answers (i.e., 3/5). If the same exercise

is repeated several times, only the best score is

reported. There may also be entry test units focused on comprehension and containing at least
three graded texts. They should last no more than
20 minutes, be timed, and drawn from a bank
containing several versions of the test (at least
three). These entry units assign both a level and
a course to the user upon completion, based on
the decision schema reported in Figure 10.
According to the SCORM standard, a tracked
course expects the user to complete the full unit
before being granted the chance to proceed any
further. Thus, if there are exercises for knowledge
assessment, they can usually add conditions to
the overall navigation procedure, for example,
preventing the user from accessing any other
section of the content until the present one has
been completed and related tests passed positively.

Figure 9. Typical fruition scheme of a learning unit (innovative case)

Activity
Activity
Activity
Activity
Activity
...

Stimulus

15-20 minutes

Stimulus

Activity
Activity
Activity
Activity
Activity
...

15-20 minutes

Time

Figure 10. Typical learner access level evaluation scheme of a language course

Level 2 stimuli (2 stimuli, 5 question each)

8/10 ok
Level 3 stimulus with 5 questions
4/5 ok
0-3/5 ok
Level 4

1362

Level 3

0-7/10 ok
Level 1 stimulus with 5 questions
4/5 ok
0-3/5 ok
Level 2

Level 1

How Technology Can Support Culture and Learning

In general, content should be interesting and

appealing. Tests should stimulate user curiosity
and willingness to access the available content,
providing exercises for the issues that generate
most interest.
From the information presented, the structure
of a typical learning course can easily be derived
(see Figure 11). Please note that in most environments, the courseware structure is structured
in levels according to the European Council
Curriculum and to Flesch-Kincaid Grade Level
Score, or to Flesch Readability Ease Score (at least
as far as the English language is concerned).
Over time, several approaches to language
teaching have been developed, starting from
the traditional ones centred on grammar and
stemming from the study of Latin and classical

Greek. This evolution has been led by the following factors:

Recognition that live-languages (currently

spoken) differ from dead-languages and
are constantly evolving; therefore, Understanding & Comprehension should come
first, and the implied grammar could then be
studied (and in some cases even derived).
Spoken languages often rely on a smaller
set of vocabulary, and the most common
phrases structure is simpler. The usage of
verb tenses and cases is also typically more
limited in spoken language than in literary
usage.
Recognition of different needs for different
audiences (business or technical trainee un-

Figure 11. Typical navigation scheme of an e-learning language course structured in levels
Target language selection

Entry Test

Language course main menu

(if the user took the entry test, the system will suggest a recomended course and therefore associated path - named "default")
Course 1

Course 2

Course 3

Course 4

Course 5

Default path

Course 6

To next course reccomendadtion

Unit 1

Unit 2

Unit 3

Unit 4

Unit 5

Unit 6

Unit 7

Unit 8

Unit 9

Unit 10

Unit 11

Unit 12 (test)

Remedial path (following low scoring)

On-line help

User results
Tools

Services

1363

How Technology Can Support Culture and Learning

dertake a learning process to be able to communicate, usually in short timeframes).

Provision of situational training better supports the retention of acquired knowledge
than the extensive study of grammar and
vocabulary.

A good example is the Berlitz method,

or other derived ones like inlingua (http://
www.inlingua.com/English/Services/main.
asp?RefNav=rnav3). It is now clearer why there
has been a progressive shift from a grammar-centric to a content-centric approach. This process
greatly contributed to the need for effective and
efficient self-learning9 that is now converging
towards a blended approach, where the self-led
activities can be even performed on the move,
with the support of specifically designed tutoring
assistance. This can be summed up in terms of
distinctions applied to language learning depending on learning scope, the institution providing
the training, trainees backgrounds, and domains,
as illustrated in the Table 1.

Language Labs
The setup of language labs dates back to at least
the 1970s, when language teachers would use this
infrastructure for practices and for assessing students progress, especially in two of the four abilities, namely listening and speaking, while reading
and writing ability would be aligned with class
work. The most frequent usage of the language lab
would be either in listening and comprehension
or in listening and repeating. Often (but this
was much dependent on the teacher) in language
labs, students would encounter audio sources that
could span a wide range of topics, from read-aloud
literature to radio programmes conversation, or
even music (folk, traditional, pop).
In the late 1980s and early 1990s, the traditional
language lab started to be sidelined by multimedia labs (initially devoted to purposes other than
language learning). Only in very recent times
has there been a convergence of language and
multimedia labs into a single entity, a multimedia
language lab, even though it is still quite popular

Table 1. Distinctions applied to language learning

Environment
Basic language
learning
Advanced learning education

Professional
learning education

1364

Provider
Learning objective
Primary and /or secondary Acquisition of the basic
education;
skills (read, write, listen,
Specific language schools. speak)
Linguistic refinement;
Interpretation schools;
Literature;
Language universities;
Language teaching;
Specific language schools.
Situational training.
Interpretation;
Translation;
Interpretation schools;
Linguistic
refinement;

Language universities;
Literature;
Specific language schools.
Language teaching;
Situational training.

Target audience
All categories of trainees;
Businesspeople;
Professionals.
All categories of trainees;
University students;
Businesspeople;
Professionals.
Interpreters;
Translators;
Dubbers;
Businesspeople;
Professionals (legal, science,
technical or medical).

How Technology Can Support Culture and Learning

(especially for certain forms of studies) to retain

the classical configuration.

Classical Language Lab

Configuration
The classical configuration of a language lab is
based on a set of audio equipment (tape recorders eventually combined with auxiliary inputs)
usually divided into three categories: the teacher
unit (the most richly featured and complete unit,
comprising several inputs), the student unit
(limited to a recording unit), and a set of supportive infrastructures (comprising the switch
and control board available to the teacher). The
control console and the switch and control board
are physical, even though there are implementations using a computer-based approach (Figures
12, 16 and 17).
A classroom may contain only a few workstations, depending on the target usage, availability
of room, complexity of cabling, and costs. The
control unit available to the teacher is usually ana-

Figure 12. Typical structure of a classical language lab configuration

logical (in the most advanced perhaps computer

based) and so is the student unit. In the latter case,
the schema becomes the one of Figure 13.

Multimedia Language Lab

Configuration
The configuration of a multimedia language lab
is based on a combination of computers and a set
of equipment (including tape recorders combined
with auxiliary inputs). Additionally, in this case,
the whole setup is divided into three categories:

The teacher unit

The student unit
A set of supportive infrastructures

While in the traditional language lab the

control unit available to the teacher is usually
analogical, in multimedia labs, the control unit
is implicitly computer based. Moreover, in multimedia labs, it is often possible to find equipment
and applications devoted to a wider set of usages

Figure 13. Typical structure of a computer controlled language lab configuration

1365

How Technology Can Support Culture and Learning

than language learning, as they were originally

created to support ICT teaching and then extended
to support language learning.

Presently, multimedia labs are also being

provided with extensions enabling the usage of
e-learning solutions. This allows for a varied set

Figure 14. Typical structure of a basic multimedia language lab configuration

Figure 15. Typical structure of an extended multimedia language lab configuration

1366

How Technology Can Support Culture and Learning

of possible configurations and setups, like the

following ones (see Figures 14, 15 and 16).
As is apparent from the previous schemas, the
usual structure is based on a network connecting
the individual positions, each equipped with the
specific sets of supportive tools needed, comprising specific HW (recorder) and SW (applications).
The class could be either fully physically bound
(all work places are in the same physical location) or dispersed (not all work places are in the

same physical location, but some are remotely

connected); furthermore, the class could also
be connected to external resources (like a VOD
server, Internet, and so forth.). Audio/video conferencing equipment can also be used to support
open and distance learning.
Additionally, in this case the overall arrangement will be then complemented with a proper
layout that will provide two different sets of
workstations for teachers and students. Again, a

Figure 16. Typical structure of a multimedia language lab class and teacher work place

Figure 17. Typical structure of a multimedia language lab teachers console

1367

How Technology Can Support Culture and Learning

classroom may contain only a few workstations,

depending on the target usage, availability of
rooms, complexity of cabling, and costs of the
overall setup, as in the following images.

(storyboards, drawings, scripts, books) that in

certain cases will have their own existence and
production cycle, which is reported in Figure 19
with the same approach of Figure 18.

Content Production

Authoring Tools and E-Learning

Content Development Applications

In the publishing environment, the content production chain follows well-codified and standardised
processes that have been developing over several
years. This process, and the actors involved, are
schematically reported in Figure 18.
Cost control and overall IPR and copyright
management are usually undertaken in a parallel
stream, to ensure constant monitoring of highrisk factors. As far as the economic impacts are
concerned, cost control has to monitor process
development and ensure that it is kept in line with
expectations and budgets, retaining its profitability (or even increasing it whenever possible) and
therefore, is a good part of the standard management process.
On the other hand, IPR and copyright management interactions occur whenever an asset cannot
be cleared and therefore, has to be replaced. This
event may occur at any step, and the impact may
be marginal or relevant, depending on a set of
possible combinations of factors:

The relevance for the publishing project of

the object that could not be cleared.
The reason for lack of clearance.
The stage of the publishing process.
The availability of a replacement/equivalent
object, and so forth.

This refers to traditional, desktop and multimedia publishing as they present the highest
number of contact points and overlaps.
A similar approach applies to the editorial process of any product, yet digital TV, serials, movies
and in general products requiring video production
are somehow different as the design and planning
phase is much longer and has several by-products

1368

Standard document creation tools rank at the top

of the list of utilities used to create e-learning content. Most of them are oriented to achieve specific
results and do not require specific programming
skills, even if they may have extensions, they are
therefore often used in combination.
According to Brandon and Hall, over 40%
of companies use an e-learning solution to provide services to their own personnel, including
language learning, mainly due to globalisation.
Most e-learning solutions adopted are used to
provide self-paced e-learning courses and live
e-learning sessions using virtual classroom applications, while tracking results using a learning
management system (LMS).
Unlike conventional training, a good instructor
cannot rescue bad online learning; the materials successful implementation is decided by the
quality of the design. Therefore, in e-learning,
content and design have remained the most critical dimensions of success. Not surprisingly, the
shortage of well-designed, engaging, and relevant
e-learning products is still high on the list of
reasons for limited acceptance of e-learning in
many organizations.

A Contextualised Example of
Content Development for E-Learning
To provide a better understanding and to highlight
the benefits of the adoption of certain technologies
as a support tool for the production of educational
content, we examine the process followed in a
successful European RTD project conducted in
IST-FP5 dealing with the application of technologies developed for the support of dyslexia to adult

How Technology Can Support Culture and Learning

Figure 18. Content production chain in the publishing environment

Activity

task

Involved roles
Management
Authors
Chief Editor
Press Office

Idea
Market survey
Title design
Go / No Go decision based on market data and production cost analysis to ensure the
expected return on investment

Research of:

Sources
References
Contacts
Multimedia

Similar titles

Management
Authors
Editorial board
Editorial staff
Press Office
Legal Department

Draft acceptance
(if positive the next step starts if not the previous is
reiterated).

Editing of:

Texts
Notes
Indexes

Multimedia
Captions

Management
Authors
Editorial board
Legal Department

IPR/ clearance

Product final acceptance

(if positive the next step starts if not the previous is
reiterated)

Finalisation of:

Chief Editor
Editorial board
Editorial staff
Instructional designer
Production department
Press Office
Legal Department

Texts
Notes
Indexes

Multimedia
Captions

Books &
Magazines
CD/ROM
DVD
Web
TV, iTV, PDA
mobile and other
new media

Authors
Chief Editor
Editorial board
Editorial staff
Instructional designer
Press Office
Legal Department
Authors
Chief Editor
Legal Department

Formal authorisation to start production

(if positive the next step starts if not the previous is reiterated. In this case IPR/
clearance should have been completed if not process may be suspended / stopped /
cancelled)

Production

Management

Cost
control

IPR/ Contracts
management

Management
Chief Editor
Editorial board
Legal Department

Production department
Outsourced service
Press Office
Marketing manager
Legal department
Company accountant

Marketing

Promoting

Marketing manager
Legal department

Distribution & rights selling

Revenue
management

Marketing manager
Legal department
Company accountant

1369

How Technology Can Support Culture and Learning

Figure 19. Content production chain in the media environment

Activity

Idea

Title design

Involved roles

Storyboard drafting

Cost estimation

Producer
Production accountant

Go/No Go decision

Producer

Storyboard finalising
Scripting drafting
Casting preparation

Design &
preparation

Casting
Scripting
Lighting drafting
Shooting plan drafting
Effects drafting
Costumes drafting
Sound track drafting
Contracts finalising

Development &
shooting

Storyboard adapting
Lighting management
Shooting plan adapting
Effects management

Post Production

Mounting
Sound addition
Effects addition
Packaging

Distribution & rights selling

Author
Script editor
Producer
Director
Casting director
Art director

Cost
estimation
refinement
Cost
estimation
refinement

Distribution & sale

IPR/ Contracts
management

Distribution & sale

IPR/ Contracts
management
Cost control

Marketing

Costume designer
Makeup designer
Composer
Audio Engineer
Graphic designer
Special effect designer
Fight arranger
Lighting cameramen
Production assistant
Location manager
Art director
Director
Casting director
Producer
Production lawyer
Production accountant
Marketing manager

Revenue
management

L2 language learning, named FLIC (Foreign Language Acquisition through the Instinct of a Child)
(FLIC, 2004, 2005a, 2005b, 2006); (MediTECH,
2005a, 2005b); (Lasnier, 2005). The results of the
project, in terms of pedagogical and neurological validation, have been conducted by Sheffield
University, while the original idea and method at
the basis of project development has been based

1370

Author
Writer
Script editor
Art director
Director
Producer
Production accountant

Marketing manager
Production accountant

on the experience gained by MediTECH in respect of the development of supporting tools for
dyslexia treatments.
Given this content production process, it is
apparent that producing content for multimedia
applications, especially in the educational field for
language learning, is a rather complex task per
se; the introduction of new technologies that alter

How Technology Can Support Culture and Learning

Table 2. Documentation tools used with e-learning

Tool used to create e-learning courses content
PowerPoint / Word
Dreamweaver (can be used by specialized and
novice users)
Flash (can be used by specialized and novice
users)
Code at the HTML tag level10
Traditional authoring tools11
Microsoft FrontPage
Learning content management system (LCMS)
Content authoring tools built-in to an LMS
Rapid e-learning development tools12
Other

such a process is often opposed due to the potential cost impact. FLIC proved that the additional
effort required to take into account the specific
solutions that will lead to product improvement
is not excessive if this is done from the product
design phase (FLIC, 2004, 2005a, 2005b); (MediTECH, 2005a); (Lasnier, 2005). For the sake of
reference, we will briefly report here the process
followed in FLIC.
In FLIC, the audio content was using dummy
stereophony (see Figure 20) for recorded sounds.
The dummy head simulates the human head in an
acoustic sense, as far as possible, as in place of
eardrums, the dummy head has suitable microphones whose directional characteristics and other
physical properties correspond to the properties
of human ears (FLIC, 2005a).
The need for this special recording procedure
was strictly connected with the special learning
support provided by the FLIC equipment, namely
the lateralisation (MediTECH, 2005a), the multichannel voice fusion (HW/SW versions) (Medi-

Freq. of
adoption
66% / 63%
61%
47%
34%
32%
26%
21%
18%
13%
16%

TECH, 2005a, 2005b), plus the time alignment

and voice modification (SW version only).
The Lateralisation process is based on a quite
simple principle related to the fact that specific
aural circumstances generate both a higher level
of attention in the listener, and also an instinctive
synchronisation effect. This is achieved in three
phases, namely:

Phase 1: Learner hears model voice from

CD and reads along silently phoneme by
phoneme hearing the model voice wandering from side to side.
Phase 2: Learner silently reads along with
the model voice (mouthing the words), again
perceiving it as wandering from side to
side.
Phase 3: Learner reads aloud in synchrony
with the model voice, hearing his and the
model voice in opposite ears and both wandering from side to side; this automatically
tends to put the two in sync.

1371

How Technology Can Support Culture and Learning

Figure 20. Dummy head and related recording setup

sound sources
headphones

dummy head
with two
microphones

Figure 21. Lateralisation principle: The wandering around voice

The multi channel voice fusion (MCVF) technology divides the audio information of the left
and right canal into eight (8) bands each, without
losing any information during the splitting. The
wave bands are differentiated in the following
way: Band 1 (low-pass filter till 200Hz), Band
2 (250-315 Hz), Band 3 (400-630 Hz), Band 4
(800-1000 Hz), Band 5 (1250-1600 Hz), Band
6 (2000-2500 Hz), Band 7(3150-4000 Hz), and
Band 8 (high-pass filter from 5000 Hz). Consequently, two times 8 canals are generated. The
mutual mixture of the canals causes an overlay
of the separate audio information (left and right)

1372

to a single frequency composition of both canals.

This principle is shown in Figure 22.
If MCVF and its effect are firstly used without lateralisation, while speaking through the
microphone and listening to ones own voice with
and without the MCVF, it is impossible to locate
the direction of the voice, which is precisely the
desired effect, produced by the FLIC-unit for
the benefit of the learner. Furthermore, a timealignment procedure was also implemented (SW
version only) to help the learner better understand
the pronunciation aspects of the reference voice.
The logic diagram of the adopted solution is shown

How Technology Can Support Culture and Learning

Figure 23. Diagram of the time synchronization

system where DTW stands for dynamic time
warping and WSOLA for waveform similarity
overlaps and add

Figure 22. Multichannel voice fusion (MCVF)

technology

proved extremely useful (and was often needed)

whenever the two voices belong to different genders. The result could be achieved with a voice
modification engine (SW version only) that would
operate according to the logics presented in Figure
24, transforming the original (reference voice) to
achieve the result presented in Figure 25.

Technology and accessibility

in Figure 23. Last, but not least, it was possible
to modify the reference voice to enhance the
possibility to understand mismatches occurring
between the learners and the reference voice. This

It was mentioned that technology could support

culture and learning by lowering access barriers. Technology has made access to content and

Figure 24. Speech modification algorithm structure

Recorded speech from

teacher

FLIC
Voice Analyzer

mean rate
mean pitch

Time-scale factor
Pitch-scale factor

FLIC
Voice Engine

Modified recorded speech from teacher

(student's mean rate and mean pitch)

mean rate
Speech from student

FLIC
Voice Analyzer

mean pitch

1373

How Technology Can Support Culture and Learning

Figure 25. Speech modification algorithm results

sources mobile and ubiquitous, and it has also

made it possible to overcome other barriers related
to impairment. This is clear to a good part of the
technical community, but not in totality, nor to
the wider public. Such a lack of understanding
of the importance of accessibility from scratch
represents, at present, a real obstacle to societal
evolution.
To get a clearer picture of the problem, it is
sufficient to note that people with disabilities
number nearly 1 billion worldwide, representing
a formidable force of employees and customers.
In Europe, according to the EU Commission,
people with disabilities comprise about 15% of
the population, and many of them face barriers
when using ICT products and services.
Despite these data, many people still consider
impairment to be a marginal problem and accessibility not a relevant issue. With aging, however,
everyone starts experiencing (to some extent)
impairments, and accessibility is a major issue

1374

The number of elderly people is actually rising

fast, according to UN figures (18% of the population was aged over 60 in 1990; while 30% will be
over 60 by 2030), and very often, elderly people
encounter knowledge or usage barriers when using ICT products and services. This, therefore,
brings back the issues of lifelong learning and
accessibility from scratch.
Surprisingly, the technology required for
achieving accessibility dovetails with technology
required for pervasive computing, as nonimpaired
customers need increasingly flexible, multimodal
interfaces to accommodate IT access under varying environments, and with numerous devices.
Therefore, taking impairment into account in
the design phase of products and services automatically widens the market perspective and the
societal benefit.
In 1998, the United States of America Congress amended the Rehabilitation Act to require
Federal agencies to make their electronic and

How Technology Can Support Culture and Learning

information technology accessible to people with

disabilities. This decision was based on the fact
that inaccessible technology interferes with an
individuals ability to obtain and use information
quickly and easily.
Section 50813 was enacted to eliminate barriers in information technology, to make available
new opportunities for people with disabilities,
and to encourage development of technologies
that will help achieve these goals. The law applies to all Federal agencies when they develop,
procure, maintain, or use electronic and information technology. Under Section 508 (29 U.S.C.
794d), agencies must give disabled employees and
members of the public access to information that
is comparable to the access available to others.
This process has brought about a major societal
change in the USA and the upcoming spread of
e-government solutions worldwide (http://www.
w3.org/WAI/Policy/ http://www.w3.org/WAI/),
and more specifically in Europe (Council Resolution of 25 March 2002 on the eEurope Action Plan
2002) means that the adoption, in the near future,
of procedures and tools to make full accessibility a reality is now a necessity. E-learning will
therefore be no more exception or something only
for professionals (including or limited to training
or retraining). Furthermore, the increase in the
societal understanding of accessibility-related
issues will favour a selection process where accessible products will be chosen over nonaccessible
ones, even overriding the factors of pricing and
brand name recognition.
The combination of constant availability of
information and content (including in education
and training sectors) and the push towards real
lifelong learning is converging towards what is
called mlearning (http://www.webopedia.
com/, http://www.odl.org/, http://en.wikipedia.
org/wiki/M-learning, http://www.pjb.co.uk/
m-learning/articles.htm, http://www.e-learningcentre.co.uk/eclipse/Resources/mlearning.
htm, http://www.academiccolab.org/initiatives/
mlearning.html)(Metcalf, 2004a, 2004b) . In

many ways, m-learning is recapitulating the evolutionary process that e-learning experienced as
it emerged from traditional classroom training. In
the mature e-learning markets in North America
and Europe, mobile e-learning exists side by side
with conventional e-learning as a general customer
option. It is the product of choice over conventional
e-learning for the mobile workforce.
All this leads back to the accessibility issue
in the wider sense; as technology has reached a
point where the objective could be reached to a
great extent; what is lacking is both the application and regulation framework, along with a broad
acceptance and recognition of this need for a real
broad accessibility to content and education.
Previously, we have already provided a definition for e-learning and how it has come to the
stage where, when talking about learning and
accessibility, it is necessary to point out that elearning is one of the terms that has emerged from
the rapidly developing world of the Internet, and
is broadly defined as Internet-enabled learning.
This is usually referring to accessibility in terms
of possibility to access, despite distance and time
constraints, to sources, not in terms of impairment
support. This is certainly not a minor issue.
Much expectation surrounds the Internet and
its role in education, as e-learning can contribute to
the improvement of standards and to the effectiveness of teaching and learning. Yet when claiming
this, almost no reference is made to impairment
support. Most Web-based content producers are
often not even aware of W3C guidelines for accessible design of Web-based applications. The
reason for this can probably be found in the fact
that Internet access has generally only been available to society as a whole for the last 10-15 years,
and to schools in the last 2-5 years. Such access
is able to provide:

Support for needs and individual learning

requirements of both children and adults.
A means of empowering children, helping
them to learn more effectively.

1375

How Technology Can Support Culture and Learning

Access to vast volumes of information,

mostly free.
Communication tools and facilities that can
be used on different levels.
Access to content and learning processes
that used to be confined to schools and colleges.

Summing up, the availability of Internet access provides a challenge to traditional forms of
teaching and learning, and opens up opportunities
previously denied to the majority. Yet all aforementioned factors are usually taken into account
regardless of proper accessibility and usability
related issues (in the sense that most content
available on the Internet may fail either to meet
accessibility issues in respect to regulations like
the U.S. 508 one or similar, or usability criteria
for certain user communities like elderly people
or people with cognitive impairments).
This has happened largely due to the most
widely adopted and common definition of e-learning. According to the IDC report E-learning:
The Definition, the Practice, and the Promise,
we can say that a good, working definition for elearning (and in more general sense of e-anything)
is electronic or Internet-enabled. Internetenabled learning, or e-learning, strictly means
learning activities on the Internet. Those events
can be live learning that is led by an instructor
or self-paced learning, where content and pace
are determined by the individual learner.
The only two common elements in this process
are a connection to the Internet (either physical
or wireless) and learning. In 1996, a program for
bringing technology into education was launched
in the U.S. Its primary goals were:

1376

All students and teachers will have access to

information technology in their classrooms,
schools, communities, and homes.
All teachers will use technology effectively
to help students achieve high academic
standards.

All students will have technology and information literacy skills.

Research and evaluation will improve the
next generation of technology applications
for teaching and learning.
Digital content and networked applications
will transform teaching and learning.

Presently, these objectives have almost been

fully met in the U.S.14 Furthermore, according to
the report A Vision of eLearning for Americas
Workforce, developed by the Commission on
Technology and Adult Learning, the global elearning industry comprises approximately 5,000
suppliers offering every imaginable method of
e-learning. The vast majority of these suppliers
are private.
Even if it is generally accepted that the driving
force behind the 21st-century economy has been
knowledge, in many European countries, these
objectives are still far from being realised.
What is driving the market in real time can
be split among present and future trends, while
drivers and inhibitors are illustrated in Tables 3
and 4.
A direct consequence of this is that technology
needs to be used in order to lower costs. This can
be achieved by reusing content (even if usually
an increase of richness is also requested), and
adding value by offering custom solutions and
new approaches (performance support, customer
education, and so forth.). Yet the most important
issue regarding e-learning systems is related to
its effectiveness in transferring knowledge to the
user. Learners comprehension depends strongly
on course structure. Therefore, particular attention
must be paid to effective navigation and lookand-feel features. In addition, content should be
designed, developed, or adapted for multimodal
delivery for effective content reusability. Such
integrated approaches are more likely to have
an impact as the meta-adaptivity system design
domain emerges.

How Technology Can Support Culture and Learning

Table 3. Today and tomorrow (Source: Online Courseware Factory Ltd)

Today
Technology training
Classes for the masses
Instructor centric
Training when scheduled
Time to train
Teaching by telling
Product-based learning
Know what
Skill and information (mastery basics
= 3Rs)
Reactive

Tomorrow
Performance improvement
Personalised learning
Learner centric
Learning on demand
Time to perform
Learning by doing
Project based learning
Know why
Inquiry, discovery and knowledge
basics
Proactive

Table 4. Drivers and inhibitors (Source: Uwe Krueger, Realtech, http://www.realtech.de)

Drivers
Lifelong learning

Inhibitors
Technology Compatibility
Economic turbulence
Limitations
Need for continued education and staff Billing systems
training
Increased Internet / eCommerce usage Security concerns
Price

Tools and approaches to

e-learning
It has been pointed out that at present, there is a
shift in education that is moving from an approach
in which individuals undergo a specific period of
training and then start their professional life to an
approach in which the learning process is a lifelong
one. In this second perspective, companies also
have to tackle the issue of continuous education.

For a rather evident set of constraints, including

but not limited to time and costs, companies have
been always trying to find alternative solutions
to class-based learning. Thus, it is not surprising
that some of the early adopters of e-learning have
been major corporations like IBM, HP, AT&T,
Boeing, and many others. In this respect, it is
interesting to note that according to Brandon
and Hall (Chapman, 2006; Chapman, 2006a
Chapman & Nantel, 2006; Nantel & Vipond,

1377

How Technology Can Support Culture and Learning

2006), organizations eLearning practices are

what turns tools and content into real learning.
Furthermore, to be successful in the design of
e-Learning courses, the following factors should
be taken into account:
1.

2.

3.

4.

5.
6.

7.

8.

Faced problem and expected results:

Matching the e-learning solution to the problem at hand; achieving intended results.
Instructional design and integrity: Structuring, relevance and quality of content;
focus on real-world competencies; selecting the right strategies for the content and
context. This also means focussing on the
intended users.
Evaluation and assessment: Applying
imagination and rigour to the design and
implementation of evaluation or assessment.
Interactivity: Using creativity and expert
design practices to achieve instructionally
powerful interactions of all kinds.
Usability and interface: Creating an effective, easy-to-use interface.
Motivation and aesthetics: Motivating
learners to follow and successfully complete the training; hitting the right tone and
aesthetic notes.
Media and technology: Smart selection and
application of media, development tools and
delivery technologies.
Money and time: Achieving excellence
under constrained budgets and time lines.
This aspect is relevant not only for companies, but also for people acquiring off-theshelf products or online solutions to solve a
specific problem like learning a language or
acquiring knowledge essential for a specific
purpose (re-employing, relocating, and so
forth).

In this process of integrating education into the

employees daily life within a company, several
aspects should be taken into account. For example,

1378

the need to keep certain components of company

know-how inside the company, or how to foster
practical skills acquisition, or how to handle the
mapping of individual characteristics onto a professional job description to then derive, thanks to
a gap analysis, the individual training needs.
Most of these items correspond in practice to
one or more tool-sets that are used by the human
resource department, the IT/ICT infrastructure,
the employees, and the company management to
address the various issues. The most relevant ones
are described next, and could also be used in the
formal education process to improve what is
presently achievable, or to complement it in order
to tackle specific issues.

Authoring Tools, KMS and (L)CMS

Authoring tools are generally desktop, single-user
applications used to construct learning content by
assembling and combining text, graphics, audio,
video, and animations into e-learning courses. Yet
there is also a current trend to have Web-based
development tools. They have been initially
derived by visual HTML editors and are often
used to create standard Web pages, but they can
also be used for creating learning applications.
In addition, there exists a growing number of
authoring tools focused on rapid, template-based
development of learning content, sometimes with
a specific focus on a particular type of learning
application, such as software simulations.
Authoring tools, even though offering a
rather broad range of solutions, represent only
one category of tools aimed at properly editing
and formatting content/knowledge. This is undoubtedly the first step in the process. Yet once
content has been prepared it has to be managed,
and a set of solutions has been developed for this
purpose over time. Notably the most interesting
are represented by:

Content management systems (CMS):

Systems for organizing and facilitating col-

How Technology Can Support Culture and Learning

laborative creation of documents and other

content, using a range of technologies and
techniques, including portal systems, wiki
systems, and Web-based groupware; sometimes Web based, though in many cases,
requiring special client software).
Knowledge management systems (KMS):
Systems for organizing and facilitating
collaborative creation and management of
knowledge and expertise, either at company
or at individual level, using a range of technologies and techniques, including portal
systems, wiki systems, and Web-based
groupware.
Learning content management system
(LCMS): A solution for the creation, management, and transfer of learning content;
it is apparent that LCMS are a subcategory
of CMS devoted to learning content.

E-Learning, ODL, and Lifelong

Learning
Computers are moving out of laboratories and
into classrooms in learning new languages,
understanding complicated math formulas, and
exploring other countries. Technology is changing
basic notions of schools and education and creating classrooms without walls that offer students
a valuable view of the world, and enable them
to experience and interact with other students
and resources around the globe. With numerous
forms of distance learning now available, public
education is moving away from a need for students
and teachers to assemble in school buildings for
education to occur. This trend has far-reaching
implications for the structure of state education
budgets.
According to analysis performed by government and research institutions, the rapid growth
in such areas as distance learning, technology-enabled assessment, and the increasingly diversified
and expanded public-private learning marketplace
require the development of new strategies for qual-

ity and consumers protection. Important priorities

for the public and private sectors include:

Providing reliable and universally accessible

quality information for consumers.
Developing quality assurance mechanisms.
Ensuring that learners have the support they
need to make the right decisions about their
e-learning options.
Developing policies and practices to ensure
privacy.

Traditional, institution-based approaches to

assessment and certification are not well suited to
an e-learning world in which the focus turns from
a record of classes taken and degrees received, to
measures of what an individual actually knows
and is able to do. As a result, private and public
sector leaders need to take steps to create new
approaches such as developing and promoting
outcome-based assessments of learning results,
and creating an electronic system for tracking
those results.
Government and commerce must play a leadership role in making quality e-learning opportunities more widely available to all, from supporting
the development of common technical standards
to promoting broader access in under-served
communities. The challenge and the opportunity
are the same: to realize e-learning potential for
reducing the divide between haves and have
nots. Therefore, it is necessary to:

Create the highest-quality e-learning experiences possible.

Implement new measures and methods for
assessing/certifying what users know and
are able to do.
Ensure broad and equal access to e-learning
opportunities.

The potential return on investment for both

the public and private sectors is enormous. The

1379

How Technology Can Support Culture and Learning

challenge for businesses is to realize the full potential of e-learning as a driver of productivity
and performance gains by making it an integral
part of organizational strategy and operations. For
government, the challenge is to create a nurturing policy environment for e-learning, firstly by
removing barriers that restrict access to e-learning
benefits and, secondly, by promoting industry
self-regulation while balancing citizensinterests
and needs.
By adopting e-learning standards like the one
defined by IEEE, AICC, and so forth, it is possible
to achieve full interoperability of e-learning platforms and to have real, full portability of produced
content. A learning object respecting all features
described in those standards will be available in
whatsoever platform and therefore, full independence from a specific vendor is achieved. Moreover,
it is possible to package objects developed with
different tools and for different delivery platforms, achieving a consistent and interoperable
object, where the tracking will also be kept across
editing, aggregation, and reuse. Other emerging
standards that focus on personalisation are likely
to have an influence in the near future, such as
the internationalization of the IMS AccessForAll
approach in ISO, the Individualized Adaptability
and Accessibility in e-learning, Education and
Training Standard.

Virtual Reality and Multisensorial

Approach
The use of virtual reality is progressively becoming widespread. Usually the primary applications
are either professional training or gaming. In
the first case, the simulation is devoted to enable
trainees to have access to equipment and situations that may be too risky and too costly to
be faced otherwise. At the same time, only big
organisations can presently afford costs related to
simulations. On the other hand, namely in gaming, customers are demanding more and more
in terms of appearance, and the huge numbers

1380

involved allow developers to achieve high levels

of quality, despite the production costs.
With todays technology, it is possible to
achieve results unforeseeable just a few years ago.
In detail, it is now possible to achieve multisensorial stimuli by combining visual, tactile, and sound
experiences into a unique new one experience.
The real challenge is to be able to generate similar
situations (potentially involving even smell) for
edutainment purposes. What is still lacking is
a paradigm to convey all those rich stimuli into
a single mainstream that may be used to both
entertain and educate at the same time.

Virtual Communities and Forums

In many cases, establishing user communities
is an extremely powerful and convenient way to
foster a learning process. Community users will
exchange ideas and cooperatively support each
other in the learning process. This approach has
a long history, and was initially born among
scholars in the scientific domain as a way to
support research results exchange, documents,
and papers peer reviews, and other forms of
interaction. With the wide diffusion of Internet,
this approach has progressively widened its application, and presently there are thousands of
communities, forums, and so forth. Given this,
it is nevertheless essential to note that starting up
a new community or forum is a difficult process.
Table 5 reports the most relevant factors that would
cause user retention or loss.
According to Table 5, actions that help in starting up the discussion in a new-born forum/community can be summarised as follows:

Promote forum existence by direct contact

with a set of people that may be interested
to take part in forum activities and provide
useful cooperation.
Provide interesting threads.
Periodically provide contributions from
experts.

How Technology Can Support Culture and Learning

Select key themes and collect them.

Generate FAQs starting from relevant
threads.
Simplify the process to manage messages
and replies, for example introducing the
concept of a follow-up message that is not
a reply but a related issue.

Another success factor that has to be taken

into account is need. If the users needs to be

Table 5. Reasons for retention or loss within a

new community
Reasons to stay & be
active
The forum is providing
interesting news
The content is complete
and gives a good
coverage of the topic
addressed
It is possible to find
replies to questions or
people able to provide
such replies
Participants are skilled,
knowledgeable and
interested
There is enough
dialogue
There is an historical
track of performed
discussions
There is a service
provision for FAQs
Case studies, white
papers, reference
documents,
presentations, articles
and papers are available
(at least for browsing,
better if downloadable)

part of a certain community, or if the users may

find, in such a community, replies to their own
needs, or even a sense of belonging/ownership,
then community success is almost assured (the
only real threat in that case may be economical
sustainability). An interesting example for this is
the German School Net; up to when it was strongly
supported (in terms of costs and equipment) by
Deutsche Telekom, it grew and developed. It
turned out to be extremely important, full of users
(both teachers and students) and data exchange.
When Deutsche Telekom stopped supporting its
activity, then community subsistence was greatly
endangered. Now with the start-up of the European
School Net, the German portal has also gained
new vitality. Also of note is the case of eSchoolnet,
the European Teachers portal.

Reasons to leave
Lack of a moderator
A number of
participants
imposing their
viewpoints
It is not easy to
understand the
content and type of
mail received from
the forum
Too much mail is
received
Shift in focus
Lack of education
in participants
behaviour

E-learning and the market

To complete the panorama provided so far to
at least the relation presently existing between
e-learning and the market should be addressed.
According to IDC (2002), the e-learning market is
segmented in three main areas: Content, Delivery
solutions, and Services.
Content providers offer courseware core material (like multimedia assets and course structure),
and testing/assessments are included in their expertise. Companies that provide delivery solutions
offer products oriented at course preparation and
organisation, such as training authoring tools and
learning management systems (LMS), including
collaborative software to help course fruition. Services include all additional features that can lead
to better course preparation (content design), and
that are related to course maintenance (management and hosting) and CRM (online mentoring).
Some products/platforms provide solutions in different areas, covering partially or totally the entire
range of e-learning offers. Figure 26 describes the
situation and introduces market players and their

1381

How Technology Can Support Culture and Learning

Figure 26. E-learning market players and their positions

Contents
Cardiff Teleform. CBIQuick,
DazzlerMax, EsaTest, EzSurvey,
Halogen Esurveyor, Intelligent Essay
Assessor, IST Author, Macromedia's
eLearning Studio, Perseus
SurveySolutions, PowerSim,
ProForma, Quandry, QuestionMark,
Quiz Rocket, Rapidbuilder,
RapidExam, ReadyGo Inc, Respondus,
Inc, Riva eTest, SecureExam,
SimCadPro, StatPac, SurveyTracker,
The Survey
System, Trainersoft,
Trivantis Lectora,
WebQuiz

Delivery Solutions

Click2Learn

Blackboard, Centra Symposium,

Docent, InterWise, Jenzabar,
Mentergy's LearnLinc, PeopleSoft
Campus Portal, Prometheus, Saba
Software, SCT's Campus Pipeline,
Total Knowledge Management
(TKM), WebCT

Learn eXact,
Lotus

ExamBuilder,
Link Systems,
Testcraft

HorizonLive

Activetouch Webex, AOL's Instant Messenger, Apple

Quicktime, Centrinity FirstClass Intranet Server, e/pop,
Eproject, eproject.com, eRoom, EZBoard, Facilitate.com,
Finali, Groove Networks, iCohere.com, ICQ, Inovie's
Team Center, Instinctive Technologies E-room, Involv
Teamspace, Jabber.com, Linktivity's WebDemo,
Microsoft Messenger & Exchange Conferencing Server &
Windows Media, Netlert, OpenText Livelink Online,
Peoplelink, Placeware Conference Center, Punch
Networks' Webgroups, RealNetworks RealVideo,
SiteScape WebWorkZone, WebCrossing, Yahoo Groups
& Messenger

Services

positions in the marketl as it appeared at the time

of the previously mentioned IDC study.
Asynchronous Web-based software suites
attempt to group suites of tools that quickly allow
an instructor to convert existing basic electronic
documents to a hierarchical system with relative
ease. Although most of these suites may require

1382

limited HTML knowledge to fully exploit them,

relatively little programming or other technical
expertise should be needed to effectively use
these software suites. Key characteristics of
major players in asynchronous suites typically
include capability for secure student login via
standard Java browser, centralized database-

How Technology Can Support Culture and Learning

centred syllabus with links to internal or external

Web pages, online, time-monitored quizzes with
randomised dynamically-generated testing, discussion groups, and integrated e-mail. Systems
also provide instructor development tools to ease
transition from other media to these products.
The academic user audience is dominated by
two vendors: WebCT and Blackboard, with only
a few additional competitors targeting academic
audiences specifically. Corporate/government
learning organizations have a much larger number
of competing systems.
Synchronous (real time) Web-based training solutions are the most appropriate to facilitate
relatively formal, instructor-led, hierarchical
learning events, such as a seminar or interactive
classroom. A number of these products incorporate
talking head video; most video-enabled products
suffer from poor video quality and tend to require
128KBPS (a.k.a. ISDN class) dial-in connections
for acceptable performance; although a compelling sales feature tending to appeal to instructors
familiar with videoconferencing, at present current talking head video quality over dial-up
Internet connections is not worth the effort; most
students ignore such poor images very early into
the course. Leading members of this class generally have also the following characteristics:

Browser-based (works with Netscape and Internet Explorer, though often PC-centric).
Online testing (designed for use over Internet dial-up connections or corporate
intranets).
Live streaming audio (generally one-way,
often two-way)
Text chat, and occasionally private text chat,
among students or between participants and
presenter.
Sequencing is controlled by a presenterleader; a secondary/copresenter is available
for the higher end products.
Ability to show PowerPoint presentations.

Ability to record sessions and make them

available for on-demand viewing at any
time.
Shared whiteboard, with ability for students
and faculty to import images that can be
annotated on the whiteboard.
Web-page sharing/cobrowsing.
Application sharing ability for presenters
to share applications running on their desktop.
Integrated polling/surveys.
Virtual hand/raising, to indicate when
participants have questions.
Firewall friendly.

Application service providers (ASP) choose

to make their products available as hosted solutions
only; they do not sell the software for organizations to create and run their own servers under
license. This license arrangement may better fit
specific needs, especially where institutions do not
have the in-house capability to sustain tech support teams and servers. Vendors in this category
generally offer the following feature set:

Browser-based: All functionality is provided within a standard browser. Updates

are added automatically.
Presentation sharing: allows any meeting participant to spontaneously share any
presentation (generally PowerPoint).
Document sharing: allows meeting attendees to view content with multilevel zooming
and annotation capabilities.
Application sharing: Run any software
application for effective live demos and
training.
Desktop sharing: Presenters can share
anything on their PC system, including applications.
Web browser sharing: Allows easy sharing
of Web-based information.

1383

How Technology Can Support Culture and Learning

Polling: Lets presenters solicit quantitative

feedback from attendees online.
Internationalization: ability to change the
menus for selected languages.
Text chat: generally available as group
chat and a separate channel for interaction
during a session. A controversial feature in
academics, as it is perceived by faculty as
the electronic equivalent of note-passing or
whispering in class.
Record and playback: Permits recording
of all interactions in a meeting for later
reference, training, or demos. Anyone can
play back a recording.
Application share control: The presenter
can share control of any software application with others in a meeting, for unmatched
interactive meetings on the Web.
Scalable: Supports multiple concurrent
meetings with thousands of participants.
File transfer: Users can conveniently upload
and download files as desired.
Event tech support: ASP vendors customer
support personnel can take control of a users
PC system (with the users approval) to instantly provide live assistance and resolve
problems.

Learning portals are an emerging variation

of Personal Collaborative Environments, and
are an attempt to combine student services and
community building through an integrated Webenabled system, very much like the search engine
portals Yahoo and Lycos. The use of the term
Learning portal could imply either the basic
software product needed to develop a portal or
the actual use of such software to create a learning experience.
Collaborative software, an emerging category of software, allows individuals to interact
one-to-one, peer-to-peer, or in small groups. The
category may also be referred to as Instant Messaging, or Buddy Systems, although a number

1384

of products in this category also represent the

older listserve e-mail-broadcast technology.
This type of tool will become integral in advanced
Web-based learning programs. Typically, tools in
this category allow awareness of user-selected
individuals and the ability to instant message
(i.e., chat one-to-one) with individuals. Most
programs provide for ad-hoc chatting in small
groups. Some of these programs permit application sharing, voice-over-IP, or other useful overthe-Web features designed to allow individuals to
interact. Currently, this category of software lacks
interoperability. The Internet Engineering Task
Forces Instant Message and Presence Protocol
represents the current form of standards proposed
in order to solve these interoperability issues.
Team groupware is a new subset of collaborative software aimed at creating virtual project
groups. Team groupware is distinguished from
personal collaborative environments by these
characteristics:

Ability to develop documents synchronously

and review asynchronously.
One-to-one, one-to-many, and many-tomany capability.
Centralized user management.
Real-time Internet communications by at
least one of the following:

Enterprise-ready systems scalable to
hundreds of users

Advanced software in this category includes

group scheduling, ability to store and review
older version of documents, and ability to route
items. Systems should also require relatively
little technical expertise for management after
installation, as projects must be easy to create,
edit, or terminate.
Web-based education development tools
attempt to collect basic and single-purpose tools
that meet specific needs but are not themselves
intended to be a turnkey implementation sys-

How Technology Can Support Culture and Learning

tem. They are generally visual authoring tools

or programmers toolkits, or meet a very specific
Web-based need. This category has undergone
substantial consolidation since the high-flying
days prior to the dot.com crash in 2000/2001. A late
2002 study from Market Data Retrieval revealed
that one-third of K-12 schools are already offering
a number of distance learning programs for their
students. High schools, in particular, have been
attracted to the virtual classroom environment,
with 53% offering such programs to students in
rural states where students do not have access to
specialized teachers. Furthermore, 36% of schools
reported to have distance learning programs for
their teachers, allowing them to access professional development programs, which is similar
to the applications corporations are finding for
e-learning.

References
Abraham, M. (1943). A theory of human motivation. Psychological Review, 370-396. Reissued in
2004 as an Appendix to The Third Force: The
Psychology of Abraham Maslow in Adobe PDF.
Brandon Hall Research. Retrieved from http://
www.brandon-hall.com/
CETIS, the Centre for Educational Technology
Interoperability Standards. Retrieved from http://
www.cetis.ac.uk/
Chapman, B. (2006). LMS knowledge base 2006.
Sunnyvale, CA: Brandon-Hall.
Chapman, B., & Nantel, R. (2006). Low-cost
learning management systems 2006. Sunnyvale,
CA: Brandon-Hall.
Council Resolution of 25 March 2002 on the
eEurope Action Plan 2002: Accessibility of public
websites and their content. (2002/C 86/02). Official
Journal of the European Communities. Retrieved

from http://eur-lex.europa.eu/LexUriServ/site/en/
oj/2002/c_086/c_08620020410en00020003.pdf
EEIG. European Economic Interest Grouping (EEIG). Activities of the European Union,
Summaries of Legislation, EUROPA.EU.INT.
Retrieved from http://europa.eu.int/scadplus/leg/
en/lvb/l26015.htm
FLIC Project. (2004). D2.1 Design of a pedagogical template.
FLIC Project. (2005a). D2.4 Voice models recording.
FLIC Project. (2005b). D2.3 Development of
materials for the evaluation.
FLIC project. (2006). Special project status report
for the EC.
Hornof, A. J., & Halverson, T. Cognitive modeling, eye tracking and human-computer interaction. Department of Computer and Information
Science, University of Oregon. Retrieved March
23, 2007, from http://www.cs.uoregon.edu/research/cm-hci/
IDC. (2002). Corporate eLearning: European
market forecast and analysis 2001/2002. IDC
Corporate, 5 Speen Street, Framingham, MA,
01701.
Kowler, E. Eye movements and visual attention.
In MITECS: The MIT encyclopedia of the cognitive sciences. Retrieved from http://cognet.mit.
edu/MITECS/Entry/kowler
Lasnier, C. (2005). How to best exploit the VOCODER. AGERCEL, 2005.
LeLoup, J. W., & Ponterio R. (2001). On the
NetFinding song lyrics online. Language
Learning & Technology, 5(3), 4-6. Retrieved from
http://llt.msu.edu/vol5num3/onthenet
Mah, A. Y. (2003). A thousand pieces of goldA
memoir of Chinas past through its proverbs (pp.

1385

How Technology Can Support Culture and Learning

293). Harper Collins Publishers. MASIE Centers

e-Learning consortium. (2002).
Making sense of learning specifications & standards: A decision makers guide to their adoption.
Retrieved from http://www.masie.com/standards/
S3_Guide.pdf
MediTECH. (2005a). FLIC prototypeInstruction manual.
MediTECH. (2005b). FLIC project public presentation.
Metcalf, D. (2004a). m-Learning: Technology
update and best practices. Presentation held
at Training 2004, Atlanta GA. Retrieved from
http://www.elearnopedia.com/mlearning/mlearning-wed.pdf
Metcalf, D., (2004b). m-Learning evaluation.
Presentation held at Training 2004, Atlanta GA.
Retrieved from http://www.elearnopedia.com/
mlearning/m-learning%20evaluation.pdf
Nantel, R., & Vipond, S. (2006). Authoring tool
knowledge base 2006. Sunnyvale, CA: BrandonHall.
Neuman, W. R. (1991). The future of mass audience. Cambridge University Press.
Novak, J. D. (1998). Learning, creating, and using
knowledge: Concept maps as facilitative tools in
schools and corporations. Mahwah, NJ: Lawrence
Erlbaum Associates.
Sussman, R., & Sedivy, J. Using Eyetracking to
detect and describe Filled Gap Effects, the electronic community for scholarly research in the
cognitive and brain sciences. MIT COgNet. Retrieved from http://cognet.mit.edu/posters/poster.
tcl?publication_id=45136
Tomita, T. (1980). The new electronic media and
their place in the information market of the future.
In A. Smith (Ed.), Newspapers and democracy:

1386

International essays on a changing medium.

Cambridge, MA: MIT Press.

ENDNOTES

The term clerices survives in some students organisations, named Goliardia,
that can still be found in ancient European
universities.
2

St. Edmund Hall (1278), Exeter (1314), Oriel
(1326), Queens (1340), New (1379), Lincoln
(1427), All Souls (1438), Magdalen (1458),
Brasenose (1509), Corpus Christi (1517),
Christ Church (1546), Trinity (1554), St
Johns (1555), and Jesus (1571).
3

Africans (RSA and South African Radio
Broadcasting Corporation), Spanish (Radio
Andorra), Chinese (Radio Beijing), Jew
(Israel Broadcasting Authority), French
(Radio France International), Japanese
(Radio Japan), Greek (Cyprus Broadcasting Corporation), English (BBC, Voice of
America in Special English, Radio Australia,
KGEI California, WYFR), Dutch (Radio
Nederland), Polish (Radio Poland), Russian
(Radio Moscow), Swedish (Radio Sweden),
German (Deutsche Welle, Deutshlandfunk,
Radio DDR).
4

Originally named APPANET (from the
combination of the names ARPA - Advanced
Research Projects Agency and Net) was born
during the late 1960s as a Defense Advanced
Research Projects (DARPA) project to
ensure efficient and robust communication
among military bases and the capitol, in
case of a nuclear attack.
5
Some of the most well-known systems are
the one used in avionics (e.g., on the Cobra
attach helicopter), or in tanks and many
other combat systems.
6

This is just a rough description of evolution
partition and sequencing in the field
1

How Technology Can Support Culture and Learning

Actually, the first emperor ordered that a 1:1
replica of the Dalai Lama palace was built
so that his guest could feel at home.
8

Matteo Ricci is the author of the first world
map printed in China, which notably is
China-centred, well in accordance with the
Chinese definition of their land: All that is
under Heaven.
9

Initially based on a combination of books and
records, then tapes, and now CD/DVD
10

NotePad or other text editor
11

Director, Authorware, ToolBook, and so
forth.
7

ReadyGo, Lectora Publisher, Trainersoft,
and so forth.
13

http://www.section508.gov/index.cfm
14

In 1999, 95% of schools in the U.S. have
Internet access, and 63% instructional
rooms with Internet access (Source: Catrina
Williams .(2000). Internet Access in U.S.
Public Schools and Classrooms, 1994-1999,
NCES 2000-086, Washington, D.C. U.S.
Department of Education, National Center
for Education Statistics. For more details
see http://nces.ed.gov/pubsearch/pubsinfo.
asp?pubid=2000086
12

This work was previously published in Interactive Multimedia Music Technologies, edited by K. Ng and P. Nesi, pp. 195-230,
copyright 2008 by Information Science Reference, formerly known as Idea Group Reference (an imprint of IGI Global).

1387

1388

Chapter 3.25

Tangible User Interfaces as

Mediating Tools within Adaptive
Educational Environments
Daria Loi
RMIT University, Australia

Abstract
This chapter proposes that, as approaches to human computer interaction (HCI), tangible user
interfaces (TUIs) can scaffold rich classroom
experiences if they are coupled and generated
within multi-pedagogical frameworks that adopt
concepts such as Multimodality, Multisensoriality,
and Multiliteracies. It overviews some necessary
conditions for these tools to be effective, arguing
that tangible user interfaces and multi-pedagogies
are efficient when they are conceptualized as part
of adaptive educational environmentsteaching
and learning ecologies where learners and teachers are seen as co-creators of content and of new
ways of interacting with such content.

Introduction
Information and communications technologies
(ICTs) enable types of learning experiences

involving HCI that other means do not easily

achieve (Simon, 1987; Underwood & Underwood, 1990). While digital spaces are traditionally manipulated via simple input devices (e.g.,
keyboard and mouse) that are used to manipulate
representations displayed on output devices (e.g.,
monitors), tangible user interfaces remove this
input-output distinction and connect physical and
digital worlds using physical objects as interfaces
to digital information (Ullmer & Ishii, 1997).
This chapter discusses the potential role of
tangible user interfaces in scaffolding rich HCI
classroom experiences and some necessary
conditions for such tools to be effective. I argue
that these interfaces can have a key role in contemporary teaching and learning environments
if they are coupled and generated within multipedagogical frameworks that adopt concepts
such as Multimodality (Kress & Van Leeuwen,
2001), Multisensoriality (Ceppi & Zini, 1998), and
Multiliteracies (Cope, Kalantzis, & New London
Group, 2000).

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Tangible User Interfaces as Mediating Tools within Adaptive Educational Environments

Tangible user interfaces and multi-pedagogies

are, however, effective when they are conceptualized as part of adaptive educational environments
(Loi & Dillon, 2006)teaching and learning
ecologies where learners and teachers are seen
as co-creators of content and of new ways of
interacting with such content.
This chapter is divided into four sections. In
the first part I overview why and how tangible
user interfaces can enrich classroom experiences,
while in the second I outline the importance of
coupling them with multi-pedagogical frameworks. The third section overviews the notions of
adaptive educational environments and proposes
that tangible user interfaces can be conceptualized as mediating tools enabling a shift of such
environments to become creative spaces. In the last
section I offer a number of concluding remarks,
highlighting future implications and the need
for new ways of conceptualizing contemporary
learning environments.
One of the key objectives of this chapter is to
highlight the importance of designing tangible
user interfaces for teaching and learning by
considering them part of larger ecological HCI
frameworks where pedagogy, people, and context
play a crucial role.

richness of physical-space interactions they want

to enhance (Ullmer & Ishii, 1997).
This understanding was the basis on which
the notion of tangible user interfacesuser interfaces that adopt surfaces, instruments, physical
objects, and spaces as physical interfaces to digital
informationwas constructed through initial
explorations by Fitzmaurice, Ishii, and Buxton
(1995) and the original work of Ullmer and Ishii
(1997). Tangible interfaces put emphasis on touch
and physicality in both input and output and are
often coupled to physical representations of actual objects (OMalley & Stanton Fraser, 2004).
Examples of tangible user interfaces include
rehabilitation tools (Edmans, Gladman, Walker,
Sunderland, Porter, & Stanton Fraser, 2004),
drawing and designing tools (Ryokai, Marti, &
Ishii, 2004), collaborative and management tools,
browsers and exploratory tools (Piper, Ratti, &
Ishii, 2002), multimodal interactive tools (Raffle,
Joachim, & Tichenor, 2002), music creation
and performance (Patten, Recht, & Ishii, 2002;
Weinberg & Gan, 2001), and toys/educational
tools (Mazalek, Wood, & Ishii, 2001; Vaucelle
& Jehan, 2002).
OMalley and Stanton Fraser (2004) stress the
beneficial role of physical manipulatives in learning environments by highlighting that:

Tangible User Interfaces and

Learning Environments

Ullmer and Ishii (1997) point out that while

graphical user interfaces (GUIs) have proven to be
a successful and durable model for human computer interaction, the GUI approach to HCI falls
short in many respects, particularly in embracing the rich interface modalities between people
and the physical environments they inhabit (p.
1). Within this context, a range of alternatives
has been explored, from ubiquitous computing
to augmented reality. However, these attempts
often rely on exporting the GUIs paradigm to
world-situated devices, failing in capturing the

physical action and concrete objects are

important in learning;
physical materials trigger mental images
that inform future problem solving in the
absence of physical materials;
learners can abstract symbolic relations from
a variety of concrete instances; and
familiar physical objects are more easily
understood by children if compared with
more symbolic entities.

This suggests that children are capable of demonstrating their knowledge via physical actions
(e.g., gestures) and can solve problems by working
with given concrete materials even if they cannot

1389

Tangible User Interfaces as Mediating Tools within Adaptive Educational Environments

solve them symbolically or in their heads. This

is confirmed through the work of Bruner (1966),
Piaget (1953), Church and Goldin-Meadow (1986),
Goldin-Meadow (2003), and more recently Martin
and Schwartz (2005).
Many thinkers have developed tools/environments that recognize the significance of childrens
interactions with the physical environment:
Pestalozzi (1894) and his method; Dewey (1938)
and his philosophy of experience and relation
to education; Montessori (1917, 1949) and her
method; Kamii & DeVries (1978, 1980) and their
take on Piagets theory; and Gardner and Hatch
(1989) and their multiple intelligences.
The role of physical materials in triggering
mental images useful for problem-solving has
been discussed by several authors and teachers,
including: Montessori (1917); Dienes (1964);
Bruner (1966); Stigler (1984); Uttal, Scudder,
and DeLoache (1997); and Chao, Stigler, and
Woodward (2000).
The capacity to abstract symbolic relations
from concrete instances is discussed by Bruner
(1966) when he talks about enactive, iconic, and
symbolic forms of representation and by Karmiloff-Smith (1992) and her theory of representational
re-description where childrens representations
move from implicit to more explicit forms.
In synchrony with the previous body of literature, tangible user interfaces and manipulatives
have a number of characteristics (OMalley &
Stanton Fraser, 2004) including:

1390

they allow for parallel input (e.g., two hands)

improving the expressiveness or the communication capacity with the computer;
they take advantage of well developed motor
skills for physical object manipulations and
spatial reasoning;
they externalize traditionally internal computer representations;
they afford multi-person, collaborative
use;

physical representations embody a greater

variety of mechanisms for interactive control;
physical representations are perceptually
coupled to actively mediated digital representations; and
the physical state of the tangible embodies
key aspects of the digital state of the system.

There are a number of examples of tangible

user interfaces that enrich HCI in classroom experience through these characteristics. OMalley
and Stanton Fraser (2004) have reviewed several
of such examples, dividing them into three main
categories.
The first category is that of digitally augmented
paper and books, such as Listen Reader (Back,
Cohen, Gold, Harrison, & Minneman, 2001),
MagicBook (Billinghurst & Kato, 2002), the
KidStory project (Stanton, Bayon, Neale, Benford,
Cobb, Ingram, et al., 2001), which adopts the
KidPad storytelling software (Druin, Stewart,
Proft, Bederson, & Hollan, 1997), and LeapPad.
Another category of tangible user interfaces is that
represented by phicons (physical objectssuch
as toys, blocks, and physical tagsused to trigger digital effects), such as the CACHET project
(Luckin, Connolly, Plowman, & Airey, 2003),
Storymat (Ryokai & Cassell, 1999), the Tangible
Viewpoints system (Mazalek, Davenport, & Ishii,
2002), and Chromarium (Rogers, Scaife, Gabrielli,
Smith, & Harris, 2002). There are also digital
devices with embedded computational properties,
such as Triangles (Gorbet, Orth, & Ishii, 1998),
Curlybot (Frei, Su, Mikhak, & Ishii, 2000), Topobo
(Raffle, Parkes, & Ishii, 2004), Thinking Tags
(Borovoy, McDonald, Martin, & Resnick, 1996),
and sensors and digital probes (tangible devices
based on physical tools that act as sensors or probes
of the environment), such as Storytent (Green,
Schndelbach, Koleva, Benford, Pridmore, &
Medina, 2002), I/O Brush (Ryokai et al., 2004),

Tangible User Interfaces as Mediating Tools within Adaptive Educational Environments

and the SENSE project (Tallyn, Stanton, Benford,

Rowland, Kirk, Paxton, et al., 2004).
Tangible user interfaces have a beneficial role
in learning environments due to their physical and
tangible characteristics and trigger HCI learning
experiences that can have different qualities from
those in which a learner can engage by using
other digital resources such as the Web. Moreover, these tools can enable many different ways
of integrating face-to-face and virtual learning,
creating hybrid interactions via their tangibility
and manipulability nature. However, tangible
user interfaces cannot operate and should not be
developed in isolation from other key factors, as
discussed in the next section.

Tangible User Interfaces and

Multi-Pedagogies
As mentioned, tangible user interfaces enable
numerous ways of integrating face-to-face and
virtual learning. However, I propose that to be
effective in the classroom and to enhance rich
HCI learning experiences, these media devices
need to be coupled with multi-pedagogical frameworks. Failing to do so implies the development
of technology for technology sake instead of rich
ICT tools that can enhance, support, and scaffold
rich HCI learning experiences. As Reimann and
Goodyear (2004) highlight, technology is not
so important in itself: what matters is how the
technology is used (p. 2)pedagogy and pedagogical choices in making good use of technology
have crucial roles.
I define multi-pedagogical frameworks as
scaffolds that refer to and integrate notions of
Multiliteracies (Cope et al., 2000), Multimodality
(Jewitt & Kress, 2003; Kress & Van Leeuwen,
2001), and Multisensoriality (Ceppi & Zini,
1998).
The term multiliteracies refers to the variability
of meaning making in different cultural, social,

or professional contexts and to the nature of new

communications technologies (Cope & Kalantzis,
2005). Cope et al. (2000), drawing from a number
of theories, generated the Multiliteracies model by
merging four approaches to teaching and learning:
situated practice, overt instruction, critical framing, and transformed practice. Deepening the notion of multiliteracies, Kalantzis and Cope (2003)
look at pedagogy in terms of eight knowledge
processes, which are selected and deployed by the
teacher and define the mindful and premeditated
use of such processes as learning-by- design
(Cope, Kalantzis, & the Learning by Design
project group, 2005). In the Learning-by-Design
project, the teacher becomes a reflective designer
of learning experiences (teacher-as-designer), and
classroom plans become shareable designs-forlearning (Burrows & Loi, 2006, p. 2). In this
model meaning is generated in the space where
visual, audio, gestural, and spatial patterns of
meaning interface written-linguistic modes of
meaning (Cope & Kalantzis, 2005). To Kress
(1999) the visual is becoming more prominent
in many domains of public communication and
is taking over many of the functions of written
language while displacing the written language
from its hitherto unchallenged central position
in the semiotic landscape (p. 68).
Building on the preceding considerations, the
notion of multimodality implies a paradigm shift
in how the communication and representation of
meanings can be interpreted. A multimodal text
adopts several modes of communication (such
as image, writing, and speech), and, while multimodality characterizes many modern texts, in
the age of digitisation, the different modes have
become the same at some level of representation,
and they can be operated by one multi-skilled
person (Kress & Van Leeuwen, 2001, p. 2). The
notion of multimodality has been to date discussed
by a number of authors from various disciplinary
domains. Besides Kress and Van Leeuwen (2001),
the works of Norris (2004), Levine and Scol-

1391

Tangible User Interfaces as Mediating Tools within Adaptive Educational Environments

lon (2004), Kramsch, ANess, and Lam (2000),

and Jewitt (Jewitt & Kress, 2003) represent key
examples.
Ceppi and Zini (1998), when discussing
the characteristics that learning environments
should have, talk about multisensoriality and
multisensorial spaces: complex environments
made up of sensory contrasts and overlappings
that are phenomenologically distinct (p. 17). To
the author, the space should not simply be about
being simply rich in stimuli, but it should have
different sensory values so that each individual
can tune into his or her own personal reception
characteristics, implying that standard univocal solutions cannot be conceived (Ceppi &
Zini, 1998, p. 17).
In multi-pedagogical frameworks teachers
deploy a number of knowledge processes in a
premeditated and mindful way (multiliteracies),
becoming reflective designers of learning experiences where meaning is represented and unfolded
in multimodal ways (multimodality) and enriched
by different sensory values (multisensoriality). As
already highlighted, pedagogy and pedagogical
choices have crucial roles in making good use of
technology in the classroom, and technology for
the classroom should be conceived and developed
in light of and interlinked with such pedagogical
choices. Within this context, this chapter argues
that the design and development of tangible user
interfaces for the classroom should be coupled
with multi-pedagogical frameworks.

together with learners, teachers, and context, and,

of course, the relationships among all actors (refer
to Figure 1). These five actors are interrelated
similarly to how context, people, practices, and
relationship interact within collaborative workspaces (Loi, 2003). Reggio Emilia schools offer
an exemplary instance of how HCI can be part
of the learning ecologies (Project Zero & Reggio
Children, 2001).
This model of learning ecology describes a
situation where actors are deeply interrelated and
co-dependentthere is a sense of wholeness and
rich complexity. This implies that tangible user
interfaces cannot be developed in isolation from
other key actors, but rather conceived in light
of their users (teachers and learners), contexts
of HCI use, and designed learning experiences
(multi-pedagogical choices).
Previous work (Loi & Dillon, 2006) discussed
how ideas about collaborative workspaces (Loi,
2005) and integrativism (Dillon, 2006) can be
incorporated and extended to look at the ecological nature of learning environments and to unfold
notions of adaptive educational environments
spaces that accommodate changing relationships
between people and resourcesand creative
spacesenvironments that have been modified
through designed interventions. Creative spaces

Figure 1. A complex web of relationships

Tangible User Interfaces and

Learning Ecologies
Tangible user interfaces have a key role in scaffolding rich HCI classroom experiences, and they
should be coupled with multi-pedagogical frameworks. To amplify these notions, I suggest that
tangible user interfaces and multi-pedagogical
frameworks are in fact part of learning ecologies,

1392

multi-pedagogical
framework

learners

TUIs
teachers
context

learning
ecology

Tangible User Interfaces as Mediating Tools within Adaptive Educational Environments

Figure 2. Tangible user interfaces as HCI mediating tools

learners
TUIs

multi-pedagogical
frameworks

teachers

I suggest that tangible user interfaces could

be employed as HCI mediating tools within learning ecologies that are conceptualized as adaptive
educational environments. This implies that the
dynamic flow underpinning the previous model
(in Figure 1) requires a refinement (refer to Figure
2). This model maintains the ecological nature
of Figure 1, while underlining the possibility to
consider tangible user interfaces as mediating
HCI agents (or brokers). This role and its positioning within educational environments impact
how tangible user interfaces can be conceived,
developed, and adopted. In line with Loi and Dillon
(2006), the notion of tangible user interfaces as
HCI mediating tools can be extended to re-conceptualize them as designed interventions that can
trigger the development of creative spaces. These
concepts will be explored in future work.
To address the development of tangible user
interfaces in the described HCI context, I suggest
that a participatory design approach (Sanoff, 1990;
Schuler & Namioka, 1993) should be adopted
to ensure that all parts of the whole are active
participants in the design process. This proposition implies that teachers and learners must be
involved in the design process; designers and
technologists must operate in multidisciplinary

learning
ecology

context

teams together with teachers and pedagogy experts; and the contexts where technology will be
adopted/embedded must be considered at all times.
There are a number of examples of teams currently
developing tangible user interfaces within these
participatory paradigms. An interesting example
within the educational context is offered by the
Children as Design Partners project.
As previously mentioned, Reimann and
Goodyear (2004) propose that in educational
environments, technology is not so important
in itself: what matters is how the technology is
used (p. 2). Their statement, which implies that
pedagogical choices in making good use of ICT
have crucial roles, should be amplified because
when considering HCI learning ecologies, the
design of ICT has an equally crucial role, as it
strongly impacts what pedagogical choices can
be conceived and enacted. The interrelations and
co-dependencies of technology and pedagogy
should not be underestimated. In light of this,
I suggest that tangible user interfaces can offer
significant opportunities for effective use of HCI
within the classroom if:

they are conceptualized as actors within

learning ecologies, together with multi-

1393

Tangible User Interfaces as Mediating Tools within Adaptive Educational Environments

pedagogical frameworks, learners, teachers,

and contexts;
there is a strong appreciation of the deep
interrelationships and co-dependencies
existing among learning ecologies actors;
they are conceived as mediating tools within
learning ecologies;
the learning experiences that are mediated
by tangible user interfaces are designed
adopting multi-pedagogical frameworks;
and
they are developed adopting a participatory
design approach.

The adoption of a participatory design approach can be extended to the development of

rich HCI learning experiences. When looked
at through such lenses, teaching and learning
practices can be expanded to notions of teacherscreators, learners-creators, and teacher-student as
space for content emergence (OMalley & Stanton
Fraser, 2004); teachers as author-publishers and
project-consultants (Burrows, 2005a, 2005b);
teacher-as-designer (Brown & Edelson, 2003);
and teachers and learners as co-researchers
(Loughran, Mitchell, & Mitchell, 2002; Phelps,
Graham, et al., 2004).
Tangible user interfaces offer some significant
benefits to classroom activities, in particular from
(1) creativity, (2) diversity, and (3) collaborative
perspectives. These are discussed in the following paragraphs.
Creativity is an essential life skill, which
needs to be fostered by the education system(s)
from the early years onward (Craft, 1999, p. 137).
In her literature review on creativity, technology
and learning, Loveless (2003, p. 13) overviews the
characteristics of learning environments that are
conducive to creative developments, including opportunities for exploration and play with materials,
information and ideas (Craft, 2000), flexibility in
time and space for the different stages of creative
activity (Claxton, 1999), and opportunities to

1394

take risks and make mistakes in a non-threatening atmosphere (Davies, 1999). These are in
synchrony with some characteristics typical of
tangible user interfaces, such as their physicality, manipulability, concreteness, and familiarity (OMalley & Stanton Fraser, 2004).
Tangible HCI enables learners to combine
and recombine the known and familiar in new
and unfamiliar ways (Hoyles & Noss, 1999) and to
unfold the world through discovery and participation
(Soloway, Guzdial, & Hay, 1994; Tapscott, 1998).
As Hoyles and Noss (1999) stress, it is this dialectic
between known and unknown, familiar and novel,
that provides a motor for creativity (p. 19). Moreover, combining familiarity with unfamiliarity can
promote reflexiveness (Rogers et al., 2002; Scaife,
2002), stimulating awareness and enhancing learning (Ackerman, 1996; Piaget & Inhelder, 1967).
Green, Facer, Rudd, Dillon, and Humphreys
(2005) have suggested that as mobile, tangible,
and networked technologies are increasingly deployed in facilitating creative work, educational
environments will be characterized by a high
degree of personalization where individuals will
have greater influence over how they choose and
use learning resources. The education system is
constantly challenged by diversity issues, and
researchers/educators are regularly looking for
ways to enable learning for diverse learners (refer,
for instance, to Cope et al., 2000). Through his
work on the Internet generation, Prensky (2001)
argues there is a need to radically rethink teaching practice to mirror current ways of learning.
The diversity issue implies a need to change the
education system so that it can conform to learners. Interestingly, learners are already shaping
their own learning outside the classroom, using
digital resources to create personalized learning
environments for themselves outside of school. As
a matter of fact, it has been suggested that by the
age of 21 the average person will have spent 15,000
hours in formal education, 20,000 hours in front
of the TV, and 50,000 hours in front of a computer

Tangible User Interfaces as Mediating Tools within Adaptive Educational Environments

screen (International Futures Forum, Futuribles,

European Journal of Education, & Universit de
Paris Dauphine, 2005). Various studies provide
practical example of this tendency (for instance,
Ba, Tally, & Tsikalas, 2002; Marsh, 2004).
The preceding data reiterates the need to find
ways to shift the educational system so that it
can conform to learners, better addressing their
diversities. As Green et al. (2005) observe, the
logic of education systems should be reversed so
that it is the system that conforms to the learner,
rather than the learner to the system. Thanks to
tangible user interfaces, their intrinsic characteristics (physicality and active-oriented nature;
sensory engagement; accessibility; concreteness
and stimuli to abstract symbolic relations; capacity to trigger mental images that inform future
problem-solving; familiarity) could offer interesting possibilities to this needed shift, enabling
interactive HCI classroom activities where diverse
learners can be in charge of how they choose and
employ available learning resources.
A third significant benefit to classroom activities offered by tangible user interfaces is represented by their capacity to encourage collaboration
as they afford multi-person, collaborative use
(OMalley & Stanton Fraser, 2004). A number of
researchers have discussed and tested the benefits
of: children working together (Rogoff, 1990;
Stanton et al., 2001; Wood & OMalley, 1996);
collaborative practices on learning (for instance,
Brandt, 1990; Hurley, Boykin, & Allen, 2005;
Johnson, 1994; Johnson, Johnson, & Scott, 1978;
Johnson, Maruyama, Johnson, Nelson, & Skon,
1981; Slavin, 1987); information technologies
in peer and cooperative learning (for instance,
Anderson, 2005; Chen & Morin, 2006; Topping,
2005); and that of tangible media in collaborative
classroom practice (Africano, Berg, Lindbergh,
Lundholm, Nilbrink, & Persson, 2004).

Conclusion and Future Work

Although ICTs are already integrated into teaching
across the curriculum, more encouragement might
be given to teachers in terms of incorporating
everyday technologies into the classroom, and the
development of physical and multisensory activities, increasingly part of the curriculum, should
occur with regard to the use of ICT (OMalley &
Stanton Fraser, 2004). In particular, tangible user
interfaces could be considered and designed as
mediating agents between pedagogy, teachers,
learners, and context. This can occur only if
more emphasis is placed on learners practices
and teachers pedagogical choices and if teachers
and learners are co-designers in the development
of tangible HCI for learning.
There is a need to expand the ways in which tangible HCIs are researched, developed, deployed,
evaluated, and implemented, and it is necessary
to bridge the gap between technologist, designers, researchers, and teachers as their technical,
pedagogical, and methodological knowledge and
skills differ while complementing each other. A
way to translate, mediate, and share ways of doing, thinking, and creating effective HCI must
be developed.
As OMalley and Stanton Fraser (2004)
stress:
More research is needed on the benefits of tangibles for learningso far the evaluation of these
technologies has been rather scarce. More effort
is also needed to translate some of the research
prototypes into technologies and toolkits that
teachers can use without too much technical
knowledge. Teachers also need training in the
use of non-traditional forms of ICT and how to
incorporate them into teaching across the curriculum. And finally, new forms of assessment
are needed to reflect the potential of these new
technologies for learning.

1395

Tangible User Interfaces as Mediating Tools within Adaptive Educational Environments

The notion of tangible user interfaces as mediating agents within educational environments
implies a shift in how the learning process is
viewed, a shift in the roles of teachers and learners, and a shift in how learning spaces are shaped
and interacted with.
OMalley and Stanton Fraser (2004) express
their hope that teachers will take inspiration from
the whole idea of technology-enhanced learning
moving beyond the desktop or classroom computer
by, for example, making links between ICT-based
activities and other more physical activities. I
share this hope, and this chapter aims at nurturing this notion by linking technological aspects
of HCI with pedagogical, ecological, and methodological considerations under the same holistic
framework. The proposed HCI scaffold fosters
a different way of looking at technology-design
and technology-deployment, while promoting the
opportunity to re-conceptualize roles, practices,
and the relationships among key actors.
As discussed in this chapter, tangible user
interfaces offer some significant benefits to
classroom activities that involve HCI, in particular from creativity, diversity, and collaborative
perspectives. However, these benefits imply a
need for educational institutions to re-consider
and re-address how they view the professional
development of teachers, the classroom space,
and how curriculum development and learning
resources should be designed and deployed.
In the next phase of this work and building on
current work around creative educational spaces,
I propose to investigate the opportunities to reconceptualize tangible user interfaces as designed
interventions. Future research in this field might
also include longer term qualitative studies to
(1) test tangible user interfaces in specific HCI
contextsliving laboratorieswith particular
emphasis on their capacity to scaffold collaboration, creativity, and diversity; (2) further explore
their agency in learning ecologies and (3) their

1396

impact on the HCI classroom space; (4) investigate

specific benefits and challenges for instructors
that decide to use the proposed model, coupling
multipedagogical framework with tangible user
interfaces; (5) look at how teachers can be scaffolded from a professional development perspective in their deployment of HCI tangibles; and (6)
what type of learning resources can be designed
for tangible user interfaces to deeply enrich the
HCI classroom experience.

Acknowledgments
My special thanks go to Prof. Patrick Dillon
for sharing with me the adaptive educational
environments journey. Thank you also to Mary
Featherston for her contagious spaces-for-learning-enthusiasm. Finally, many thanks to Dr. Peter
Burrows for our discussions and shared visions,
which truly enrich my work.

References
Ackerman, E. (1996). Perspective-taking and
object construction: Two keys to learning. In Y.
Kafai & M. Resnick (Eds.), Constructionism in
practice: Designing, thinking and learning in a
digital world. NJ: Lawrence Erlbaum.
Africano, D., Berg, S., Lindbergh, K., Lundholm,
P., Nilbrink, F., & Persson, A. (2004, April).
Designing tangible interfaces for childrens
collaboration. Paper presented at the CHI 04,
Vienna, Austria.
Anderson, L. S. (2005). A digital doorway to the
world. T.H.E. Journal, 33(4), 14-15.
Ba, H., Tally, W., & Tsikalas, K. (2002). Investigating childrens emerging digital literacies. The
Journal of Technology, Learning and Assessment,
1(4), 3-50.

Tangible User Interfaces as Mediating Tools within Adaptive Educational Environments

Back, M., Cohen, J., Gold, R., Harrison, S., &

Minneman, S. (2001). Listen reader: An electronically augmented paper-based book. Paper
presented at the ACM SIGCHI Conference on
Human Factors in Computing Systems (CHI01),
Seattle, WA.
Billinghurst, M., & Kato, H. (2002). Collaborative augmented reality. Communications of the
ACM, 45(7), 64-70.
Borovoy, R., McDonald, M., Martin, F., & Resnick,
M. (1996). Things that blink: Computationally
augmented name tags. IBM Systems Journal,
35(3), 488-495.
Brandt, R. (1990). On cooperative learning: A
conversation with Spencer Kagan. Educational
Leadership, 47(4), 8-11.
Brown, M., & Edelson, D. C. (2003). Teaching
as design. Detroit, MI: The Center for Learning
Technologies in Urban Schools.
Bruner, J. (1966). Toward a theory of instruction.
New York: WW Norton.
Burrows, P. (2005a). The emergence of pedagogical mentors: The project-consultant. In M. Kalantzis, B. Cope, & the Learning by Design project
group (Eds.), Learning by design (pp. 177-196).
Melbourne: Common Ground Publishing.
Burrows, P. (2005b). The role of the teacher as
theory-enactor and author-publisher. In M. Kalantzis, B. Cope, & the Learning by Design project
group (Eds.), Learning by design (pp. 197-216).
Melbourne: Common Ground Publishing.
Burrows, P., & Loi, D. (2006). Learning-by-design: Combining a pedagogical framework and
epublishing medium to create, plan and share
teaching practice. The International Journal of
Learning, 13.
Ceppi, G., & Zini, M. (1998). Children, spaces
& relationsMetaproject for an environment
for young children. Reggio Emilia, Italy: Reg-

gio Children S.r.l. & Domus Academy Research

Center.
Chao, S., Stigler, J., & Woodward, J. (2000). The
effects of physical materials on kindergartners
learning of number concepts. Cognition and
Instruction, 18(3), 285-316.
Chen, L., & Morin, M. (2006). Trading files or
learning? Exploring the educational potential of
instant messaging. Orbit, 36(1), 18-21.
Church, R., & Goldin-Meadow, S. (1986). The
mismatch between gesture and speech as an index
of transitional knowledge. Cognition, 23, 43-71.
Claxton, G. (1999). Wise up: The challenge of lifelong learning. London: Bloomsbury Publishing.
Cope, B., & Kalantzis, M. (2005). Multiliteracies.
Retrieved November 11, 2005, from http://www.
cgpublisher.com/publishers/1/web/index.html
Cope, B., Kalantzis, M., & New London Group.
(2000). Multiliteracies: Literacy learning and
the design of social futures. London; New York:
Routledge.
Cope, B., Kalantzis, M., & the Learning by
Design project group. (2005). Learning by design. Melbourne, Australia: Common Ground
Publishing.
Craft, A. (1999). Creative development in the early
years: Some implications of policy for practice.
The Curriculum Journal, 10(1), 135-150.
Craft, A. (2000). Creativity across the primary
curriculum: Framing and developing practice.
London: Routledge.
Davies, T. (1999). Taking risks as a feature of
creativity in the teaching and learning of design
and technology. The Journal of Design and Technology Education, 4(2), 101-108.
Dewey, J. (1938). Experience and education. New
York: Collier Books.

1397

Tangible User Interfaces as Mediating Tools within Adaptive Educational Environments

Dienes, Z. (1964). Building up mathematics (2nd

ed.). London: Hutchinson International.

Human Factors in Computing Systems (CHI98),

Los Angeles, CA.

Dillon, P. (2006). Creativity, integrativism and a

pedagogy of connection. International Journal
of Thinking Skills and Creativity, 1(2), forthcoming.

Green, H., Facer, K., Rudd, T., Dillon, P., &

Humphreys, P. (2005). Personalisation and digital
technologies. Retrieved January 11, 2006, from
http://www.futurelab.org.uk/research/personalisation/report_01.htm

Druin, A., Stewart, J., Proft, D., Bederson, B., &

Hollan, J. (1997). KidPad: A design collaboration
between children, technologists and educators.
Paper presented at the ACM SIGCHI Conference on Human Factors in Computing Systems
(CHI97), Atlanta, Georgia.
Edmans, J., Gladman, J., Walker, M., Sunderland, A., Porter, A., & Stanton Fraser, D. (2004).
Mixed reality environments in stroke rehabilitation: Development of rehabilitation tools. Paper
presented at the 4th International Conference
on Disability, Virtual Reality and Associated
Technologies, Oxford.
Fitzmaurice, G., Ishii, H., & Buxton, W. (1995).
Bricks: Laying the foundations for graspable
user interfaces. Paper presented at the SIGCHI
Conference on Human Factors in Computing
Systems (CHI95), Denver, Colorado.
Frei, P., Su, V., Mikhak, B., & Ishii, H. (2000).
Curlybot: Designing a new class of computational
toys. Paper presented at the SIGCHI conference on
Human Factors in Computing Systems (CHI00),
The Hague, Netherlands.
Gardner, H., & Hatch, T. (1989). Multiple intelligences go to school: Educational implications of
the theory of multiple intelligences. Educational
Researcher, 18(8), 4-9.
Goldin-Meadow, S. (2003). Hearing gesture:
How our hands help us think. Harvard University
Press.
Gorbet, M., Orth, M., & Ishii, H. (1998). Triangles:
Tangible interface for manipulation and exploration of digital information topography. Paper
presented at the ACM SIGCHI Conference on

1398

Green, J., Schndelbach, H., Koleva, B., Benford,

S., Pridmore, T., & Medina, K. (2002). Camping
in the digital wilderness: Tents and flashlights as
interfaces to virtual worlds. Paper presented at
the ACM SICGHI Conference on Human Factors
in Computing Systems (CHI02), Minneapolis,
MN.
Hoyles, C., & Noss, R. (1999). Playing with (and
without) words. Paper presented at the 7th European Logo Conference - Eurologo 99, Sofia,
Bulgaria.
Hurley, E. A., Boykin, A. W., & Allen, B. A.
(2005). Communal versus individual learning of a
math-estimation task: African American children
and the culture of learning contexts. The Journal
of Psychology; Provincetown 139(6), 513-527.
International Futures Forum, Futuribles, European Journal of Education, & Universit de
Paris Dauphine. (2005). Futures of learning
seminars, future learning practice (Seminar
report). Glasgow.
Ishii, H., & Ullmer, B. (1997). Tangible bits: Towards seamless interfaces between people, bits
and atoms. Paper presented at the ACM SICGHI
Conference on Human Factors in Computing
Systems (CHI97), Atlanta, GA.
Jewitt, C., & Kress, G. (Eds.). (2003). Multimodal
literacy. New York: Peter Lang.
Johnson, D. W. (1994). The new circles of learning: Cooperation in the classroom and school
(Report): Association for Supervision and Curriculum Development.

Tangible User Interfaces as Mediating Tools within Adaptive Educational Environments

Johnson, D. W., Johnson, R. T., & Scott, L. (1978).

The effects of cooperative and individualized
instruction on student attitudes and achievement.
Journal of Social Psychology, 104, 207-216.
Johnson, D. W., Maruyama, G., Johnson, R., Nelson, D., & Skon, L. (1981). Effects of cooperative,
competitive, and individualistic goal structures
on achievement: A meta-analysis. Psychological
Bulletin, 89(1), 47-62.
Kalantzis, M., Cope, B., & et al. (2003). Assessing
multiliteracies and the new basics. Assessment in
Education, 10(1), 15-26.
Kamii, C., & DeVries, R. (1978). Physical knowledge in preschool education: Implications of
Piagets theory. Englewood Cliffs, NJ: PrenticeHall, Inc.

Loi, D. (2003, April). Shared work environments

as ecologies: New ways of working and designing.
Paper presented at the Techn5th European
Academy of Design Conference, Barcelona,
Spain.
Loi, D. (2005). Piazzas where to meet: Organisational settings and their un-manageable spaces
in-between. International Journal of Knowledge,
Culture and Change Management, 4.
Loi, D., & Dillon, P. (2006). Adaptive educational
environments as creative spaces. Cambridge
Journal of EducationSpecial Issue: Creativity
in Education, 36(3).
Loughran, J., Mitchell, I., & Mitchell, J. (2002).
Learning from teacher research. St Leonards,
N.S.W.: Allen & Unwin.

Kamii, C., & DeVries, R. (1980). Group games in

early education: Implications of Piagets theory.
NAEYC Publication.

Loveless, A. (2003). Literature review in creativity, new technologies and learning. Bristol:
NESTA Futurelab.

Karmiloff-Smith, A. (1992). Beyond modularity.

MIT Press.

Luckin, R., Connolly, D., Plowman, L., & Airey,

S. (2003). Childrens interactions with interactive
toy technology. Journal of Computer Assisted
Learning, 19, 165-176.

Kramsch, C., ANess, F., & Lam, W. (2000).

Authenticity and authorship in the computermediated acquisition of L2 literacy. Language
Learning and Technology, 4(2), 78-104.
Kress, G. (1999). English at the crossroads:
Rethinking the curricula of communication in
the context of the turn to the visual. In G. E.
Hawisher & C. L. Selfe (Eds.), Passions, pedagogies, and 21st century technologies. Utah State
University Press.

Marsh, J. (2004). The techno-literacy practices

of young children. Journal of Early Childhood
Research, 2(1).
Martin, T., & Schwartz, D. (2005). Physically
distributed learning: Adapting and reinterpreting physical environments in the development
of fraction concepts. Cognitive Science, 29(4),
587-626.

Kress, G., & Van Leeuwen, T. (2001). Multimodal

discourse: The modes and media of contemporary
communication. London; New York: Arnold;
Oxford University Press.

Mazalek, A., Davenport, G., & Ishii, H. (2002).

Tangible viewpoints: A physical approach to
multimedia stories. Paper presented at the ACM
Multimedia Conference, Juan-les-Pins, France.

Levine, P., & Scollon, R. (Eds.). (2004). Discourse

and technology. Multimodal discourse analysis. Washington, DC: Georgetown University
Press.

Mazalek, A., Wood, A., & Ishii, H. (2001). genieBottles: An interactive narrative in bottles.
Paper presented at the SIGGRAPH 2001Spe-

1399

Tangible User Interfaces as Mediating Tools within Adaptive Educational Environments

cial Interest Group on Computer Graphics, Los

Angeles, CA.
Montessori, M. (1917). The advanced Montessori
method. New York: Frederick A Stokes.
Montessori, M. (1949). The absorbent mind. New
York: Dell.
Norris, S. (2004). Analyzing multimodal interaction. A methodological framework. New York &
London: Routledge.
OMalley, C., & Stanton Fraser, D. (2004).
Literature review in learning with tangible
technologies. Retrieved August 5, 2005, from
http://www.nestafuturelab.org/research/reviews/
reviews_11_and12/12_01.htm
Patten, J., Recht, B., & Ishii, H. (2002). Audiopad:
A tag-based interface for musical performance.
Paper presented at the New Interface for Musical
Expression (NIME 02), Dublin.
Pestalozzi, J. (1894). How Gertrude teaches her
London: Swan Sonnenschein.
children.
Phelps, R., Graham, A., et al.
(2004). Teachers
and ICT: Exploring a metacognitive approach to
professional development. Australasian Journal
of Educational Technology, 20, 49-68.
Piaget, J. (1953). How children form mathematical
concepts. Scientific American, 189(5), 74-79.
Piaget, J., & Inhelder, B. (1967). The coordination
of perspectives. In J. Piaget & B. Inhelder (Eds.),
The childs conception of space. New York: Norton
and Company.
Piper, B., Ratti, C., & Ishii, H. (2002). Illuminating clay: A 3-D tangible interface for landscape
analysis. Paper presented at the ACM SIGCHI
Conference on Human Factors in Computing
Systems (CHI02), Minneapolis, MN.
Prensky, M. (2001). Digital natives, digital immigrants. On the Horizon, 9(5).

1400

Project Zero, & Reggio Children. (2001). Making learning visible: Children as individual and
Reggio Emilia, Italy: Reggio
group learners.
Children.
Raffle, H., Joachim, M., & Tichenor, J. (2002).
Super cilia skin: An interactive membrane. Paper
presented at the ACM SIGCHI Conference on
Human Factors in Computing Systems (CHI02),
Fort Lauderdale, FL.
Raffle, H., Parkes, A., & Ishii, H. (2004). Topobo:
A constructive assembly system with kinetic
memory. Paper presented at the ACM SIGCHI
Conference on Human Factors in Computing
Systems (CHI04), Vienna, Austria.
Reimann, P., & Goodyear, P. (2004). ICT and
pedagogy stimulus paper. In Review of national
goals: Australias common and agreed goals
for schooling in the twenty-first century (pp.
1-42). Sydney: Ministerial Council for Education, Employment, Training and Youth Affairs
(MCEETYA) Task Force.
Rogers, Y., Scaife, M., Gabrielli, S., Smith, H.,
& Harris, E. (2002). A conceptual framework
for mixed reality environments: Designing novel
learning activities for young children. Presence:
Teleoperators & Virtual Environments, 11(6),
677-686.
Rogoff, B. (1990). Apprenticeship in thinking:
Cognitive development in social context. New
York: Oxford University Press.
Ryokai, K., & Cassell, J. (1999). StoryMat: A
play space for collaborative storytelling. Paper
presented at the ACM SIGCHI Conference on
Human Factors in Computing Systems (CHI99),
Pittsburgh, PA.
Ryokai, K., Marti, S., & Ishii, H. (2004). I/O brush:
Drawing with everyday objects as ink. Paper
presented at the ACM SIGCHI Conference on
Human Factors in Computing Systems (CHI04),
Vienna, Austria.

Tangible User Interfaces as Mediating Tools within Adaptive Educational Environments

Sanoff, H. (1990). Participatory design: Theory

& techniques. Raleigh, NC: Henry Sanoff (distributor).
Scaife, M. (2002). External cognition, innovative
technologies and effective learning. In P. Gardenfors & P. Johansson (Eds.), Cognition, education
and communication technology. Lawrence Erlbaum Associates.
Schuler, D., & Namioka, A. (1993). Participatory
design: Principles and practices. Hillsdale, NJ:
Lawrence Erlbaum Associates.
Simon, T. (1987). Claims for LOGO: What should
we believe and why? In J. Rutkowska & C. Crook
(Eds.), Computers, cognition and development
(pp. 115-133). Chichester: John Wiley & Sons.
Slavin, R. E. (1987). Cooperative learning and
the cooperative school. Educational Leadership,
45(3), 7-13.
Soloway, E., Guzdial, M., & Hay, K. (1994).
Learner-centered design: The next challenge for
HCI. ACM Interactions, 1(2), 36-48.
Stanton, D., Bayon, V., Neale, H., Benford, S.,
Cobb, S., Ingram, R., et al. (2001). Classroom
collaboration in the design of tangible interfaces
for storytelling. Paper presented at the ACM SIGCHI Conference on Human Factors in Computing
Systems (CHI01), Seattle, WA.
Stigler, J. (1984). Mental abacusThe effect of
abacus training on Chinese childrens mental calculation. Cognitive Psychology, 16(2), 145-176.
Tallyn, E., Stanton, D., Benford, S., Rowland, D.,
Kirk, D., Paxton, M., et al. (2004). Introducting

e-science to the classroom. Paper presented at the

UK e-science All Hands Meeting, EPSRC.
Tapscott, D. (1998). Growing up digital: The rise
of the net generation. New York: McGraw Hill.
Topping, K. J. (2005). Trends in peer learning.
Educational Psychology; Dorchester-on-Thames,
25(6), 631.
Ullmer, B., & Ishii, H. (1997). The metaDESK:
Models and prototypes for tangible user interfaces. In 10th Annual ACM Symposium on User
Interface Software and Technology (UIST97)
(pp. 223-232). Banff, Alberta, Canada.
Underwood, J., & Underwood, G. (1990). Computers and learning: Helping children acquire
thinking skills. Oxford: Basil Blackwell.
Uttal, D., Scudder, K., & DeLoache, J. (1997).
Manipulatives as symbols: A new perspective on
the use of concrete objects to teach mathematics.
Journal of Applied Developmental Psychology,
18(1), 37-54
Vaucelle, C., & Jehan, T. (2002). Dolltalk: A computational toy to enhance childrens creativity.
Paper presented at the ACM SIGCHI Conference
on Human Factors in Computing System (CHI02),
Minneapolis, MN.
Weinberg, G., & Gan, S. (2001). The squeezables:
Toward an expressive and interdependent multiplayer musical instrument. Computer Music
Journal, 25(2), 37-45.
Wood, D., & OMalley, C. (1996). Collaborative
learning between peers: An overview. Educational Psychology in Practice, 11(4), 4-9.

This work was previously published in Enhancing Learning Through Human Computer Interaction, edited by E. McKay, pp. 178191, copyright 2007 by Information Science Reference, formerly known as Idea Group Reference (an imprint of IGI Global).

1401

1402

Chapter 3.26

Facilitating E-Learning with

Social Software:

Attitudes and Usage from the Students

Point of View
Reinhard Bernsteiner
University for Health Sciences, Medical Informatics and Technology, Austria
Herwig Ostermann
University for Health Sciences, Medical Informatics and Technology, Austria
Roland Staudinger
University for Health Sciences, Medical Informatics and Technology, Austria

Abstract
This article explores how social software tools can
offer support for innovative learning methods and
instructional design in general, and those related
to self-organized learning in an academic context
in particular. In the first section, the theoretical
basis for the integration of wikis, discussion forums, and Weblogs in the context of learning are
discussed. The second part presents the results of
an empirical survey conducted by the authors and
explores the usage of typical social software tools
that support learning from a students perspective.
The article concludes that social software tools

have the potential to be a fitting technology in a

teaching and learning environment.

Introduction
One major task of higher education is to train
students for the requirements of their future work
by applying and adapting their knowledge to specific workplace-related requirements and settings.
Due to the ongoing pressure on enterprises to cut
costs, the periods of vocational adjustment in a
company will become shorter and shorter.

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Facilitating E-Learning with Social Software

On the one hand, the rising pressure of innovation and fast-paced development in the economy
results in increased demand for continuous
employee training. On the other, growing global
competition forces enterprises to use available
resources very economically so that employee
training is considered to be necessary and desired
even though it is conducted under considerable
time and cost pressure (Kllinger, 2002).
According to these goals, the settings of the
education must be changed adequately: While
most of higher education still ascribes to traditional models of instruction and learning, the
workplace is characterized by rapid changes
and emergent demands that require individuals
to learn and adapt in situ and on the job without
the guidance of educational authorities (Sharma
& Fiedler, 2004, p. 543).
In the field of higher education, it has become
an important goal to develop digital literacy
and educate learners as competent users and participants in a knowledge-based society (Kerres,
2007), but it can be assumed that there is a new
generation of students, the digital natives, who
are accustomed to digital and Internet technology
(Prensky, 2001a, 2001b).
Oblinger and Oblinger (2005) characterize
next-generation students (called n-gen, for Net
generation) as digitally literate, highly Internet
savvy, connected via networked media, used to
immediate responses, preferring experiential
learning, highly social, preferring to work in
teams, craving interactivity in image-rich environments, and having a preference for structure
rather than ambiguity.
According to a study conducted by Lenhart
and Madden (2005), half of all teens in the USA
may be considered content creators by using
applications that provide easy-to-use templates
to create personal Web spaces.
Classical face-to-face learning is seen as
rigid and synchronous, and it promotes one-way
(teacher-to-student) communication. Thus, it
is not surprising that more and more students

are opting for Web-based education as a more

flexible and asynchronous mode (Aggarwal &
Legon, 2006).
The higher education system should provide
answers to this new generation of students who
enter the system with different backgrounds and
skills. They are highly influenced by social networking experiences and are able to create and
publish on the Internet (Resnick, 2002).
Educators and teachers therefore have to consider the implications of these developments for
the future design of their courses and lectures.
In 2002, a new term, social software, entered
the stage to refer to a new generation of Internet
applications. One focus of this new generation is
the collaboration of people in sharing information
in new ways such as through social networking
sites, wikis, communication tools, and folksonomies (Richter & Koch, 2007).
Wikis, Weblogs, and discussion forums will
play a central role in the new context, so the areas
of application and possibilities will enlarge enormously. It can be assumed that this will also have
considerable influence on learning and the usage
of these instruments as learning tools.
This article presents the results of an empirical survey in order to highlight the benefits of
the above-mentioned Web-based social software
tools from the students point of view; 268 firstsemester students, all in the first term of their
studies at Austrian universities from different
study programs, took part in this survey. The
students were asked to use one or more of these
tools as a learning tool. Participation in this survey
was voluntary.
The presentation of the results of this survey
is divided into three parts: first, the use of the
tools by the students (before they started their
studies); second, the experiences the students had
made with the tools during the study; and third,
the potential future usage.
The article concludes with a discussion of
the results of this survey in contrast with other
empirical studies already published. Also, the

1403

Facilitating E-Learning with Social Software

limitations of this survey and ideas for further

research are pointed out.

Theoretical Framework
This part refers to the necessary theoretical
background required for the following empirical study, especially the areas of social software
and learning.

Social Software
The term social software emerged and came into
use in 2002 and is generally attributed to Clay
Shirky (2003). Shirky, a writer and teacher on the
social implications of Internet technology, defines
social software simply as software that supports
group interaction.
Another definition of social software can
be found in Coates (2005), who refers to social
software as software that supports, extends, or
derives added value from human social behaviour.
Users are no longer mere readers, audiences,
or consumers. They have the ability to become
active producers of content. Users can act in
user and producer positions and they can rapidly
change the position.
Nowadays the term social software is closely
related to Web 2.0. The term Web 2.0 was introduced by Tim OReilly (2005), who suggested
the following definition:
Web 2.0 is the network as platform, spanning all
connected devices; Web 2.0 applications are those
that make the most of the intrinsic advantages of
that platform: delivering software as a continually
updated service that gets better the more people
use it, consuming and remixing data from multiple sources, including individual users, while
providing their own data and services in a form
that allows remixing by others, creating network
effects through an architecture of participation,

1404

and going beyond the page metaphor of Web 1.0

to deliver rich user experiences.
Web 2.0 technologies such as blogs, wikis,
podcasts, and RSS feeds or discussion forums
have been dubbed social software because they
are perceived as being especially connected and
allow users to develop Web content collaboratively
and publicly (Alexander, 2006).
Until now, the Internet (Web 1.0) has one big
disadvantage: It is easy to get information in it,
but it is quite complicated and inconvenient to act
as an author and take part in the development of
content. Web 2.0 should enable all Internet users
to actively take part in the further development of
the Internet. Everyone should be able to contribute
easily. The focus of Web 2.0 is on the behaviour
of the user. It should empower people to communicate, collaborate, contribute, and participate.
This growing phenomenon is very interesting
and ought to be examined carefully in order to
understand how the Web is evolving and how this
continuously regenerative cycle of performance
and technological innovation empowers learning
by sharing (Thijssen & Vernooij, 2002).
Based on the key principle of the architecture
of participation, social software can be seen as
part of Web 2.0. Wikis, Weblogs, and discussion
forums are tools that are seen as social software
applications and were selected for further research
and the empirical study presented below.

Related Empirical Research

Institutions in the field of higher education have
made efforts to introduce various IT-supported
learning tools in the daily routine of students and
lecturers (Aggarwal & Legon, 2006; Dooley &
Wickersham, 2007; Duffy & Bruns, 2006; Evans
& Sadler-Smith, 2006; McGill, Nicol, Littlejohn,
Grierson, Juster, & Ion, 2005).
Published results of the usage of Weblogs in
the Prolearn project (http://www.prolearn-project.
org) have shown that a large majority of respon-

Facilitating E-Learning with Social Software

dents considers personalization and adaptation

of the learning environment as important and
crucial factors. Learning should be individualized
to become more effective and efficient. Personalization is a key element of the learning process,
and specific problems need specific solutions as
students differ greatly in their backgrounds and
capabilities.
Learning materials are typically too general
in order to cover a very wide range of purposes
and personal learning needs. Compared to classical learning, personalization can be the most
important added value that e-learning can offer.
With it, education can be optimised and adjusted
to various working conditions and needs because
students have different goals, interests, motivation
levels, learning skills, and endurance (Klamma
et. al., 2006).
Chao (2007) explored the potential uses of
wikis in the field of software engineering (38
participants), especially for software project team
collaboration and communication. Overall, 25
students agreed and 1 student disagreed (2 were
neutral) that the wiki is a good tool for project collaboration. Concerning the applications of wikis,
more than 23 students found that a wiki is a good
tool for maintaining a group diary, managing user
stories (project requirements), and project tracking and reporting. While a majority of students
found that a wiki is a good tool for updating a
project plan, managing acceptance tests, tracking
defects, and developing user documents, there
were also a significant number of students who
disagreed.
First results using wikis for collaborative writing (about 40 participants) also reported similar
results. In this study, students used wikis to write
articles partly together with the lecturer.
After early problems with using the software
and writing contributions in the wiki, students
were able to write articles by themselves or in
teams. The motivation among students was on

different levels, so the lecturer had to increase

it during lessons. Other students, however, were
highly motivated and were creating the content
and adding them to the wikis (Bendel, 2007).

Constructivism and Learning:

Presentation of the Learning Model
A constructivist point of view focuses on the
learning process by looking at the construction of
knowledge by an individual. As a consequence,
there is a recommendation to align learning environments, especially in the academic context,
and associated complex learning objectives with
constructivist learning principles (Du & Wagner, 2005; Jonassen, Mayes, & McAleese, 1993;
Tangney, FitzGibbon, Savage, Mehan, & Holmes,
2001). Learning is not seen as the transmission
of content and knowledge to a passive learner.
Constructivism views learning as an active and
constructive process that is based on the current
understanding of the learner. Learning is embedded in a social context and a certain situation
(Schulmeister, 2005).
The constructivist approach shifts learning
from instruction and design centered to learnercentered learning and teaching. The role of the
educator changes from directing the learner toward supporting and coaching the learner.
Baumgartner (2004) has suggested three different prototypical modes of learning and teaching. These three different modes of learning and
teaching can be neutral or specific so they can
be applied across all subject domains. Therefore,
each teaching model can be used to teach, for
example, sociology subjects as well as to teach
technical sciences. Learning can be portrayed
as an iterative process that can subsequently be
subdivided into different phases, which are summarized in Figure 1.
In particular, these three different prototypical
modes for learning encompass the following.

1405

Facilitating E-Learning with Social Software

Figure 1. Prototypical modes of learning and teaching (Baumgartner, 2004)

Learning and Teaching I:

Transferring Knowledge
At the starting point, the learner needs to be
provided with abstract knowledge to lay the
theoretical foundations and to understand relevant
signposts, road markings, and orientation points.
This kind of factual knowledge is static and has
little value by itself in real and complex situations.
It merely serves as a shortcut to prevent pitfalls
and to help to organize the students learning
experiences.
The knowledge of the student is based on
knowledge possessed by the teacher. Students
have to learn what teachers ask them to learn.
The teacher has the responsibility to make the
knowledge transfer as easy as possible.

Learning and Teaching II:

Acquiring, Compiling, and
Gathering Knowledge
In this section of the individual learning career,
the student actually applies the abstract knowledge

1406

and gathers his or her own experiences. In order

to limit the action and reflection possibilities, the
learner interacts within a somewhat restricted,
artificial environment, which is reduced in
complexity and easy to control by the teacher.
To provide feedback, the learning environment
is designed to include relevant devices where
students can deposit their interim products and
teachers can inspect them.
The emphasis in this model lies on the learning process of the student. Teachers try to help
the students overcome wrong assumptions and
wrong learning attitudes, and assist in the reflection process of the subject domain.

Learning and Teaching III:

Developing, Inventing, and
onstructing Knowledge
Teacher and learner work together to master
problems. This model includes problem generation
and/or invention. The environment is constructed
in such a way that it represents, at least in certain
aspects, reality or reality in a constrained form.

Facilitating E-Learning with Social Software

This model includes two-way communication on

equal terms, using either linguistic representations
or other adequate kinds of language.
Teaching III has strong links to constructivism.
From a constructivist point of view, learning is
considered as an active process in which people
construct their knowledge by relating it to their
previous experiences in complex and real situations in life. In their practical lives, people are
confronted with unique, unpredictable situations
whose inherent problems are not readily observable (Baumgartner, 2004).
Students should be enabled to invent new
things, and produce or generate new knowledge.
Consequently, learning and teaching at universities in most cases can be assigned to the requirements presented in Learning and Teaching II and
III. In order to achieve this goal, a special learning
environment must be provided.

Consequences for IT-Supported

Learning and Teaching
Computer software can be used for all three
models, ranging from programmed instruction (Learning/Teaching I) to problem-solving
software (Learning/Teaching II), to complex
simulations and/or so-called micro worlds (Learning/Teaching III). It is said that the inherent
nature of the Internet brings the real world into
the classrooms, and with its hyperlink structure
it clearly advocates the model of Teaching III
(Baumgartner, 2004).
The use of the Internet, especially through its
social software, gains importance because it can
contribute to exceed the limits of classical teaching models. By adapting learning and teaching
models to the new technical possibilities, the
roles of learner and teacher are becoming more
indistinct because the learner can take a central
part in the design and arrangement of the learning
process (Kerres, 2006).
Systems that support learners with respect
to the Learning Model III are called personal

learning environments (PLEs). PLEs are mostly

Web-based applications and are based on learning
management systems (LMSs; Seufert, 2007).
PLEs are personal and open learning environments, and they are suitable for cross-linking
content and people. Learners can use PLEs to
manage individual learning progress. They are
ideally available for lifelong learning and are
supported by the following processes.

setting up individual learning goals

planning and controlling ones own learning concerning the content as well as the
learning process
combining formal and informal learning
activities
communicating with peers during the learning process
establishing social networks or communities
of practice
using Web-based services, for example,
syndication
verifying the learning process with respect
to the learning goals

Unlike an LMS, which is usually related to one

special institution or to one special course, a PLE
is focused on the individual learner. A PLE should
combine a broad mixture of different resources
and subsystems in a personally-managed space
(Attwell, 2006).
In the previous decade, learning management
systems were developed that moved toward enterprise-level applications, but the wealth of new,
user-friendly, tools in the Web 2.0 environment
suggests that the all-in-one monolithic e-learning
systems may be entering a phase of obsolescence
by the ongoing development of the web (Craig,
2007).
Social software applications have the potential to cope with these requirements (Brahm,
2007).

1407

Facilitating E-Learning with Social Software

Description and
Classification of Social
Software Tools
In the following section, three social software
toolsWeblogs, discussion forums, and wikisare described in more detail and the tools
are compared. Students were able to select these
tools during the empirical study.

Weblog
A Weblog, a compound of Web and logbook, usually just called blog, is a Web site that contains
new articles or contributions in a primarily chronological order, listing the latest entry on top.
Primarily, a Weblog is a discussion-oriented
instrument especially emphasizing two functions:
RSS feeds and trackback. RSS feeds, also called
RSS files, can be read and processed for further use
by other programs. The most common programs
are RSS readers or RSS aggregators that check
RSS-enabled Web sites on behalf of the user to
read or display any updated contribution that can
be found. The user can subscribe to several RSS
feeds. Thus, the information of different Web sites
can be retrieved and combined. Preferably, news
or other Weblogs are subscribed to.
Trackback is a service function that notifies
the author of an entry in a Weblog if a reference
to this contribution has been made in another
Weblog. By this mechanism, a blogger (person
who writes contributions in a Weblog) is immediately informed of any reactions to his or
her contribution on other Weblogs (Hammond,
Hannay, & Lund, 2004).

Forum
A discussion forum or Web forum is a service
function providing discussion possibilities on
the Internet. Usually, Web forums are designed

1408

for the discussion of special topics. The forum

is furthermore subdivided into subforums or
subtopics. Contributions to the discussion can be
made and other people may read and/or respond
to them. Several contributions to a single topic
are called a thread.
The application areas of the two instruments,
Weblogs and forums, are quite similar. The most
essential differences between Weblogs and discussion forums can be described as follows:

A forum is usually located on one platform

while many bloggers develop their own,
individual environments. They connect
their Weblogs via RSS feeds and trackback
functions.
Through the integration of RSS files and
trackback functions, a discussion process
can be initiated and continued, crossing the
boundaries of the bloggers own Weblogs
without authors having to observe other
Weblogs.
Weblogs tend to be more people centered
whereas forums are more topic focused.
Through the use of Weblogs, learner-specific
learning environments can be constructed
without interfering with the learning environments of others (Baumgartner, 2004).

Wiki
A WikiWikiWeb, shortly called wiki, is a hypertext system for storing and processing information.
Every single site of this collection of linked Web
pages can be viewed through a Web browser.
Furthermore, every site can also be edited by
any person. The separation between authors and
readers who write their own text, and change
and delete it is obsolete as also third parties can
carry out these functions (Augar, Raitman, &
Zhou, 2004).

Facilitating E-Learning with Social Software

Learning Activities Supported by

Social Software
The integration of different social software tools
offers support in the following learning activities:

Learning from different perspectives: The

integration supports the exchange of ideas
as well as finding like-minded people. Furthermore, social software tools simplify the
process of establishing connections between
people of the same or similar interests.
Simultaneously, its open and expandable
philosophy supports going beyond the
thinking in groups (of a common interest) by
supporting diversity and bringing together
different perspectives and backgrounds
(Efimova & Fiedler, 2004; Schulmeister,
2004).
Synergies of self-organized and joint learning: Social Software tools provide a personal
learning area for their authors. However,
this does not force a general learning flow
or learning style. Learners are not alone
and can profit from the feedback of a community in order to examine and enhance the
development of their own ideas (Bttger &
Rll, 2004; Efimova & Fiedler, 2004; Fiedler,
2004).
Digital apprenticeship: Through reading
other wikis, forums, or Weblogs regularly,
beginners are enabled to learn from experts.
At the same time, they can actively participate in discussions beyond geographic or
thematic borders (Efimova & Fiedler, 2004;
Fiedler, 2004).
Support for the development of the ability to
learn (learning to learn): Through the publication of ones own thoughts and reflections,
content is made available for assessment as
well as for further development, thereby improving self-observation and self-reflection

skills. The knowledge change of the learner

will be improved (Baumgartner, 2005).
Support for reflexive writing: The simple but
efficient and rather robust encoding standard
usually used in social software allows for the
explicit modeling of content flows, feedback
loops, and monitoring procedures of various
kinds, thus supporting an ongoing reiterative
process of explication and reflection (Fiedler,
2004).

Integration of Social Software Tools

and the Learning and Teaching
Modes
Baumgartner (2004) has integrated different
types of content management systems in relation
to the most suitable learning and teaching mode.
He clearly states that the boundaries are overlapping and that every tool, in one way or another,
could be used for every teaching model. Figure
2 presents the integration of the social software
tools and the learning and teaching modes.
Weblogs and forums can be defined as discussion-oriented tools because the discourse and
exchange of ideas related to a certain topic is the
preeminent aim. Weblogs offer the possibility to
support all three phases of the learning process.
However, the main focus can be assigned to
Modes II and III.
Based on the multitude of interaction possibilities, wikis can be attached to Teaching III
(Baumgartner, 2004). Additional functions were
added to Weblog tools that go beyond the scope
of the central use of Weblogs; for example, longer
articles can also be stored. Through the creation
of directories, a structured collection of links can
be implemented.
Through the additional linking of Weblogs,
wikis, and forums, there is the possibility to develop a personal knowledge collection (Kantel,
2003).

1409

Facilitating E-Learning with Social Software

Figure 2. Prototypical modes and social software tool

Empirical Survey
The purpose of this survey was to determine
if the integration of Web-based social software
tools (wikis, discussion forums, and Weblogs)
are suitable to foster learning from the students
point of view.

Aim of the Survey and Methodology

Scrutinizing the possibilities and constraints of
social software tools (wikis, discussion forums,
and Weblogs) as personal learning environments,
students at Austrian universities were asked to use
one or more of the offered tools for their research,
homework, and documentation purposes. In most
cases, the collaboration of students was required
to perform the assigned tasks.
The students were asked to use the tools for one
course only during the winter term of 2006. Furthermore, there was no obligation for the students
to use a tool at all; they were just encouraged to
do so. Students were also offered the possibility
to use two or three tools; the selection was up to
the students.

1410

The courses were organized as blended-learning courses so they included on-campus lessons
and off-campus work in which the students could
work face-to-face or using the social software
tools.
More than 90% of the students attending the
courses took part in this survey. In order to give
the participants an impression of the functionality and usage of the tools, short presentations of
the tools were made by an instructor before the
students made their choice.
At the end of the testing phaseafter 4 weeks
of using the toolsselected students reported their
experiences with the tools used. Students who
had decided not to use the tools in the first place
got an impression about the usage, advantages,
and disadvantages of the tools from their fellow
students. Following these short presentations, a
questionnaire was completed that provided the basic findings for further inspection and research.
A total of 268 first-semester students of different Austrian universities in five selected courses
took part in this survey. The majority of the participants were between 18 and 20 years old. The
portion of female students was about 17%.

Facilitating E-Learning with Social Software

According to a survey conducted by Seybert

(2007) concerning gender differences in computer and Internet usage by young people (aged
between 16 and 24), there is no gap between
men and women in Austria. The proportion of
women and men (in the relevant age class) that
use a computer (almost) once a day is 72% the
same. A study by Otto, Kutscher, Klein, and Iske
(2005) indicates that there is a positive correlation
between a formal educational background and
the usage of the Internet in Germany: Beside
socio-cultural resources like family background,
peer structures and social support in general, the
formal educational background turns out to be the
main factor for explaining differences in internet
usage (p. 219). As a consequence, for the analysis
of the results of this survey, no distinction between
male and female students was made.
Table 1 presents the distribution of the participants concerning the degree programs the
students are attending.
For the further analysis of the results, no
distinction according to degree programs will
be made.
The questionnaire asked each participant questions about her or his subjective impression of the
application of the tools. It included 5-point Likert
scales for rating constructs such as eligibility,
perceived quality, or enjoyment.

Table 1. Distribution of students regarding degree

program

Management & Law

Management & IT
Management & Industrial
Engineering
Mechanical Engineering,
Electronics

Distribution
17%
31%
22%
30%

The study was conducted to find answers

about the:

usage of social software before the study

started,
selection of the offered tools,
perceived quality of the contributions and
the support for learning,
applicability of the instruments to support
communication and community building,
correlation of the usage for private and
educational purposes of the tools,
fun factor in using the instruments, and
potential future usage.

The results of the study are presented in three

parts:

Part 1: Analysis of the usage of wikis, discussion forums, and Weblogs of the students
before the study was started
Part 2: Experiences made with the tools
during the study
Part 3: Potential future usage of the tools

Part 1: Tool Selection and Prestudy

Usage
Due to the fact that the students could select the
tools on their own, the Table 2 shows the results
of this selection process.
According to Table 2, the combination of wikis
and discussion forums is the most selected combination of tools (42.9%), followed by wikis only
(23.1%) and discussion forums only (22.4%). In the
end, only five students (1.9%) did not take part in
the study; they did not select a tool, although they
first had the intention to do so. Only one student
used Weblogs only. Generally, Weblogs were not
used very intensively by the participants.
Table 3 shows the usage of the tools by the
participants before they took part in the study.
It indicates that wikis (76%) and discussion forums (78%) are currently the most widely used

1411

Facilitating E-Learning with Social Software

Table 2. Tools selected by the students

Only one tool selected

Wikis only
Discussion forums only
Weblogs only
More than one tool selected
Wikis and discussion forums
Wikis and Weblogs
Discussion forums and Weblogs
Wikis, discussion forums, and
Weblogs
No tool selected
No tool selected

tools. Weblogs are only used by 11% of the asked

students.
The results clearly show that the Weblog hype
had not yet reached the surveyed students. Due

Table 3. Students already using the tools

Yes
No

Wiki
76%
24%

Forum
78%
22%

Weblog
11%
89%

Table 4. Usage for private purposes

I totally agree
I generally agree
Neither...nor (neutral)
I slightly disagree
I disagree

1412

Wiki
33%
35%
9%
16%
8%

Percent

Number

23.1%
22.4%
0.4%

62
60
1

42.9%
1.9%
0.7%

115
5
2

6.7%

18

1.9%

to the fact that only about 11% of the students

are currently using Weblogs, the results for this
instrument are not published for the first part of
the analysis. When it comes to the potential future
usage of the instruments, Weblogs are taken into
consideration again.
The following section presents the results for
statements used in analyzing the usage in more
detail.
Tables 4 and 5 present the current usage of
the tools for private and educational purposes.
First, the statement I often use wikis or forums
for private purposes was presented.

Table 5. Usage for educational purposes

Forum
33%
29%
9%
17%
12%

Wiki
I totally agree
I generally agree
Neither...nor (neutral)
I slightly disagree
I disagree

57%
33%
3%
8%
1%

Forum
22%
29%
12%
24%
12%

Facilitating E-Learning with Social Software

Table 5 presents the results for the statement

I often use wikis or forums for educational
purposes.
A huge majority (90%) stated that they use
wikis for educational purposes and about two
thirds (68%) used wikis for private purposes.
Wikis are therefore more intensively used for
educational purposes than for private purposes,
whereas the usage of forums is exactly the opposite: They are more used for private purposes
than for education.
The responses of the students concerning these
statements were that wikis are foremost considered
as a source of serious information, whereas forums
are ideal for getting hints or clues to problems because of their privacy. Questions about computer
problems, computer games, leisure activities, and
so forth were mentioned. A repetition of this image can be identified when the disagreement with
the statement is analyzed; 29% of the students
do not or rarely use forums for private purposes
compared to 36% for education.

Quality and Support for Learning

This section presents the results of the study

concerning experiences with the usage of the
tools during the study.

The following refers to statements concerning the

quality of the contributions of wikis and discussion
forums and their support for learning.
The results of the statement The quality of
contributions in wikis or forums is in general
good are presented in Table 6. The contributions
of wikis are evaluated to be much better than
those of forums.
The surveyed pupils had the opportunity to
give reasons for their assessment concerning the
quality of contributions via additional qualitative feedback. The following summarizes the
addressed reasons.
One reason for the excelling grade for the quality of wikis is the Wikipedian community. The
term wiki is often seen as a synonym for the free
online encyclopedia Wikipedia (http://www.wikipedia.org). Wikipedia is widely used for a great
variety of tasks, including research on all topics
needed for educational and private purposes.
In contrast to the good evaluation of the
contributions of wikis, the open architecture of
wikis was also mentioned. In most cases, this
open architecture allows everyone to edit entries,
which results in the uncertainty of whether the
knowledge presented is correct or not. The quality
of contributions in discussion forums was rated

Table 6. Perceived quality of contribution

Table 7. Clarity of contributions

Part 2: Experiences Made During the

Study

I totally agree
I generally agree
Neither...nor (neutral)
I slightly disagree
I disagree

Wiki
38%
52%
10%
2%
0%

Forum
10%
31%
41%
15%
4%

Wiki
I totally agree
I generally agree
Neither...nor (neutral)
I slightly disagree
I disagree

2%
6%
29%
34%
29%

Forum
4%
18%
37%
27%
14%

1413

Facilitating E-Learning with Social Software

rather mediocre. Forums are primarily used for

technical problems, especially computer-related
problems; to get in contact with experts on certain
topics, and to get information on online games.
The next statement, The usage of wikis or
forums leads to misunderstandings and confusion, is about the clarity of the contributions.
Only a minority think that the contributions
are not clear and may lead to misunderstandings. In this case, wikis are also rated better than
forums.
The next statements addressed the support of
these instruments for learning. Table 8 summarizes the results for the statement When reading
contributions in wikis or forums, it is easier for
me to acquire the learning contents.
More than half of the students express that reading contributions in wikis is helpful for learning,

whereas only about 8% think that it is not helpful.

Compared to forums, wikis were again much better
evaluated, especially considering the big difference
from the negative evaluations of forums.
Table 9 presents the learning support achieved
by writing contributions. (When writing contributions in wikis or forums, it is easier for me to
acquire the learning contents.)
A different picture emerges in the statistics
when comparing the evaluation of how writing
an article or post supports the learning process.
Here, forums take the lead when it comes to positive assessment. In both cases, there was a large
number stating that writing is neither positive nor
negative. The majority of the students read rather
than wrote, while more students wrote in forums
than in wikis.

Table 8. Reading contributions helps to acquire

contents

Table 10. Applicability for communication

I totally agree
I generally agree
Neither...nor (neutral)
I slightly disagree
I disagree

Wiki
23%
36%
32%
5%
3%

Forum
8%
21%
31%
25%
15%

Table 9. Writing contributions helps to acquire

contents

I totally agree
I generally agree
Neither...nor (neutral)
I slightly disagree
I disagree

1414

Wiki
8%
13%
45%
14%
19%

Forum
7%
19%
34%
22%
17%

Wiki
I totally agree
I generally agree
Neither...nor (neutral)
I slightly disagree
I disagree

9%
33%
29%
15%
15%

Forum
39%
37%
17%
4%
3%

Table 11. Support for community building

I totally agree
I generally agree
Neither...nor (neutral)
I slightly disagree
I disagree

Wiki
10%
25%
39%
15%
10%

Forum
28%
32%
23%
11%
6%

Facilitating E-Learning with Social Software

Applicability for Communication and

Community Building
The statement was formulated as follows: Wikis
or forums are appropriate to support communication.
The results clearly demonstrate that discussion
forums are made for communication whereas wikis are rather seen as a kind of reference book or
encyclopedia, as already mentioned above.
The results of the next statement, Wikis or
forums support the setup of communities, can
be seen in Table 11.
Opinions about the applicability of wikis to
establish a community is split. About 35% say
that wikis are supportive of building a community
compared to 25% who said that wikis do not support
community building. The support of forums to build
a community is rated much better: 50% indicated
that forums are well suited to build a community.
These results were to be expected because they
confirm the nature of the instruments.

A majority enjoy using wikis (62%) and forums

(56%). Considering the percentage of students
who said that there is no (I disagree) or little
(I slightly disagree) fun when using these instruments, wikis (6%) are much better rated than
forums (18%).

Part 3: Potential Future Usage of the

Tools
The third section of the empirical study deals
with the potential usage by students who had not
used the instruments before the study. Students
gained knowledge and experiences by using the
tools during the study by themselves or on the
basis of the reported experiences made by their
fellow students.

Table 13. Future usage in educational context

(current nonusers)

Fun Factor in Using the

Instruments
In surveying whether students gain pleasure (I
enjoy using wikis or forums), wikis again came
out on top.

Table 12. Fun Factor in using the instruments

I totally agree
I generally agree
Neither...nor (neutral)
I slightly disagree
I disagree

Wiki
26%
36%
31%
5%
1%

Forum
19%
37%
26%
14%
4%

I totally agree
I generally agree
Neither...nor (neutral)
I slightly disagree
I disagree

Wikis
18%
36%
30%
9%
7%

Forums
16%
23%
16%
12%
33%

Weblogs
13%
23%
24%
13%
27%

Table 14. Future usage in private context (current nonusers)

I totally agree
I generally agree
Neither...nor (neutral)
I slightly disagree
I disagree

Wikis
11%
36%
30%
14%
9%

Forums
14%
23%
25%
7%
32%

Weblogs
9%
22%
24%
16%
28%

1415

Facilitating E-Learning with Social Software

The first statement, I will use wikis, forums,

or Weblogs for educational purposes in the future,
yielded the results in Table 13.
According to this study, wikis will have a
bright future and will be used often for educational purposes, whereas forums will be used less
often. About 54% of the surveyed students had
the intention of using wikis more or less often in
the future. About 16% did not think that they will
use wikis often in the future and 30% are not yet
sure if they will use this instrument.
The results for forums and Weblogs indicate
no clear trend, but forums were rated slightly
higher than Weblogs; 39% of the students stated
that they can imagine using forums in the future
for their education compared to 36% for Weblogs.
At the other end of the scale, 45% did not have the
intention to use forums (40% for Weblogs).
The statement I will use wikis, forums, or
Weblogs for private purposes in the future leads
to similar results.
From this point of view, wikis are again the
leading instrument, followed by forums and then
Weblogs.
It must be said that the responses to this set
of statements represented feelings, attitudes, and
opinions about instruments that had not yet been
used by the asked participants. The purpose of
posing these statements was to gain insight into
the mind-set in regard to these instruments.

Discussion
The results clearly show that wikis are currently
the most often used instrument and furthermore
have the greatest potential as a tool for learning
and knowledge management in the field of learning; these findings are in line with other empirical
studies (Bendel, 2007; Chao, 2007).
Other studies (McGill et al., 2005; Nicol &
MacLeod, 2004) report that a shared workspace
helps to support collaborative learning; the possibility of being able to access and contribute to

1416

the development of resources at any time and

from any location was especially appreciated by
the students.
The survey at hand made a distinction between
reading and writing contributions to wikis and
discussion forums. The results show that 59%
of students said reading contributions in wikis
is helpful for learning (stating I totally agree
and I generally agree) while only 21% stated
that writing in wikis is helpful for learning.
Reading contributions in forums helped 29% of
the participants, whereas writing in forums was
helpful to 26%. This survey supports the general
statement that a shared workspace that supports
a constructivist and learner-centered approach is
helpful for learning.
The pedagogical value in the context of learning is described in several publications (Babcock,
2007; Hurst, 2005). Weblogs can foster the establishment of a learning and teaching environment
in which students and teachers experience a
greater degree of equality and engagement. Du
and Wagner (2007) published results of a study
of an information systems undergraduate course
(31 participants). This study indicated that the
performance of students Weblogs was a significant predictor for learning outcomes, while traditional coursework was not. Moreover, individuals
cognitive construction efforts to build their own
mental models and social construction efforts to
further enrich and expand knowledge resources
appeared to be two key aspects of constructivist
learning with Weblogs. According to this study,
there is a potential benefit of using Weblogs as a
knowledge construction tool and a social learning medium.
According to the survey at hand, Weblogs are
not yet widely used, and their potential seems to
be limited. It can be assumed that these limited
prospects will change when the penetration of
Weblogs into the daily routine of the students
increasefor private as well as for educational
purposes.

Facilitating E-Learning with Social Software

To avoid possible pitfalls about the application

of these instruments in the context of learning,
some social and psychological issues must be
taken into consideration (Kreijns, Kirschner, &
Jochems, 2003). Social interaction is essential
for members of a team to get to know each other,
commit to social relationships, develop trust, and
develop a sense of belonging in developing a learning community. The size and the composition of
the learning communities seem to be important
factors in how interaction and communication
within the learning community will take place
(Dooley & Wickersham, 2007). There are also
many unresolved issues, like the provision of the
technology and the services, intellectual property
rights and digital rights management, the security
of data, access restrictions to the content, and
information ethics (Attwell, 2006; McGill et al.,
2005; Sharma & Maleyeff, 2003).

CONCLUSION
The aim of this contribution was to investigate
the experiences of students using social software
tools in the context of learning. Wikis, Weblogs,
and discussion forums are typical social software
tools and were used for this survey.
The results clearly show that wikis and discussion forums can support learning and collaboration. The usage of Weblogs in this study
was limited and hence no statements about their
applicability can be made. In order to assure a
successful implementation of these tools, social
and psychological issues must be taken into consideration as well.
The results of this study are the basis for the
introduction of social software into education
to help students set up individual learning environments. These learning environments should
support lifelong learning.
There are likely to be other unplanned consequences of the intensive use of the Internet in

general and social software especially. Further

research is needed to explore possible problems
and solutions.
The results of the empirical survey indicate
that a long-term study in combination with the
further development of social software tools may
be promising.

References
Aggarwal, A. K., & Legon, R. (2006). Web-based
education diffusion. International Journal of
Web-Based Learning and Teaching Technologies, 1(1), 49-72.
Alexander, B. (2006). Web 2.0: A new wave of
innovation for teaching and learning? Educause
Review, 41(2), 32-44.
Attwell, G. (2006). Personal learning environment. Retrieved May 17, 2007, from http://www.
knownet.com/writing/weblogs/Graham_Attwell/
entries/6521819364
Augar, N., Raitman, R., & Zhou, W. (2004, December 5-8). Teaching and learning online with wikis.
In R. Atkinson, C. McBeath, D. Jonas-Dwyer,
& R. Phillips (Eds.), Beyond the Comfort Zone:
Proceedings of the 21st ASCILITE Conference,
Perth, Western Australia. Retrieved May 17,
2007, from http://www.ascilite.org.au/conferences/perth04/procs/contents.html
Babcock, M. (2007). Learning logs in introductory
literature courses. Teaching in Higher Education,
12(4), 513-523.
Baumgartner, P. (2004). The Zen art of teaching.
Communication and Interactions in eEducation.
Retrieved May 17, 2007, from http://bt-mac2.fernuni-hagen.de/peter/gems/zenartofteaching.pdf
Baumgartner, P. (2005). Eine neue lernkultur
entwickeln: Komptenzbasierte ausbildung mit
blogs und e-portfolios. In V. Hornung-Prhauser

1417

Facilitating E-Learning with Social Software

(Ed.), ePortfolio Forum Austria: Proceedings of

the ePortfolio Austria 2005, Salzburg, Austria
(pp. 33-38).
Baumgartner, P., Hfele, H., & Maier-Hfele, K.
(2004). Content management systeme in e-education: Auswahl, potenziale und einsatzmglichkeiten. Innsbruck, Austria: Studienverlag.
Bendel, O. (2006). Wikipedia als methode und
gegenstand der lehre. In K. Hildebrand (Ed.),
HMD: Praxis der wirtschaftsinformatik: Vol.
252. Social software (pp. 82-88). Heidelberg,
Germany: dpunkt-Verlag.
Bttger, M., & Rll, M. (2004, December 15-17).
Weblog publishing as support for exploratory
learning on the World Wide Web. In P. Isaias, K.
Demetrios, & G. Sampson (Eds.), Cognition and
Exploratory Learning in Digital Age (CELDA
2004): Proceedings of the IADIS International
Conference, Lisbon, Portugal. IADIS Press.

whole-class online discussions. Quarterly Review

of Distance Education, 8(1), 1-8.
Du, H. S., & Wagner, C. (2005). Learning with
Weblogs: An empirical investigation. In Proceedings of the 38th Annual Hawaii International
Conference on System Sciences (HICSS05):
Track 1 (p. 7b). Washington, DC: IEEE Computer
Society.
Du, H. S., & Wagner, C. (2007). Learning with Weblogs: Enhancing cognitive and social knowledge
construction. IEEE Transactions on Professional
Communication, 50(1), 1-16.
Duffy, P. D., & Bruns, A. (2006). The use of blogs,
wikis and RSS in education: A conversation of
possibilities. In Proceedings of the Online Learning and Teaching Conference 2006, Brisbane,
Australia (pp.31-38).

Brahm, T. (2007). Social software und personal

broadcasting: Stand der forschung. In S. Seufert
& D. Euler (Eds.), Ne(x)t generation learning:
Wikis, blogs, mediacasts & Co. Social software
und personal broadcasting auf der spur (pp. 2038). SCIL Arbeitsbericht.

Efimova, L., & Fiedler, S. (2004, March 24-26).

Learning webs: Learning in Weblog networks.
In P. Kommers, P. Isaias, & M. B. Nunes (Eds.),
Web based communities: Proceedings of the
IADIS International Conference 2004, Lisbon,
Portugal (pp. 490-494). IADIS Press. Retrieved
May 17, 2007, from https://doc.telin.nl/dscgi/ds.py/
Get/File-35344

Chao, J. (2007). Student project collaboration using wikis. In Proceedings of the 20th Conference
on Software Engineering Education & Training
(pp. 255-261). Washington, DC.

Evans, C., & Sadler-Smith, E. (2006). Learning styles in education and training: Problems,
politicisation and potential. Education + Training,
48(2/3), 77-83.

Coates, T. (2005). An addendum to a definition of

social software. Retrieved March 4, 2007, from
http://www.plasticbag.org/archives/2005/01/an_
addendum_to_a_definition_of_social_software

Fiedler, S. (2004). Personal Webpublishing as a

reflective conversational tool for self-organized
learning. In T. Burg (Ed.), BlogTalks (pp. 190-216).
Retrieved May 7, 2007, from http://seblogging.cognitivearchitects.com/stories/storyReader$963

Craig, E. (2007). Changing paradigms: Managed

learning environments and Web 2.0. CampusWide Information Systems, 24(3), 152-161.
Dooley, K. E., & Wickersham, L. E. (2007).
Distraction, domination and disconnection in

1418

Hammond, T., Hannay, T., & Lund, B. (2004,

December). The role of RSS in science publishing: Syndication and annotation on the Web.
D-Lib Magazine. Retrieved April 20, 2007, from

Facilitating E-Learning with Social Software

http://www.dlib.org/dlib/december04/hammond/
12hammond.html
Hurst, B. (2005). My journey with learning logs.
Journal of Adolescent & Adult Literacy, 49(1),
42-46.
Jonassen, D. H., Mayes, T., & McAleese, R. (1993,
May 14-18). A manifesto for a constructivist approach to uses of technology in higher education.
In T. M. Duffy, J. Lowyck, D. H. Jonassen, & T.
Welsh (Eds.), Xpert.press: Vol. 105. Designing Environments for Constructive Learning: Proceedings of the NATO Advanced Research Workshop
on the Design of Constructivist Learning Environments Implications for Instructional Design
and the Use of Technology, Leuven, Belgium (pp.
231-247). Berlin, Germany: Springer.
Kantel, J. (2003). Vom Weblog lernen: Community,
peer-to-peer und eigenstndigkeit als ein modell
fr zuknftige wissenssammlungen. Retrieved
March 12, 2007, from http://static.userland.com/
sh4/gems/schockwellenreiter/blogtalktext.pdf
Kerres, M. (2006). Web 2.0 and its implications
to e-learning. In T. Hug, M. Lindner, & P. A.
Bruck (Eds.), Micromedia & E-Learning 2.0:
Gaining the Big Picture: Proceedings of Microlearning Conference 2006, Innsbruck, Austria.
University Press.
Kerres, M. (2007). Microlearning as a challenge
for instructional design. In T. Hug & M. Lindner
(Eds.) Didactics of microlearning. Mnster,
Germany: Waxmann. Retrieved March 10, 2007,
from http://mediendidaktik.uni-duisburg-essen.
de/files/Microlearning-kerres.pdf
Klamma, R., et al. (2006). Social software for professional learning: Examples and research. In B.
Kinshuk, R. Koper, P. Kommers, P. Kirschner, D.
Sampson, & W. Didderen (Eds.), Advanced Learning Technologies: ICALT 2006, Los Alamitos,
CA (pp. 912-917). IEEE Computer Society.

Kllinger, P. (2002). Report e-learning in

deutschen unternehmen: Fallstudien, konzepte,
implementierung (1. Aufl.). Dsseldorf, Germany:
Symposion.
Kreijns, K., Kirschner, P., & Jochems, W. (2003).
Identifying the pitfalls for social interaction in
computer-supported collaborative learning environments: A review of the research. Computers
in Human Behaviour, 19, 335-353.
Kutscher, N., Klein, A., & Iske, S. (2005). Differences in Internet usage: Social inequality and
informal education. Social Work & Society, 3,
215-233.
Lenhart, A., & Madden, M. (2005). Teen content
creators and consumers. In Pew Internet Project
Data Memo. Retrieved June 21, 2007, from http://
www.pewinternet.org/pdfs/PIP_Teens_Content_Creation.pdf
Lewinson, J. (2005). Asynchronous discussion
forums in the changing landscape of the online
learning environment. Campus-Wide Information
Systems, 22(3), 162-167.
McGill, L., Nicol, D., Littlejohn, A., Grierson,
H., Juster, N., & Ion, W. J. (2005). Creating
an information-rich learning environment to
enhance design student learning: Challenges
and approaches. British Journal of Educational
Technology, 36(4), 629-642.
Nicol, D. J., & MacLeod, I. A. (2005). Using a
shared workspace and wireless laptops to improve
collaborative project learning in an engineering
design class. 44(4), 459-475.
Oblinger, D., & Oblinger, J. (2005). Is it age or
IT: First steps towards understanding the Net
generation. In D. Oblinger & J. Oblinger (Eds.),
Educating the Net generation. Educause. Retrieved June 12, 2007, from http://www.educause.
edu/educatingthenetgen/

1419

Facilitating E-Learning with Social Software

OReilly, T. (2005). Web 2.0: Compact definition?

Retrieved May 17, 2007, from http://radar.oreilly.
com/archives/2005/10/web_20_compact_definition.html
Otto, H.-U., Kutscher, N., Klein, A., & Iske, S.
(2005). Social inequality in the virtual space: How
do young people use the Internet? Results from
empirical research about online use differences
and acquiring patterns of young people. Retrieved
January 16, 2008, from http://www.kib-bielefeld.
de/externelinks2005/Social_Inequality%20KIB.
pdf
Prensky, M. (2001a). Digital natives, digital immigrants: Part 1. Retrieved April 13, 2007, from
http://www.marcprensky.com/writing/Prensky
-Digital Natives, Digital Immigrants-Part1.pdf
Prensky, M. (2001b). Digital natives, digital immigrants: Part 2. Retrieved April 13, 2007, from
http://www.marcprensky.com/writing/Prensky
-Digital Natives, Digital Immigrants-Part2.pdf
Resnick, M. (2002). Rethinking learning in the
digital age. In G. S. Kirkman, P. K. Cornelius, J.
D. Sachs, & K. Schwab (Eds.), Global information
technology report 2001-2002: Readiness for the
networked world (pp. 32-37). New York: Oxford
University Press.
Richter, A., & Koch, M. (2007). Social software:
Status quo und zukunft (Tech. Rep. No. 200701). Fakultt fr Informatik, Universitt der
Bundeswehr Mnchen. Retrieved June 1, 2007,
from http://www.unibw.de/wow5_3/forschung/
social_software/
Schulmeister, R. (2004). Diversitt von studierenden und die konsequenzen fr elearning.
In D. Carstensen & B. Barrios (Eds.), Kommen
die digitalen medien in die jahre? Medien in
der wissenschaft (Vol. 29, p. 133-144). Mnster,
Germany: Waxmann.

1420

Schulmeister, R. (2005). Lernplattformen fr

das virtuelle lernen: Evaluation und didaktik (2.
Aufl.). Mnchen, Germany: Oldenbourg.
Seufert, S. (2007). Ne(x)t generation learning:
Was gibt es neues ber das lernen? In S. Seufert
& D. Euler (Eds.), Ne(x)t generation learning:
Wikis, blogs, mediacasts & Co. Social software
und personal broadcasting auf der spur (pp. 219). SCIL Arbeitsbericht.
Seybert, H. (2007). Gender differences in the use
of computers and the Internet. Retrieved January 15, 2008, from http://epp.eurostat.ec.europa.
eu/portal/page?_ pageid=1073,46587259&_
dad=portal&_schema=PORTAL&p_product_
code=KS-SF-07-119
Sharma, P., & Fiedler, S. (2004, June 30-July
2). Introducing technologies and practices for
supporting self-organized learning in a hybrid
environment. In K. Tochterman & H. Maurer
(Eds.), Proceedings of I-Know 04, Graz, Austria
(pp. 543-550).
Sharma, P., & Maleyeff, J. (2003). Internet education: Potential problems and solutions. 17(1),
19-25.
Shirky, C. (2003). Social software and the politics
of groups. Retrieved May 3, 2007, from http://
shirky.com/writings/group_politics.html
Tangney, B., FitzGibbon, A., Savage, T., Mehan,
S., & Holmes, B. (2001). Communal constructivism: Students constructing learning for as well
as with others. In C. Crawford, D. A. Willis, R.
Carlsen, I. Gibson, K. McFerrin, J. Price, & R.
Weber (Eds.), Proceedings of Society for Information Technology and Teacher Education
International Conference 2001 (pp. 3114-3119).
Norfolk, VA: AACE.
Thijssen, T., & Vernooij, F. (2002). Breaking
the boundaries between academic degrees and
lifelong learning. In T. A. Johannessen (Ed.),

Facilitating E-Learning with Social Software

Educational innovation in economics and business: Vol. 6. Teaching today: The knowledge of

tomorrow (pp. 137-156). Dordrecht, The Netherlands: Kluwer Academic.

This work was previously published in the International Journal of Web-Based Learning and Teaching Technologies, edited
by L. Esnault, Volume 3, Issue 3, pp. 16-33, copyright 2008 by IGI Publishing, formerly known as Idea Group Publishing (an
imprint of IGI Global).

1421

Section IV

Utilization and Application

This section introduces and discusses the ways in which particular technologies have impacted humanity
and proposes new ways in which IT-related innovations can be implemented within organizations and
in society as a whole. These particular selections highlight, among other topics, ubiquitous computing
applications in an educational setting and computer-mediated communication. Contributions included
in this section provide excellent coverage of todays world and insight into how our interaction with
technology impacts the fabric of our present-day global village.

1423

Chapter 4.1

Human Computer Interaction

and the Best Mix of Face-toFace and E-Interactions in
Educational Settings
Bolanle A. Olaniran
Texas Tech University, USA

Abstract
Recent trends and rapid improvement in technology such as computer-mediated communication
(CMC) and increasing bandwidth in the Internet
are facilitating increased electronic interactions
(i.e., e-interactions otherwise known as or commonly referred to as the human computer interaction (HCI)). CMC technology systems are a
common occurrence in educational institutions as
administrators attempt to encourage technology
usage and instructors race to learn and implement
CMC use in their classrooms and students demand
greater flexibility and control in how they learn.
Notwithstanding is the need to decide which forms
of HCI technology to use, how to use them, and
what benefits can accrue from such usage. The
discussion here explores each of these issues,
but more specifically will focus on addressing
the case for blending e-interactions with the

traditional face-to-face (FTF) communication

medium while addressing the appropriateness of
such combination.

introduction
Human computer interaction (HCI) occurs
through a host of information communication
technologies (ICT), specifically computer-mediated systems (e.g., e-mail, computer conferencing, video-conferencing) that facilitate electronic
interaction among people. A multitude of organizations involved in the knowledge management
industry are finding ways to incorporate computermediated communication (CMC) technologies
into their day-to-day operations as coordination
tools and learning curriculum dissemination
tools. HCI has also found explosive growth in
educational settings such as traditional and non

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Human Computer Interaction and the Best Mix of Face-to-Face and E-Interactions in Educational Settings

traditional universities to extend current instructional content delivery and to tap into distance
educational settings. The idea of CMC systems
for the instructional delivery tool to provide both
uniform and customized training is called electronic learning, which is otherwise referred to as
e-learning. Contemporary research has discussed
at length the potential benefits of CMC systems
in general and within distance education in particular (Barnard, 1997; McIsaac & Gunawardena,
1996; Yakimovicz & Murphy, 1995), but research
on joint use of the CMC system and face-to-face
(FTF) communication in regular classrooms is
scarce (Olaniran, 2004). This chapter will address
how using a combined CMC and (FTF) interaction could benefit courses in communication and
other social sciences.
Teachers are challenged to incorporate
information communication technology (ICT)
into their curriculum to enhance learning and
to prepare students for future careers (Althaus,
1997; Craig, 2001-2002; Snoeyink & Ertmer, 20012002; Witmer, 1998). ICT offers opportunities for
facilitating discussion and presenting forums for
varieties of opinions (McComb, 1994; Olaniran,
2001; Olaniran, Savage, & Sorenson, 1996, Witmer, 1998). HCI technology, specifically, CMC,
offers the opportunity for active and experiential
learning and its benefits in group activities have
been acknowledged (Craig, 2001-2002; Gasker
& Cascio, 2001; Olaniran, 1994; Olaniran, et al.,
1996).
In spite of the identified advantages, incorporating ICT especially CMC into classrooms
remains challenging. Some challenges facing
instructors regarding implementation of CMC
in classrooms and courses include selection and
usage long after adoption decisions are made. In
addition to technology issues, instructors need to
focus on pedagogical issues surrounding course
structure, and course management.

1424

Computer Mediated
Communication and Learning
Facilitation
The rampant effect of information communication technology (ICT) is constantly being felt
in contemporary organizations, interpersonal
interactions, and academic settings. ICTs are
instrumental in facilitating human computer
interaction, which underlies the computer mediated communication (CMC) in organizations,
classrooms, groups, and interpersonal contexts. As
a matter of fact, the issue facing most institutions
today (academic and non academic organizations)
is not whether to use CMC systems but rather how
to use them effectively.
There is a significant volume of literature
on CMC systems and distance education (e.g.,
Abrahamson, 1998; Twigg, 1997, Wegrif, 1998),
however, not all usages of CMC in learning environment are exclusively distance education. Some
research in CMC technologies concentrates on the
social effects of communication technologies (e.g.,
Hiltz & Turoff, 1978; Sproull & Kiesler, 1986).
No doubt, the demand for distance education
contributes to the proliferation of ICT as a way to
provide education and training. However, there is
a need to look at how CMC and other technologies facilitate learning and to identify other key
variables that must be accounted for in order for
effective learning to occur. This chapter seeks to
provide a practical guide to educators and learners
for enhancing learning through communication
technology.
The globalization trends in contemporary
organizations are putting priority on uniform and
customized training to the extent that institutions
are looking at e-learning to meet their curriculum
needs. As a result, varieties of online universities
like the University of Phoenix and Westwood College Online are developing in order to meet the
needs of non traditional students (e.g., corporate
travelers and expatriates). Furthermore, several
universities are throwing their support behind

Human Computer Interaction and the Best Mix of Face-to-Face and E-Interactions in Educational Settings

the development of technological centers such

as the teaching, learning, and technology center
(TLTC) to assist instructors in incorporating
CMC into teaching and learning. Specifically, the
developments in the use and adoption of ICTs in
the educational arena necessitate a closer look at
learning that takes place where these technologies
are being deployed.
There are several areas to examine when
dealing with e-learning or incorporating CMC
into courses. These include areas such as: the
interaction between students and technology,
the interaction between students and teacher, the
interaction between students and students. Robson
(2000) argues that learning theories are mediated
by technologies given that students often establish
relationships with varieties of technologies. In
essence, it is important to examine how students
learn when using CMC systems (Alavi, 1994; Du
& Wagner, 2005). Furthermore, it is hoped that
examining how students learn and interact with
technology will help instructors and, perhaps,
designers to determine the best way to apply
technology in course curricula.

How Students Learn with

Computer Mediated
Communication System

Perhaps the most prominent reason administrators and instructors offer for incorporating HCI
and other CMC technologies into course design
is the vision of personalizing learning to meet
the idiosyncrasies of students learning styles.
Innovative teachers see computers as tools to
help students in their learning needs with the
goal of tailoring course content to suit each student. Similarly, personalized learning is closely
related to active or transformational learning
in the sense that it allows for students to apply
instructional content while retaining information
longer than the lecture and print mode that is
characterized by traditional modes of instruction

(Olaniran, Stalcup, & Jensen, 2000). Students

apply computers to learning in several ways (1)
to access information, (2) to develop skills, and
(3) to cultivate appropriate behaviors and other
social skills (Robson, 2000). Each of the uses is
explored in the following paragraphs.

Information Access
CMC offers students the opportunity to scan
information through Website navigation (browsing) and by searching for particular information. For instance, in Web-assisted methods of
CMC, students are able to navigate instructors
Web pages to access course notes and other
instructional resources provided by instructors.
However, access to information is not restricted
to just instructors Web pages. Students can use
a browser to seek additional information or to
research similar topics and other topics of interest.
General browsing is categorized as low cognitive
activity in learning (Jacobson, 1995). Browsing
can, however, facilitate a high cognitive thinking
because students learn as they seek varieties of
information and when they develop a method to
categorize information in a structured manner
(i.e., retention and recall) at a later time (Jacobson,
1995; Robson, 2000). At the same time, one must
not forget that students differ even in the way they
search and learn. For instance, Robson (2000)
argues that some students prefer general browsing
to specific keyword searches. Students also differ
in the way they avoid disorientation or get lost
in the multitude of information (Hara, Bonk, &
Angeli, 2000; Wegrif, 1998). For example, some
students are able to navigate through multiple sites
and keep track of information obtained, while
others are stressed and are unable to distinguish
one network from another (Duchier, 1996).
As far as information access and personalized
learning is concerned, Olaniran (2004) argues
that a mixed or combined CMC and face-to-face
communication media (CMC/FTF) offers a good

1425

Human Computer Interaction and the Best Mix of Face-to-Face and E-Interactions in Educational Settings

way of addressing students needs and at the

students convenience and comfort levels. In a
study evaluating the combined CMC/FTF media,
Olaniran (2004) finds that students using CMC
were three times more likely to initiate contact
with the instructor about course content than in an
FTF setting. Specifically, the results also indicate
that students are twice as likely in the use of the
CMC medium when compared to FTF medium to
address questions relating to individualized learning. Thus, the CMC/FTF design structure offers
a way to compare both media while presenting
a method for accounting for individual students
interest, knowledge level, and learning styles in a
gradual manner (Gasker & Cascio, 2001; Olaniran,
et al., 1996). CMC structure allows teachers to
target different learning styles and knowledge
levels without confusing other students through
individualized contacts and course designs.
Other factors influencing access include the
propensity to adopt and use communication technologies. Adopting CMC for course design and
students learning starts with access to technology
and the willingness to use it (Olaniran, 1993). A
study on attitude and perceptions of e-learning
found ease of access as the top reason in students
use of CMC at 47 percent, while course relevance
to future careers and user friendliness were at a
distant second and third (i.e., 29 percent and 24 percent respectively) (Vaughan & MacVicar, 2004).
This finding reinforces the need for instructors to
pay specific attention to issues of accessibility. It
is often taken for granted that students have access to e-mail and Internet especially in the USA.
This may be obvious because some students have
access to internet while on campus whereas they
do not have access at home. More troubling on
most university campuses is the fact that students
often shy away from using a university assigned
e-mail account for other third party client e-mail
such as Yahoo, MSN, and Hotmail and the likes.
The challenge is that some university are clamping down on the use of these third party e-mail
accounts as a preventive measure against the

1426

spread of viruses and worms to their systems

(Curran, McIntyre, Meenan, McCloy, & Heaney,
2004). Consequently, some students are unable
to access materials sent by instructors.

Skill Development
Skills take two forms. The first deals with the
technical skills that allow both student and instructor to use hardware and software or evaluate
data. The second involves the social skills that are
deployed by technology users during interactions
(Robson, 2000). The technical skills address the
issue of competence with the technology in use,
while social skills deal with issues of etiquette
and discernment of contextually appropriate
behaviors. In CMC research, Olaniran (2004)
addresses the adjustment process that accompanies students interaction, which range from
use of all caps to protocols of posting messages
in a threaded discussion. Personal experience in
threaded discussions also suggests that students
often participate in threaded discussion by breaking the thread; that is, starting a new thread while
still on the same topic.

Nurturing Appropriate
Attitudes
As students use CMC they develop attitudes
toward the technology medium (Olaniran, 2004)
as well as developing professional attitudes and
values consistent with the course content or discipline. Robson (2000) argues that the learning
of concepts, skills, and development of attitudes
equates cognitive development on the part of the
student, which represents an important goal in all
forms of education. However, it is not certain that
attitude development toward technology translates
positively to professional attitude development.
For instance, how users (i.e., students) perceive a

Human Computer Interaction and the Best Mix of Face-to-Face and E-Interactions in Educational Settings

CMC medium based on ease of use affects decisions to adopt and continue using the technology
(e.g., Olaniran, 1993; 2004; Vaughan & MacVicar,
2004). Thus, one could argue that when students
are frustrated within a CMC system, they may
abandon the technology or fail to adopt it.
At the same time, students never get to develop
professional attitudes or achieve high cognitive
development when they fail to use a particular
communication technology and thus, they rob
themselves the opportunity to learn the technology even when they know it might be beneficial to
them. For instance, in a course utilizing Web authoring tools, I have found few students deciding to
skip certain assignments (designing personal Web
page) simply because of the perceived difficulty
with the technology software. Students discuss
the lack of competence in the Web authoring tools
offered through the university computing services
center. Therefore, it is imperative for instructors
to realize that simply because the tool is available
and the instructor has offered some training on
the tools does not mean that students grasp the
concept and are ready to perform certain tasks
on their own. This is especially the case with
non-technically oriented courses in liberal arts
and social sciences. Robson (2000) echoes this assessment when arguing that access to knowledge
through technology does not guarantee improved
outcomes (p. 157). Instructors are then critical
about facilitating students learning; how they design the learning experience is what will facilitate
the desired outcomes. This raises the issue of how
instructors can assist students in learning, which
is discussed in the next section.

Helping Students to Learn

with CMC
Incorporating CMC to a course design is not a
panacea. However, how technologies are applied
will be the major determinant of the learning that
takes place. CMC platforms can be used simply

as a conduit to create printed materials or it can

be used as an all-encompassing learning forum
(access, knowledge, interaction forum etc.). Electronic media provide a platform for interaction
and can be a source of knowledge (Du & Wagner,
2005; Leidner & Jarvenpaa, 1995). For instance,
the Internet is viewed as a gateway to information
(Uys, 1997) and a source of knowledge.

Critical Thinking and

Transformative Learning

The use of CMC in teaching courses can foster
transformational learning. The transformation
learning provides an opportunity to contrast
experiential knowledge (i.e., previously acquired
information) with new information in order to
make informed decisions about which to accept
(Olaniran, 2004). The transformational learning
identifies three factors (1) Validation of prior
knowledge, (2) modification of prior knowledge,
and (3) replacement of prior knowledge (Craig,
2001-2002; Salaberry, 2000). The process is
facilitated through the direct communication
process enhanced in CMC (Alavi, 1994; Mason
& Bascich, 1998). Olaniran (2004) reported evidence of transformation learning when students
challenged their existing beliefs about both CMC
and FTF media, that is, some changed their prior
beliefs after they were confronted with new information and experience. The best evidence of
transformation learning reported by the learners
is illustrated in the understanding of the reason
for the change however.
It is possible that the change in opinion of both
communication media could be attributed to the
course design. The CMC and FTF format helped
fortify the transformational learning experience
by providing students with practical use of communication technology, which provides students
with the opportunity to validate information
(beliefs and facts) about both study and the communication media (CMC vs. FTF). The learning

1427

Human Computer Interaction and the Best Mix of Face-to-Face and E-Interactions in Educational Settings

process that occurs can then enhance greater

retention of information about course content
and attitudes toward technology. Furthermore,
interaction occurring when course materials are
provided with CMC is said to be in line with the
principles of contemporary learning strategies
(Du & Wagner, 2005; Hoare & Race, 1991). As a
result students are able to develop skills in critical
thinking, which requires the ability to reflect on
the interaction with the aid of the record storage
provided by CMC technologies. While self-reflection is possible in FTF setting as well, it is
usually more difficult to do in the absence of good
note taking with FTF context than it is in CMC,
because note taking is flawed with inaccuracies
due to failed memory recall. Robson (2000) puts
things into perspective when arguing that in
the past, students achieved cognitive learning
by linear texts and discussion. However, with
technology, students are able to seek information
in order to extend their knowledge, reflect on information, and then, develop their own views of
the information and the foundation upon which
the information is based. Du and Wagner (2005)
on Weblogs refer to power of CMC to prevent
free-riding that often occurs in FTF groups and
other collaborative learning because assessment
is graded on individual effort. Students have to
take ownership of their individual contribution
knowing that it will be evaluated in direct comparison with their peers.

Giving Students the Control

Incorporating CMC into a course allows students
to take charge of how they learn. The interaction
with technology provides some level of flexibility
in how students adapt technology to meet their
individual needs. Olaniran (2004) argues that
the ability and the freedom to ask the instructor
questions directly show that CMC has a tendency
to foster a greater sense of control in how they
learn and in the process of learning. For instance,
students do not have to wait in line to talk to the
1428

instructor or wait for appointment. The significance is that the traditional gate-keeping role of
a secretary is reduced. The notion of direct
manipulation (Fong & Kwan, 2003) speaks to
the idea, where a task as simple as icon dragging
or cursor position on a computer screen provide
students the knowledge about the technology and
also about the task. Indirectly, Rubens et al. (2005)
address control when discussing the flexibility that
technology design affords for knowledge building.
By knowledge building, students instead of passive
participation in class discussion are individually
responsible for identifying ideas and to expand
their knowledge base on those ideas.
Furthermore, Olaniran (2004) finds that
increased control in access to instructors with
CMC provides students opportunity to share more
personal information with the instructor, which
further aids learning as students are able to develop
a rapport and comfort with an instructor, course,
and the technology. On the contrary, traditional
FTF interaction limits time for sharing such information. At the same time, a balance needs to
be struck between offering users or students the
control to accomplish a task versus control that
can be counter productive. Fong and Kwan (2003)
offer the following advice for effective electronic
learning that users need to have technologies that
provide them with knowledge of getting their
work done without destroying their data. While
technology is supposed to provide a means for
accomplishing a task in both an effective and efficient manner, they are known to be problematic
either through mistake or malfunctioning. Thus,
it is advised that technology offer tool box or
alert box (i.e., wizards) that can notify students of
impending mistakes such as accidental deletion
of document (Fong & Wang, 2003).

Course Management
The goal of any course regardless of the communication medium (technological or traditional) is

Human Computer Interaction and the Best Mix of Face-to-Face and E-Interactions in Educational Settings

to help students build knowledge and demonstrate

mastery of the concepts. Thus, it is important that
instructors use technologies in a way that facilitates this process by selecting CMC hardware and
software that create the context where students
are actively engaged in the knowledge building
and overall learning process. The emphasis on
individualized or personalized instruction can
enable students to learn, however, a mismanagement of the process can be counter productive and
inhibit learning. For instance, technology media
can serve as a teaching mechanism in which
case a good interface between the technology
and students is required. Students must not be
left to fend for themselves (e.g., learning to use
the technology features). Robson (2000) hints at
this when she suggests that teachers roles call
for them to manage rather than mere pointing
students to information. As instructors perform
their managerial duties, attention must be given
to how students learn, process information, and
select technology that can help accomplish these
goals. For example, in CMC using computer conferencing or threaded discussion, students need to
know that they do not have to depend only on the
instructor or their own ability to obtain answers
to questions (Olaniran, 2004). Students can post
their questions on the discussion board and ask
others for help. Then instructors can monitor the
discussion to mine information for areas of course
or learning that instructor can offer assistance and
foster deeper knowledge.
As managers, instructors can input messages
to keep discussions focused on task goals and
to prevent disruptive students from taking over
class interaction. The CMC/FTF course design
can allow instructors to identify concepts from
the electronic discussion that are suitable for
further FTF discussion. Similarly, tools offered
within CMC media should allow for students to
focus on learning rather than housekeeping tasks.
Features such as auto save, auto archiving, and
Wizards are tools that instructor can use to assist
students to focus on learning rather than technical

component that can frustrate students experience

and learning. The degree which CMC technologies incorporate course management tools will
influence the quality of the course and ultimately
students learning and experience.
Instructors must decide ahead of time the degree to which a course needs to facilitate relational
development among students. Therefore, it may
be necessary for CMC users to establish relationships with one another and gain confidence in the
system before proceeding in its use. The need
for relational development in student-to-student
interaction reinforces the need for supplementing
CMC design courses with the FTF interaction
(Olaniran, 2004; Wilson & Whitelock, 1998).
Another issue of course management in CMC
courses is the degree of shift in learning control
and responsibilities. CMC can shift the burden of
learning from teachers to others such as students
and other experts, such that teachers are able to
concentrate on other aspects of learning. For
instance, Olaniran (2004) finds a considerable
drop in the proportion of direct teacher-student
FTF interaction in comparison to the traditional
FTF course. However, the reduction in time spent
in FTF interactions was supplanted by increase
in time needed for course management because
instructors had to spend more time reading
students contributions and providing individualized feedbacks. Specifically, the study reported
increase in the proportion of students contributing to in-class discussions on a consistent basis
from 56 percent for FTF to 79 percent in CMC
(Olaniran, 2004, p. 152).

Designing CMC into Courses

A major challenge facing institutions and instructors today is not whether the CMC technology
works. The challenge is making the technology
work within different pedagogical goals in a way
that accomplishes the idiosyncratic needs of different instructors and disciplinary demands. This

1429

Human Computer Interaction and the Best Mix of Face-to-Face and E-Interactions in Educational Settings

is where CMC teachers, vendors, and software

designers come in and can make a difference.
One of the major benefits of any communication technology in a learning environment is
to increase interactivity or active participation
among students. But more importantly, CMC
technologies must offer students a way to apply and
extend course contents beyond the confines of their
classrooms. CMC, if used appropriately, can foster
critical thinking along with greater interactivity
with other students (Olaniran, et al., 1996). By
implication, it appears that students are expected
to take initiative not only in their classrooms but
also in their socio-cultural environments. Thus,
educators are increasingly demanding more out of
their students with the expectation that students
will become active cultural change agents who
are critical consumers of information (McComb,
1994). Contemporary education must find a way
to engage learners experiences, concerns, and
voices (Alavi, 1994; Du & Wagner, 2005; Freire,
1983; Leidner & Jarvenpaa, 1995). This is a
paradigmatic shift from the traditional mode of
education that stresses passive reaction to learning contents to the one that emphasizes active or
proactive creators and participants in how they
learn (McComb, 1994; Olaniran, 2004; Olaniran,
et al., 1996). CMC systems designers must find
a way to create suites of hardware and software
that help accomplish the following:

1430

Allows students to connect with other learners experiences

Puts teachers in a position to study and
understand how best to adapt technology
to students experiences in a way that leads
to learning
Allows learning to take place outside the
linear mode of instructors transmitting or
being the sole provider of knowledge
Allows technology to extend the boundaries
of learning

CMC is not a cure-all but it can offer instructors

the opportunity to turn over learning responsibilities to the direct control of learners. Olaniran
(2004) argues that simply adding CMC to a course
will be counterproductive, however, when it is
used to facilitate course goals, then it will make
a significant contribution to learning processes
and students, as well as instructors, would benefit
from its deployment. McComb (1994) concurs
when she suggests that CMC should not be used
as a simple replacement for FTF communication. Other scholars have argued that using CMC
in a course design does not mean that students
will take advantage of it by taking the required
initiatives (Olaniran, et al., 2000; Witmer, 1998).
Quite the contrary, as Olaniran (2004) reported
an initial indifference in communication course
among graduate students who are expected to
show greater latitude in taking responsibility for
how they learn than the undergraduate students.
Students resistance to technology is to be expected but this can be overcome by instructors.
One way of making sure that initial resistance
is overcome is by making sure that a significant
portion of course evaluation is tied to performance
and usage of the system (Olaniran, 1995, 2004).
Furthermore, CMC can be explored or promoted
by teachers from the standpoint that it offers
students opportunity to allow their voices to be
heard and consequently, it is useful for curricula
that aim to facilitate students involvement and
in encouraging them to develop initiatives (Du
& Wagner, 2005; McComb, 1994).

Feedback and Assignment

Turnaround
CMC facilitates direct feedback without having
to go through traditional gate-keeping roles of
secretaries. However, affecting feedback is the
issue of assignment turnaround. While CMC
allows for assignment to be electronically submitted and graded, it is important for instructors to

Human Computer Interaction and the Best Mix of Face-to-Face and E-Interactions in Educational Settings

understand that they would have to develop the

habit of turning assignments and projects around
expeditiously. Students can become impatient
and frustrated when their assignments are not
returned in a timely manner. The resulting
frustration can also lead to failure to continue to
use the technology (Olaniran, 2004). Thus, it is
imperative that instructors understand that rules
that work in other classes where CMC is not in
use may fail miserably when applied in the CMC
course design.
Another challenge with feedback in CMC lies
in the fact that CMC is immune to large class size
when participation is concerned. Accompanying
large class sizes is also the need to grade a large
number of assignments, yet the demand for fast
turnaround time persists. One way to overcome
this is to develop an electronic toolkit consisting
of a glossary of commonly found errors, which
consists of pre-written short text files about different errors (McComb, 1994). When an error is
encountered, an instructor can quickly insert the
accompanying text comments in student papers.
First the short text files allow standardization of
certain feedback on papers while allowing flexibility for personalized feedback. Second, the
process allows instructors to meet the goal of fast
turnaround. Third, the process allows instructor
to move quickly through large volume of papers
with relative ease. Fourth, the process safeguards
against bad handwriting that may defeat the
original purpose of feedback.
In the process of feedback, McComb (1994)
considered the record-keeping feature of CMC
as a useful tool for training students from comparison standpoint. Specifically, she indicates
that she can refer to her notes to see if students
follow her suggestions. The record also provides a
good way to monitor students progress on activities and over a period of time. The record offers
teachers the means for comparing a particular
assignment with those of other groups and to
present examples of poorly completed assignments
against those that are exceptional (Du & Wagner,

2005; McComb, 1994; Phillips & Santoro, 1989).

Furthermore, CMC allows students, when given
access, to monitor their individual or collective
performance on a regular basis and at their own
convenience. For instance, my university provides
online course management tool (i.e., Raiderlink)
that instructors can use to provide students their
grades and other performance evaluation measure.
However, students also use it to stay on top of
their grades and they use it to contact instructors about any discrepancy during the course,
rather than having to wait until after the course
is over or when grades have been submitted to
the registrars office.
When designing CMC system into courses,
one of the benefits is the ability to post messages
and share files with participants. This becomes
an invaluable tool that instructors can explore
with learners especially in group interactions
and other problem solving contexts. Readings,
course notes, and supplemental resources can
be disseminated via Internet and CMC forum.
Learners can interact by sharing their comments
with one another and engage in in-depth discussion with their peers and instructors. According
to Bailey and Cotlar, (1994) instructors can ask
different questions to stimulate interaction and
thinking in CMC forum between students and
multiple instructors.

Challenges and
Recommendations for
CMC Structure
A major challenge facing instructors in incorporating CMC into course design involves deciding
which technology to use and when to use it. Asynchronous CMC (e.g., threaded discussion, bulletin
board, e-mail) are popular because of the set up
ease. The fact that asynchronous design allows
participants (students and Instructors) to reflect
before posting contributions is also appealing.

1431

Human Computer Interaction and the Best Mix of Face-to-Face and E-Interactions in Educational Settings

At the same time, the design offers the quickest

solution because most existing university and IT
infrastructure require little modification to add
CMC to course structure. However, synchronous
CMC course design (e.g., chat, video conferencing)
where participants interact, at the same time, offer
a greater level of interactivity and immediate feedback (Olaniran, 1994). Furthermore, synchronous
interaction allows participants to interact using
combination of voice, video, and audio channels,
which can enhance students learning experience.
With multiple channel cues media the barriers
of time, distance, and culture can be overcome.
(e.g., Solomon, 2001)
Synchronous CMC course structure could
be problematic for distance learning where participants are spread across different geographic
time zones. However, a recommendation for
overcoming the challenge is to offer a mixed
structure design, where asynchronous CMC may
be the primary design. At the same time, offering
synchronous interaction features such as the chat
feature would significantly enhance student-tostudent interactivity. Courses requiring group
writing or decision making would also benefit
when participants can either set up a time when
all or most group members can meet to put the
project together or write a paper. An alternative is
for the instructor to devote class time to accommodate synchronous interactions for students,
provided the infrastructure exists.
It is also recommended that software designers needs to help instructors overcome some
of the challenges in CMC. One is the issue of
bandwidth, which prevents against implementing
more sophisticated CMC applications such as
those allowing audio and video channel cues, the
likes of computer videoconferencing and internet
protocol television (IPTV). Creating third-party
software that offers multiple cues and ease of
deployment by instructors and easy access by
students is called for. In the meantime, Olaniran
(2004) recommends a combined method that allows for both FTF and CMC such that users do

1432

not feel like they are giving up one medium for

another. Furthermore, a combined media design
simulates interactions similar to those found in
most organizational settings, where members
are often allowed to use multiple communication
media for interaction. Similarly the combined FTF
and CMC design provide students time to adapt
to the CMC system. Thus, while people may be
using the CMC medium, they are still able to
interact via FTF. The use of combined media in
course design is not about CMC technology, but
rather about the use of CMC to facilitate interaction and learning.

Addressing Students Needs

Another challenge facing CMC and course
design involves the students needs. Outside
course content, the primary challenge facing an
instructor is in accommodating different needs
(e.g., learning styles and preferences) of the students. Consequently, instructors must develop a
way to identify students needs and use the CMC
system in a way that facilitates personalized
learning through multiple strategies. Therefore,
it is recommended that teachers break away from
the linear mode of teaching that characterizes a
lecture style. This can be done via prompt feedback through e-mail or posting to bulletin boards,
discussion forums, along with reflective Weblogs.
Another pressure placed upon instructors is that
they need to be well informed by having current
and adequate information to deal with individual
students problems. Instructors can overcome
this challenge by offering students toolkits or
resource links where students can go to obtain
desired information. Resources links are useful
in empowering students to take control of the
way they learn and recognize that learning exists
beyond the classroom walls. The need for stressing individualized learning also demands some
changes in procedures and course design. It is
recommended that the training manual or Web

Human Computer Interaction and the Best Mix of Face-to-Face and E-Interactions in Educational Settings

information must be easy enough for inexperienced users to understand. This simplification
will help to reduce some anxieties with computer.
However, the CMC training manual or Web
page should not prevent sophisticated users from
enjoying the systems benefits. Olaniran (2004)
advises that CMC should be used in a way that
facilitates creativity. Encouraging creativity can
be done when students are allowed to explore
the medium for other uses that deviate from the
courses primary purpose such as socialization
and recreation activities (Olaniran, 1993). A
benefit from creative use of CMC is that it allows
meta-interaction whereby students are able to
explore and develop the novel use of the system
as well as to seek help from one another when
they encounter problems. Finally, the opportunity
for students to help or coach one another in CMC
forum may alleviate the need for instructors to
be the only source of help.

References
Abrahamson, C. E. (1998). Issues in interactive
communication in distance education. College
Student Journal, 32(1), 33-43.
Alavi, M. (1994). Computer-mediated collaborative learning: An empirical evaluation. MIS
Quarterly, 18(2), 159-174.
Althaus, S. L. (1997). Computer-mediated communication in the university classroom: An experiment with on-line discussions. Communication
Education, 46, 158-174.
Bailey, E. K., & Coltar, M. (1994). Teaching via
the Internet. Communication Education, 43(4),
pp. 182-193.
Barnard, J. (1997). The World Wide Web and
higher education: The promise of virtual universities and online libraries. Educational Technology,
37(3), 30-35.

Bonk, C. J. & Cummings, J. A. (1998). A dozen

recommendations for placing the student at the
centre of Web-based learning. Educational Media
International, 35(2), 82-89.
Craig, D. V. (2001-2002). View from electronic
learning environment: Perceptions and patterns
among students in an online graduate education
course. Journal of Education Technology Systems,
30(2), 197-219.
Curran, K., McIntyre, S., Meenan, H., McCloy,
F., & Heaney, C. (2004). Civil liberties and computer monitoring. American Journal of Applied
Sciences, 1(3), 225-229.
Du, H. S., & Wagner, C. (2005). Learning

with
Weblogs: An empirical investigation. IEEE Proceedings of Hawaii International Conference on
System Sciences, pp. 1-9.
Duchier, D. (1996). Hypertext, NY: Intelligent
Software Group. Retrieved 11/10/2005 www.isg.
sfu.ca/~duchier/misc/hypertext_review/chapter4.
htm
Fong, J., & Kwan, I. (2003). Effective e-learning
by using HCI and interactivity on data modeling.
International Journal of Computer Processing of
Oriental Languages, 16, 293-310.
Freire, P. (1983). Pedagogy in process: the letters
to Guinea-Bissau. New York: Continuum.
Gasker, J. A., & Cascio, T. (2001). Empowering
women through computer-mediated class participation. Affilia: Journal of Women & Social Work,
16(3), 295-313.
Hara, N., Bonk, C. J., & Angeli, C. (2000). Content
analysis of an on-line discussion in an applied
educational psychology course. Instructional
Science, 28, 115-152
Hiltz, S. R., & Turoff, M. (1978). The network
nation: Human communication via computer.
Reading, MA: Addison-Wesley.

1433

Human Computer Interaction and the Best Mix of Face-to-Face and E-Interactions in Educational Settings

Jacobson, F. (1995). From Dewey to Mosaic:

Considerations in interface design for children.
Internet Research, 5(2), 67-73.
Hoare, R. M., & Race, W. P. (1991). Computer conferencing: A versatile new undergraduate learning
process. SIGUCCS Newsletter, 20, 18-24.
Leidner, D., & Jarvenpaa, S. (1995). The use of
information technology to enhance management
school education: A theoretical view. MIS Quarterly, 19(3), 265-291.
Mason, R., & Bacsich, P. (1998). Embedding
computer conferencing into University Teaching.
Computers & Education, 30, 249-258.
McComb, M. (1994). Benefits of computer-mediated communication in college courses. Communication Education, 43, 159-170.
McIsaac, M. S. and Gunawardena, C. N. (1996).
Distance education. In D. Johnassen (Ed.), Handbook of research for educational communications
and technology (pp. 403-437). New York: NY:
Macmillan.
Olaniran, B. A. (1993). Integrative approach for
managing successful computer-mediated communication technological innovation. Ohio Speech
Journal, 31, 37-53.
Olaniran, B. A. (1994). Group performance in
computer-mediated communication and face to
face meetings. Management Communication
Quarterly, 7, 256-281.
Olaniran, B. A. (1995). Perceived communication
outcomes in computer-mediated communication:
An analysis of three systems among new users.
Information Processing and Management, 31,
525-541.
Olaniran, B. A. (2001). The effects of computermediated communication on transculturalism.
In V. Milhouse, M. Asante, & P. Nwosu (Eds.)
Transcultural realities (pp. 83-105). Thousand
Oaks, CA: Sage.

1434

Olaniran, B. A. (2004). Computer-mediated communication as an instructional Learning tool:

Course Evaluation with communication students.
In P. Comeaux (Ed.), Assessing online teaching &
learning (pp. 144-158). PA: Anker Publishers.
Olaniran, B. A., Savage, G. T., & Sorenson, R. L.
(1996). Experiential

and experimental approaches

to face-to-face and computer mediated communication in group discussion. Communication
Education, 45, 244-259.
Olaniran, B. A., Stalcup, K., & Jensen, K. (2000).
Incorporating computer-mediated technology to
strategically serve pedagogy. Communication
Teacher, 15, 1-4.
Phillips, G.M., & Santoro, G.M. (1989). Teaching
group communication via computer-mediated
communication. Communication Education, 38,
151-161.
Robson, J. (2000). Evaluating on-line teaching.
Open Learning, 15(2), 151-172.
Rubens, W., Emans, B., Leinonen, T., Skarmeta,
& A. G., Simons, R. (2005). Design

of Web-based
collaborative learning environments. Translating
the pedagogical principles to human computer-interface. Computers & Education, 45, 276-294.
Salaberry, M. R. (2000). Pedagogical design
of computer-mediated Communication Tasks:
Learning objectives and technological capabilities. The Modern Language Journal, 84, 28-37.
Snoeyink, R., & Ertmer, P. (2001-2002). Thrust
into technology. Journal of Education Technology
Systems, 30, 85-110.
Solomon, C. M. (2001). Managing virtual teams.
Workforce, 80(6), 60-65.
Sproull, L., & Kiesler, S. (1986). Reducing social
context cues: Electronic mail in organizational
communication. Management Science, 32, 14921512.

Human Computer Interaction and the Best Mix of Face-to-Face and E-Interactions in Educational Settings

Twigg, C. A. (1997, March/April). Is technology

a silver bullet? Educom Review,28-29.
Uys, P. (1997). Managing a hypermedia courseware development project: Dynamics and conclusions. Paper presented at the International
Council on Distance Education Conference,
Pennsylvania State University.
Vaughan, K., & MacVicar, A. (2004). Employees
pre implementation attitudes and perceptions
to e-learning: A banking case study analysis.
Journal of European Industrial Training, 28(5),
400-413.
Wegerif, R. (1998). The social dimension of
asynchronous learning networks. Journal of
Asynchronous Learning Networks, 2 (1) Retrieved
from Available:http://www.aln.org/alnWeb/journal/vol2_issue1/wegerif.htm
Wilson, T., & Whitelock, D. (1998). What are the
perceived benefits of participating in a computermediated communication (CMC) environment
for distance-learning computer science students?
Computers & Education, 30, 259-269.
Witmer, D. F. (1998). Introduction to computermediated communication: A master syllabus for
teaching communication technology. Communication Education, 47, 162-173.
Yakimovicz, A. D., & Murphy, K. L. (1995). Constructivism and collaboration on the Internet: Case
study of a graduate class experience. Computers
and Education, 24(3), 203-209.

key terms
Computer-Mediated Communication
(CMC): Computer-mediated communication
involves communication interactions that exist
over computer networks.
Critical Thinking: Involves a mental process
of analyzing or evaluating information in an
attempt to attain higher level of understanding.
person or learner is what has been learnt.
E-Learning: involves the process of knowledge dissemination and acquisition taken place
over electronic networks
Electronic Toolkit: represents wizard-like
database that can be invoke in electronic communication by both learners and instructors to
provide general but customized feedback.
Human Computer Interaction (HCI): Involves the study of interaction between people
and computers where the interface of computer
and people enables goal accomplishment.
Globalization: Involves economic and sociocultural ideas where organizations are able transcend national geographic and cultural boundaries
through convergence of space and time in attempt
to accomplish goals.
Transformational Learning: Involves knowledge acquisition that individuals and students can
adapt to transform old knowledge about a given
idea or topic. In other words it represents a process
of getting beyond factual knowledge alone.

This work was previously published in Handbook of Research on Computer Mediated Communication, edited by S. Kelsey
and K. St.Amant, pp. 49-61, copyright 2008 by Information Science Reference, formerly known as Idea Group Reference (an
imprint of IGI Global).

1435

1436

Chapter 4.2

Human Computer Interaction

for Computer-Based
Classroom Teaching
Wolfgang Hrst
Albert-Ludwigs-Universitt Freiburg, Germany
Khaireel A. Mohamed
Albert-Ludwigs-Universitt Freiburg, German

Abstract
This chapter focuses on HCI aspects to overcome
problems arising from technologies and applications that may hinder the normal teaching process
in ICT-ready classrooms. It investigates different
input devices on their usage and interactivity for
classroom teaching and argues that pen-based
computing is the mode of choice for lecturing in
modern lecture halls. It also discusses the software
design of the interface where digital ink, as a first
class data type is used to communicate visual
contents and interact with the ICT.

Introduction
Utilizing information and communication
technology (ICT) in modern classrooms for the

purpose of teaching offers several advantages.

These include (worldwide) access to digital materials during lectures, recording and archiving of
both the presented materials as well as the whole
live event, and transmitting the lecture (live or
recorded) to remote locationsa concept of teleteaching. Consequently, conventional equipment,
such as blackboards and chalk as well as overhead
projectors and transparencies, have almost disappeared from the classrooms and lecture halls at
many institutions and have been replaced with
computers and data projectors. Integrating ICT
in the classroom raises a myriad of technical
questions related to the implementation of the
respective devices and services. It is generally
agreed, that details of the technical installation
should be hidden from the presenters and teachers. Thus, in an ideal scenario, the teacher walks
into the lecture room and, if anything, just has to

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Human Computer Interaction for Computer-Based Classroom Teaching

press one button in order to switch on the whole

equipment and having all the infrastructure readily available. In this chapter, we are concentrating
on the users perspective once all equipment is up
and running and operated by the teacher during
the lecture. In doing so, we focus on questions
related to human computer interaction (HCI):
How can people access the functionality offered
by the installed ICT? How do they interact with
the available infrastructure? What are the best possible approaches for the interface design in order
to support the teaching process? And so on.
Generally, the way in which people interact
with any kind of ICT is influenced by two aspects.
The first aspect is the input devices, that is, the
hardware that specifies what information is exchanged between man and machine and how this
exchange takes place. For example, by moving a
mouse, a user sends events to a computer that
encodes information such as relative coordinates
representing mouse movements. The second
aspect is the software design of the user interface and the way in which the technical signals
exchanged by the hardware are interpreted. For
example, incoming signals from mouse movements (representing relative coordinates) are
generally mapped to absolute pointer movements
on the screen and as such visualized to the user.
Unfortunately, traditional HCI concepts, such
as conventional desktop input devices (the
keyboard and mouse) and their related classical interface designs are not well suited for the
tasks appearing in a classroom, forcing teachers
to modify and adapt their teaching styles to the
used equipment.
Following the point of view described before,
this chapter is organized into two parts. First, we
address input devices, that is, hardware-related
issues for operating and using ICT in classrooms.
We describe current and future trends in the
usage of different devices using examples and
case studies from our past and ongoing research
projects as well as our experiences in actual teaching scenarios. We identify a subtle, yet promis-

ing trend that can be observed developing in the

educational worldmore and more pen-based
input devices are used in classrooms and lecture
halls. Examples of such devices include interactive
electronic whiteboards, graphics tablets, Tablet
PCs, and personal digital assistants. The second
part of this chapter is devoted to the software
designs and interaction paradigms for interactive
classroom teaching using ICT. We describe different aspects in relation to pen-based interaction
with the installed ICT. Digital ink is described
as a new, first-class data type serving both, an
exchange of information between teacher and
students, as well as a new metaphor for interacting with the installed ICT via newly introduced
gesture commands. Generally, we concentrate
on instructional talks and presentations, where
a presenter (i.e., the teacher) faces an audience
(i.e., the students) and presents some content
in a typical classroom setting. More interactive
classroom settings, such as seminars or other arrangements involving group work, are discussed
in a concluding future trends section at the end
of this chapter.

Hardware-Related Issues:
Input Devices
Background: Ubiquitous Computing
Environments
As mentioned in the introduction, ICT usage in
classrooms offers great opportunities to support
and improve teaching. However, traditional input
and output devices often prove to be a hindrance
for the direct interaction between teacher and
students. They draw too much of the lecturers attention to the operation of the respective hardware.
Keyboard and mouse cannot be used to provide
content to the students as naturally as chalk is
used to write on blackboards. Monitors, even
the small ones from laptop computers, can stand
between the audience and the teacher building a

1437

Human Computer Interaction for Computer-Based Classroom Teaching

virtual barrier, and so on.

In a perfect scenario, all ICT equipment should
be seamlessly integrated into the environment,
that is, the classroom, and not interfere with the
activity of teaching but support the lecturer in
the best possible way. Presenters must be able
to naturally access and handle the additional
functionality provided by the respective devices
alongside in a way that enables them to fully concentrate on their teaching and the interaction with
the audience. Hence, the appropriate usage of ICT
in classrooms is an ideal example for a concept of
computing introduced by Marc Weiser in 1991 as
ubiquitous computing (Weiser, 1991). Generally,
the term ubiquitous computing refers to an environment in which a large number of computers,
ranging from hundreds of palm-sized to a few
wall-sized units will be seamlessly integrated
into our immediate surroundings, connected by
wireless networks (MacIntyre & Feiner, 1996,
p. 251). While originally being applied mainly in
the academic world, we now have the technology
available to make Weisers vision of a ubiquitous
computing environment a reality by building a
classroom where all computing functionality is
integrated into the environment. Many universities are already equipped with wireless networks.
Large touch sensitive displays for integration into
classrooms are available at prices affordable by
most universities, and so forth.
However, to really create an intelligent
classroom that does not constrain teachers but
fully supports them, some open problems and
challenges still remain, especially related to HCI.
In the following, we give a historical overview
about the usage of computers and different input
devices in the classroom over the last couple of
years. We discuss our experiences gained from
utilizing such devices at our university and from
observing other project- and user-groups. The aims
of this overview are to highlight the advantages
and disadvantages of the respective devices, identify interaction and usage styles, provide advice
for the appropriate usage of the technology, and

1438

pinpoint open problems for further HCI-related

research. Pen-based interfaces are identified as a
trend-setting and essential aspect for ICT-based
classroom teaching. Afterward, two case studies
of innovative usage of pen-based input-devices
are presented, which give direction to the future
usage of pen-based interaction in an interactive
classroom.

Input Devices for Computer Usage in

Classrooms: An Historical Overview
Traditional teaching: blackboards, overhead
projectors, and freehand writings. Teaching a
class involves a lot of interaction between teacher
and students, through the freehand writings on
blackboards or chalkboards. Blackboards have
traditionally been by far the most widely used
medium for lecture presentations in institutes of
higher learning, and they are very effective. Not
only do they allow instructors to communicate all
or part of the course materials in natural handwriting, but they also take away the attention to details
for their use. Freehand writings and annotations
take an important role in classroom teaching, even
if a lesson is delivered via previously prepared
slides or transparencies that are presented to the
students using an overhead projector. Essentially,
as classes progress, teachers will write, highlight,
underline, draw, and formulate on those slides to
put forth their points to the students. By using
empty slides, they can spontaneously write
on them, what actually facilitates discussions
amongst students, helps to illustrate complex
mathematical proofs, and demonstrates diagrammatic facts more clearly. Studies show that the
combination of handwritten annotations, the development of thoughts through freehand writings,
and the verbal explanations of the lecturers, are
indeed crucial in assisting students understanding
of the subjects (Mukhopadhyay & Smith, 1999;
Zupancic & Horz, 2002).
Early computer usage: data projectors and
presentation software. With the rising popularity

Human Computer Interaction for Computer-Based Classroom Teaching

of presentation software tools, such as Microsofts

PowerPoint, in the mid and late 1990s, a sizable
number of teachers moved from using chalk and
blackboards or overhead transparencies to using
electronic slides and computers for classroom
teaching. The increase of processing power and
the decrease in hardware pricing led to the augmentation of lecture halls and classrooms with
data projectors connected to background computer systems. By observing how teachers took
to using these new tools, three different trends
can be discovered:
1. Due to the lack of a flexible and effective
input device such as the pen and partly due
to a less than perfect interface design, some
lecturers adapted from their very lively
teaching style, to a scenario where the materials presented to the students were prepared
electronically in advance and shown in a
pure slide show style presentation. Hence
our observation that whenever computers
are exploited in classrooms, the teaching
styles of the instructors are often made to
adapt to the currently available interfaces
and not vice versa.
2. Some teachers have tried to revert back to
their old, traditional approach of slowly developing a topic by continuously adding and
annotating their contents. This often results
in their sitting down behind a computer
screen and using traditional input devices,
such as the keyboard and mouse, to add and
modify the presented materials during the
lecture. Where instead of writing in freehand
and annotating on the slides, they type the
text with the keyboard and make annotations
by drawing straight lines and boxes with the
mouse. So, again, the teachers go off their
traditional way of teaching and adapting to
a new and less appropriate style of interaction. In addition, sitting down and hiding
behind a big computer monitor led to a loss
of contact with the students.

3. Chong and Sakauchi (2000) observed that

when presentation slides are available (either
as overhead projected slides or PowerPoint
slides from a data projector), the use of the
chalkboard is significantly reduced during
the course of the lecture. However, when it
comes to facilitating questions and answers,
the chalkboard is conveniently used again to
demonstrate facts, confirm equations, and
diagrammatically illustrate flowcharts. But
making a direct reference from the chalkboards to any of the slides can be a hassle,
and sometimes the instructors resort to annotating on those slides in whatever little
space is available, and may even compromise
on legibility.
In all three scenarios, we can observe that
forced by inappropriate user interfaces, teachers generally adapt their teaching style in a less
preferable way to the available technology. As for
the students, they normally appreciate the fact
of having an electronic version of the presented
material after the lecture. It enables them to pay
more attention to the class, and it gives them the
time to make more appropriate and intelligent
annotations during the live event because they do
not have to copy everything the presenter shows
them. However, pure slide shows are generally
considered too exhaustive. Students often complain that such presentations are too fast for their
liking, while the required speed for handwriting
naturally slows down the pace of the lecture to a
speed that is generally considered more appropriate and pleasant.
Initial usage of pen-based devices: Graphics tablets and touch screens. Considering the
negative observations described previously, it is
no surprise that with the advent of devices for
pen-based computing, presenters quickly started
to experiment with them in their classrooms and
lecture halls. People first started using touch
screens (where users communicate with the
computer by direct touching of the screen) and

1439

Human Computer Interaction for Computer-Based Classroom Teaching

digitizing or graphics tablets (where a digital

pen is moved over a separate surface in order to
remotely control and operate the computer, similar
to moving a mouse in order to operate the mouse
pointer on the screen, cf. Figure 1). These input
modalities improved the situation in so far as
teachers are at least able to write natural freehand
annotations. The devices, however, still lacked
the usability and simplicity of markers or chalk.
For example, in the case of graphics tablets, not
every teacher feels comfortable with the remote
control of a digital pen, but prefers to write directly onto the visualized electronic slide. Touch
screens are hard to write on when the monitor is
standing upright. Ideally, the screen should be
mounted horizontally. However, touch screens
do not only react to pen input but to any kind of
physical contacts, thus making it impossible for
teachers to rest their hand on the surface while
writing. A detailed discussion about the initial
usage of pen-based input devices in the classroom
can be found in [*].
Advanced pen-based input devices: Interactive LCD panels and electronic whiteboards. The
previously described input devices represented
an improvement over traditional ways for human
computer interaction such as keyboard and mouse.
However, it was only after large, blackboard-like

electronic whiteboards and LCD panels with integrated tablet became available at a reasonable
price when pen-based computing became a real
alternative to chalk and blackboard usage.
Originally targeted toward special interest
groups and professions (such as graphic designers), the first LCD displays with integrated tablet,
in the following called interactive LCD panels
(cf. Figure 2), became available for the average
consumer by the end of the 1990s. These devices
operate like regular graphics tablets described
before with the main difference being that the
tablet is transparent and mounted on top of an
LCD panel. Hence, compared to common touch
screens, the surface does not react to any kind
of contact, but only to the input of a special pen,
thus enabling presenters to naturally write on
a horizontally mounted screen. Using the pen,
people can directly interact with the applications
(e.g., by making annotations on the presented
slides, navigating through subsequent slides, etc.).
Teachers generally see this as a big improvement
compared to normal touch screens or graphics
tablets. However, early versions often proved to
be too small, had a resolution that was too low,
and a limited processing power that resulted in
a noticeable time delay during freehand writing.
Nowadays, we observe significant advancements

Figure 1. Examples of pen-based input via graphics tablets in different sizes

1440

Human Computer Interaction for Computer-Based Classroom Teaching

Figure 2. Examples of interactive LCD panels where the user can write directly on the computer screen
using a special digital pen. While original versions were rather small (top right), newer versions (left
and bottom right) are now available at reasonable sizes and prices.

in display size, screen resolution, as well as processing power, paired with lower prices. Hence,
when mounted horizontally such devices provide
the same ease of use as traditional overhead projectors while at the same time enabling users to
access and use the full functionality offered by
the underlying ICT installation.
An electronic whiteboard (cf. Figure 3) is
the equivalent of a chalkboard, but on a large,
wall-mounted, touch-sensitive digital screen that
is connected to a data projector and a computer
(see, for example, Smart Technologies Inc. [Online]). This computer can be controlled directly
from the digital screen with a digital pen or by
simply using a finger to touch it. Other known
terminologies of the electronic whiteboards include the eBoard, digital whiteboard, smart
board, or interactive whiteboard. They carry
slightly different meanings to different people
depending on their environment of application.
However, these terms all describe in general, the
group of technologies that are brought together to
support classroom activities. Different versions
exist that rely on front or rear projection. In the

case of front projection, there are also whiteboard solutions that do not react to any kind of
contact, but (similar to interactive LCD panels)
use a special pen for interaction. While such an
approach is preferable for LCD panels or any
kind of horizontally mounted input surface, it is
less important for vertically wall-mounted boards
since users generally do not rest their hand on the
board during writing.
Similar to early interactive LCD panels, first
versions of electronic whiteboards had limitations
regarding size, resolution, and processing power
and were often too expensive for large-scale usage. Again, the situation is improved significantly
today, and we can observe how more and more
classrooms are becoming equipped with such
devices. In fact, there is a growing trend amongst
learning institutions approving the integration
of electronic whiteboards into their classrooms.
There are also many reports in the emerging body
of literature suggesting that vendors of education
and teaching instructors alike find the electronic
whiteboards relatively easy and compelling to use
(BECTA, 2003; Glover & Miller, 2001). However,

1441

Human Computer Interaction for Computer-Based Classroom Teaching

Figure 3. Examples of electronic whiteboards which enable direct interaction and freehand writing to the
users (top right). Earlier versions were rather small and often used front projection (top left) while newer
versions increased in size and offer rear projection at reasonable prices (top middle and bottom).

the authors also caution that becoming confident

in their use often requires effort in commitment
in terms of both training and independent exploration.
The situation today. As described previously, we saw that pen-based input devices offer
promising perspectives for usage in classroom
teaching, but initial versions had limitations
and disadvantages. Due to advancements in
hardware (such as size, resolution, and processing power) and dropping prices we can now
say that electronic whiteboards and interactive
LCD panels have become a real alternative to
traditional blackboards and overhead projectors,
respectively. Both devices offer a similar ease of
use as their traditional counterparts while at the
same time opening various new opportunities by
providing full access to the underlying computer
systems and infrastructure. In addition, new
developments such as Tablet PCs (cf. Figure 4)
incorporate the pen-based input technology and

1442

are more and more used for giving presentations

and classroom lectures as well. Tablet PCs are a
special kind of computer in the size of a laptop
with a pen-operated graphics tablet integrated into
the screen, similar to the interactive LCD panels
described previously. Some Tablet PCs purely
rely on pen-based user interaction, while others,
called convertibles or hybrids, offer pen-based
operation in addition to a regular keyboard and
mouse or touch pad input.
Based on the feedback we received in various
situations at our university, we argue that the
equipment available today can finally be used
to replace blackboards and overhead projectors
through their electronic counterparts in a reasonable way. We observed that even people who are
rather skeptical about using computers in classrooms and still stood with the old, traditional form
of classroom teaching are more and more switching to computer supported presentations due to
these new input devices. Overall, we can conclude

Human Computer Interaction for Computer-Based Classroom Teaching

that pen-based computing will be an important

if not essential part of computer usage in future
classrooms. It seems likely that in the near future,
more and more classrooms will be equipped with
pen-based input devices, in a similar way as not
so long ago, blackboards and overhead projectors
were complemented and slowly replaced with
electronic data projectors.
In the following section, we describe two case
studies of actual equipment used at our university
for teaching. The purpose of this description is to
describe the respective technology and how it is
best used, to illustrate how people take advantage
of it, and to identify open problems and future
trends. We will see that with this technology
we are able to build real ubiquitous computing
environments as described before as the ultimate
goal in the Background section.

Case Studies of Actual Equipment

Used in a Ubiquitous Computing
Environment
Electronic whiteboards as blackboard replacement. As mentioned before, todays electronic
whiteboards are larger in size, offer a higher
resolution, are technically more reliable, and are
easier to handle. However, their absolute size still
remains an issue when compared to traditional
blackboards. As a consequence, some people
started placing two boards next to each other in
order to increase their writing space by forming
a large electronic whiteboard (Prante, Streitz,
& Tandler, 2004). Examples of such twofold
whiteboard installations at our university can be
found in Figure 5. The boards shown at the top and
bottom left of Figure 5 are integrated into the front
wall of the lecture room thus hiding the required
technology from the presenter and providing an
environment, where the computing functionality
can be directly accessed by the teacher through
interacting with the electronic whiteboards, but the

Figure 4. Examples of Tablet PCs which enable pen-based input via a transparent graphics tablet mounted
on top of the LCD screen. While some versions purely rely on pen-based input (top) others, called hybrids
(bottom), feature a rotating display thus supporting both, traditional as well as pen-based interaction.

1443

Human Computer Interaction for Computer-Based Classroom Teaching

Figure 5. Putting two electronic whiteboards next to each other, in order to form a larger input and
output area, resemble traditional blackboards

whole technology does not interfere with regular

teaching. All technologies are hidden as much as
possible in order to let the presenter concentrate
on the talk without distraction from any of the
hardware. In our installation, both boards can
be connected to one computer. People are able
to place one application window (e.g. from the
used presentation software) over two boards thus
creating one large virtual electronic whiteboard.
Experience with users has shown that the border
in the middle of these two boards is generally not
considered critical since most presenters prefer
to write on the board in columns anyway. While
connecting several input and output devices to
one computer was still considered critical a few
years ago (MacIntyre & Feiner, 1996), todays
windowing systems generally support this kind
of functionality without any problems.
When observing how people are using such
an installation, we realized different approaches.
Some presenters like to have just one application
window that extends over both whiteboards, either
using it to present two slides next to each other or

1444

to start with two empty slides and an increased

writing surface. Others prefer to have two separate
applications running, for example, in order to show
an overview slide of the lecture on the left board
while presenting the current slide on the right side.
Some teachers use one board for slides and the
other one to show a different application, such as
a player for video clips or animations. People can
also connect their own laptop to one of the boards,
which is used, for example, by some presenters to
show their slides from the laptop, while referring
to the other board (and the connected computer
in the classroom) when accessing and presenting
information from the Internet. Using the two
output devices for different applications is a common practice often chosen when using multiple
monitors for desktop computing as, for example,
Grudins studies pointed out; there, participants
used multiple monitors rather for partitioning
than to increase space (Grudin, 2001).
Overall, we can observe that people take great
advantage of the additional working space. No
single trend in using this installation can be ob-

Human Computer Interaction for Computer-Based Classroom Teaching

served, but a variety of different usage scenarios

exists. Hence, instead of having the teachers adapt
to a certain interface or style of interaction, this
setup enables them to choose among different
approaches and to use the respective tools in the
way they personally prefer.
Integrated interactive LCD panel as overhead
projector replacement. In addition to the boards,
we built a high desk or lectern with integrated
computing facilities that can serve as a true replacement of the traditional overhead projector
(cf. Figure 6). As input device, a horizontally
mounted LCD display with pen-based interaction
functionality is integrated into the surface of the
lectern. All other equipment needed for a talk is
embedded into the lectern, but again hidden to
the user as much as possible. This includes even
audio recording facilities if the presenter wants
to record the lecture with some lecture recording
software (Mller & Ottmann, 2000) and preserve
it for further, off-line usage. Hence, the normal
user just has to connect the lectern to the power
source and data projector and press a button to
turn it on. Including all required equipment into

the lectern provides several advantages, because

it enables us to make the whole device mobile
by putting it on wheels. Thus, it can be easily
maintained and replaced by technical staff if
necessary. In addition, it can be transferred to
different rooms, which is important in case not
all rooms can be provided with the full range of
technical equipment. It also allows one to easily
put it away and lock it, in case it is not possible
to always lock the room.
Considering the usage by the presenters, most
of them reported that handwriting and making
annotations is much easier, since they can do this
on a horizontally mounted surface placed in front
of them. Some also expressed their appreciation
that they can face their material and the audience
at the same time and do not have to continuously
turn their back to the students when presenting
their material. Meanwhile, several of these lecterns
were built and used in various lecture halls at our
as well as other institutions.
Both the lectern as well as the large interactive
whiteboard described before turned out to be reasonable replacements for traditional devices used

Figure 6. Lectern with integrated computing facilities and pen-based input device. The interactive LCD
panel is integrated into the surface in order to support freehand writing on the presented slides and
pen-based operation of the used applications.

1445

Human Computer Interaction for Computer-Based Classroom Teaching

in classrooms, that is, blackboards and overhead

projectors. If integrated into the environment as
shown, for example, in Figure 5 and 6, they do
not interfere with the normal teaching process of
the presenter while at the same time offering full
access to the computing functionality offered by
the underlying computing system and network
infrastructure.

Software-Related Issues:
Interaction Paradigm
Background: Digital Ink
In the previous section we argued that the pen is the
natural choice for interaction with wall-mounted
boards and interactive tablets. It enables users to
interact directly with the respective tools instead
of remotely as with a mouse pointer. Freehand
writing proves to be more intuitive and flexible
to enter content during a lecture than a keyboard.
Hence, handwriting becomes a first-class data
type, on an equal footing with text entered via
keyboards. However, common interface designs
and graphical user interfaces (GUIs) usually
follow the desktop metaphor and are therefore
optimized for input devices such as keyboard and
mouse. For example, navigation through nested
menus can easily be done with a mouse, but is
hard to do with a pen. For pen-based interaction,
earlier research on pie menus proved an advantage for users of digital screens, as people tend
to better remember the cyclic position, and as
such expedites the selection process (Callahan,
Hopkins, Weiser, & Shneiderman, 1988). Hence,
we see ourselves confronted with two different
demands: on the one hand, the aim of using penbased input devices in order to present information to the students, while on the other hand, the
need to rely on traditional, mouse-based input in
order to interact with the ICT and to operate the
respective programs. To avoid the unfavorable
switching between different devices, we propose

1446

a new interaction mode purely based on pen input

and the notion of digital ink described in the following sections.
Digital Ink. Motivated by the treatment of
handwriting as a first-class data type, we introduce the notion of digital ink, which is used
not only to enter and represent contents, but
also to operate the ICT via gesture commands.
A gesture command is the resultant process of
invoking gesture-like movements with the pen,
that resemble special shapes drawn in a certain
distinctive way, recognized by the program as
certain predefined instructions for it to carry
out. The digital ink derives its name from the
data class that represents any kind of information
created when using a digital pen to draw strokes
on a screen of a pen-based input device. While
the strokes received on the pen-computers surface get recorded as digital ink, its consequential
markings are reflected noticeably onto the screen
simultaneously. Any digital ink that was written
is either kept in its raw form or goes through a
process that translates it into recognizable texts
or graphic objects. Often, in many pen-based
applications, the ink goes further to be tried as a
possible pen-gesture. In this case, when the ink
is confirmed to be a gesture command, its visual
trace on the screen is removed upon the execution
of the respective gesture command.
Digital ink in its raw representation as processed by the respective hardware and software
holds far greater amounts of information than
are immediately apparent. These include, among
others, the type of pen tip, the amount of pressure used to create a stroke, the height of the pen
above the screen surface, opacity value, color,
and timing data. As a result, the digital ink is
acknowledged in serious enterprises such as one
for secure handwritten signatures (Cyber SIGN,
Inc., 2005; Gries, 2000; Hangai, Yamanaka, &
Hamamoto, 2000), as well as in everyday norms
of handwriting e-mail messages (Pen&Internet
LLC, 2005), sketching design posters (CorelTM,
2005), and annotating digital documents (GRAHL

Human Computer Interaction for Computer-Based Classroom Teaching

Software Design, 2005; Mller & Ottmann, 2000).

As more electronic devices with pen interfaces
continue to become available for entering and
manipulating information, efforts have been
made to ensure that supporting pen-applications
are effective at leveraging this method of input.
Handwriting is an input modality that is very
familiar for most people since everyone learns
to write in school. So, there is a high tendency
for people to use this mode of input and control
not only in the classroom scenario described
previously, but for a great variety of different
applications as well.
Ink traces and symbolic representation. The
domain of the digital ink is unlike any others,
such as text and graphics, that can have all its
related information easily organized and crossreferenced, and be presented in front of users
to allow them direct manipulation of the data.
A trace refers to a trail of digital ink data made
between a successive pair of pen-down and penup events representing a sequence of contiguous
ink pointsthe X and Y coordinates of the pens
position. If the sampling property of the input
device is not constant, it becomes advantageous
to include timestamps for each pair of the sampled
coordinates. A sequence of traces symbolically accumulates to meaningful graphics, forming what
we humans perceive as characters, words, drawings, or gestures representing commands. Each
trace can be categorized in terms of the timings
noted for its duration, lead, and lag times.
The importance of digital ink has also been
recognized by the Internet community, leading to the specification of the World Wide Web
Consortiums Ink-Markup-Language (InkML)
standard (Russel, Chee, Seni, Yaeger, Tremblay,
Franke, Madhvanath, & Froumentin, 2004).
InkML allows the description of precise digital
ink information that is not only interchangeable
between systems, but also allows the convenient
archival and streaming of ink data between
applications. Among many of its properties, it is
important to note that:

InkML provides a common format for the

exchange of ink data between components
such as handwriting and gesture recognizers, signature verifiers, and ink-aware
modules;
InkML supports the representation of handdrawn ink with full attributed information,
on top of capturing the pen positions over
time;
InkML allows the specification of extra
information for additional accuracy such
as pen tilt and pen tip force (pressure) for
the enhanced support of applications such
as character and handwriting-style recognition, and authentication;
InkML provides the means for extension
by allowing application-specific information
to be added to the ink files to support issues
of compatibility.

Active digital ink in recordings and replay. One

of the advantages of the usage of ICT in classrooms
is the possibility to record and archive the live event
for further usage and studying. Today, numerous
systems exist for automatic lecture recording that
automatically turn the live event into a multimedia
document. These files are used, for example, by
students to reflect a lecture. Others use them to
deepen their understanding of particular topics.
They can even be used as the core for building
Web-based online courses.
Independent of the respective application scenario, the studies mentioned previously (Mukhopadhyay & Smith, 1999; Zupancic & Horz, 2002)
highlight the importance of the development of
thoughts through freehand writings. They also
confirm the evidence that simply capturing the
static end-result of the lecturers annotations for
recalling the subject at a later time is inferior to
unambiguously maintaining the whole sequence
and dynamics of those freehand annotations when
recorded for interactive replay.
As a consequence, user interfaces that allow
replaying of real-time multimedia documents

1447

Human Computer Interaction for Computer-Based Classroom Teaching

usually include command buttons to quickly

fast-forward and rewind their data sets for
systematic browsing. We see this very often in
many learning applications that try to cater to
both students and teachers in assisting them to
effectively retrieve desired segments of those
documents. Random access navigation, that is, the
ability to access any position along the time-line
by using, for example, a slider interface, provides
an additional advantage on top of this simple fastforward and rewinding of real-time information.
It gives users the ability to straight away focus on
the intended portions of their documents without
having to wait a second longer for these processes
to arrive at the stipulated point of time, albeit the
processes being a little faster than the normal
replay mechanism. Hence, users should be able
to navigate through digital ink recordings along
the timeline. However, realizing such a random
visible scrolling is a challenging task.
Summary. In this section, we introduced digital
ink as a notion to describe pen-based input, which
is used not only to model chalk or markers in the
digital world, but also to interpret particular ink
traces, which are the building blocks of digital
ink, as gesture commands. In such a setup, we
need sophisticated techniques, not only to represent, process, and manipulate traces of ink, but
also to store, access, and replay them. The ability
for random access navigation poses a particular
challenge in this context.
In the following section, we first describe
aspects about the personalization, archival,
and streaming of digital ink. Subsequently, we
introduce a new paradigm for interaction with
electronic whiteboards via ink traces that are
interpreted as gesture commands.

Personalizing Digital Ink for Archival

and Streaming
In addressing the proper support interfaces for
digital ink mediums that are to be stored and
indexed in accessible archives or streamed in

1448

broadcast networks, appropriate tools are required

for practical personal manipulation of the digital
ink within multimedia and multivalent materials (Phelps & Wilensky, 1997). These tools are
for the producing, capturing, and replaying of
freehand writings, and should therefore contain
features deemed to let users feel that their digital
handwritings are acceptable substitutes for their
normal ones when applied on digital documents.
We have identified six main feature-themes that
relate to what users perceive in the contents of the
formulated digital strokes and what they would
expect when interfacing with them. These featurethemes refer to the personal flavor, manipulation
in pixels, manipulation by objects, compactness
of representation and scalability, replay and synchronization, and facilities for random-access.
1. Personal flavor: Every individual person
has a unique and particular writing style.
This is portrayed in the calligraphic output
characteristics from their written works.
A handwriting tool needs to allow the
expression of this style, as accurately as
possible. The recording samples and the
replay stream of this tool should maintain
all these personal calligraphic attributes.
Various pen tip styles, such as coarse whiteboard marker-pens or fine pencil leads, and
pressure-sensitive information should be
supported during both recording (creation)
and replay. Note that most currently available
digital ink software offered in presentation
tools is either nothing more than fixed-size
pixel-traces on screen, or does not support
the capture and replay of traces in a manner
that maintains the dynamics of the generation process. That being mentioned, users
should also expect simple anti-aliasing
techniques on the interface tool to improve
the stroke appearance of the traces rendered
on screen (Klassen, 1991).
2. Manipulation in pixels: Removing traces,
or parts of it, and inserting empty spaces

Human Computer Interaction for Computer-Based Classroom Teaching

on the grid-like digital screens are typical

examples of pixel-oriented operations.
3. Manipulation in objects: Generating or
selecting whole objects, stretching, contracting, reflecting, pivoting, copying, and
pasting are typical examples of object-oriented operations applied to trace-objects.
Arguably, the typical editing commands in
modern graphics tools (insert, copy, paste,
and the like), however, may also be considered as pixel-based operations.
4. Compactness of representation and scalability: Extensive writing ultimately generates large volumes of data if the complete
transcript of the raw digital ink is stored in
an uncompressed form. Inevitably, eliminating any superfluous information will be
a reasonable course of action. One way to
ensure that the integrity of the data remains
intact would be to try and symbolically
represent the handwritten letters/alphabets
as spline-curves that are independent of
the screen resolution. That is, scaling such
spline-curve-writings up or down would
preserve and not deteriorate the rendering
quality of the letters/alphabets during replay
of the recorded traces.
5. Replay and synchronization: At this point,
it should now be possible to replay all recorded handwritings in their best quality,
the same way as they were first recorded/
presented. Here, both the inter- and the intrastream time constraints should be observed,
such that the full dynamics of the generation
process of handwritings are maintained in
exactly the same way as they appeared at
generation time. Furthermore, the interface
must allow all (time-stamped) data streams
to be synchronized with other recorded
data streams, particularly with the audio
stream of the lecturers voice. The latter is
usually considered to be the master stream
in a master-slave synchronization model for
replay (Mller & Ottmann, 2000), whereas

the recorded ink traces are considered as the

slave stream that is synchronously replayed
with the master stream.
6. Facilities for random-access: Easy navigation, particularly random visible scrolling
(cf. previous) through a recorded document,
requires random real-time access to the
entire data collection. Metaphorically, this
ideology resembles the thumb-flipping of a
book. Given an instance of time, the complete
list of objects visible at this time instance
must be made available for display almost
immediately. That is, it should be possible
to compute and display all of these objects
fast enough, so that the user gets the impression of immediacy, as opposed to making
the user wait due to latency issues.
Table 1 summarizes the following discussion
about how current enterprise tools fulfil these
six requirements.
The Web tool riteMail (Pen&Internet LLC,
2005) allows for handwritten messages in an
interactive, electronic ink e-mail application.
Users can create high resolution, fully interactive
handwritten notes and drawings on hand-held
devices, Tablet PCs, as well as on every Java-enabled desktop or laptop computer. The notes can
then be sent to any e-mail address. The riteMail
software employs unified Extensible Markup
Language (XML)-based protocols for digital
ink transmission between all components of the
service. However, a replay facility is not included
within (and, in order to be fair, not essential for
its purpose).
Corel Grafigo (CorelTM, 2005) targets
pen-computers and is designed specially for
the Windows XP Tablet PC Edition. It supports
pressure-sensitive pens, allows the marking
up of documents and images individually, and
provides the features of shape recognition, and
gesture-based commands. Data are saved in SVG
format.

1449

Human Computer Interaction for Computer-Based Classroom Teaching

Table 1. Summarized assessment of various tools on freehand writings

Criteria

Writing
style

Support of
pixel-based operations

Object-oriented
mani-pulation

Symbolic
representation

Recording

Replay

Random
access

riteMail

yes

yes

yes

yes

static

no

no

Corel
Grafigo

yes

yes

yes

yes

static

no

no

Painter
Classic

yes

no

yes

no

static

no

no

Tablet PC
Software
Development
Kit

yes

yes

yes

yes

static

no

no

Lecturnity

no

yes

yes

no

dynamic

yes

yes

Software

Painter Classic, marketed by Corel (Wacom,

2005), is an excellent paint program for graphic
designers that simulates the various graphic
markers and coloured pencils and pens, as well
as water color and oil paints. As an image-editing
program, it offers numerous high performance
functions (edged and smooth drawing, distorting,
cloning, etc.).
The AOF-tools and their commercial spin-off,
Lecturnity (imc AG, 2003), fulfill many of the
interface requirements discussed in the previous
section, except that they do not fully preserve and
regurgitate their users personal style of writing.
Moreover, the data volume on recorded documents
with extensive handwritings is intolerable. The
reason is its highly redundant storage method;
for while Lecturnity offers a real-time random
access facility during replay, it also computes
the entire list of visible objects on the screen for
every single time instance.

Using Digital Ink on Electronic

Whiteboards
In the preceding discussion about input devices
we already described different usage scenarios of
electronic whiteboards in a ubiquitous computing classroom. Now we take a closer look at the

1450

actual applications running on those devices,

how people are operating them, and what kind
of interaction takes place.
Instructors and students in class do not only
see the prepared course materials projected onto
the electronic whiteboards, but also the active
annotations made by the instructors using the
digital pens as the lessons progress. Combined
with dedicated software applications, the supporting whiteboard technologies can be made to
function more effectively with better pen control,
and appropriately displaying both static and moving images at suitable resolutions. Among other
things, these applications generally offer:

Flipchart capabilities to save and print

images as well as annotated ink-trace markings;
A blackboard to manipulate primitive
handwritten text and drawings;
Integrated slides and blackboard environment;
Movie screens to showcase video and
software demonstrations;
Tools to facilitate discussions and group
work; and
Access to the network, where shared teaching resources may be available.

Human Computer Interaction for Computer-Based Classroom Teaching

More importantly, the way the electronic

whiteboards are utilized will shape the kind of
impact they will have on the overall learning
environment. The basic guidelines highlighted
by Glover and Miller (2001) for good practices
and outcomes of engaging classroom interaction
and participation with the electronic whiteboards
across the curriculum are:

To increase efficiency by enabling teachers to

draw upon a variety of ICT-based resources
without disruption or loss of space;
To extend learning by using more engaging
materials to explain concepts; and
To transform learning by creating new learning styles stimulated by interaction with the
electronic whiteboards.

All of the previously mentioned technological

and social benefits have positive indications from
the wider observational point of view. Nevertheless, there are also subtle cases that can make
instructors hesitant to change their pedagogy to
incorporate the electronic whiteboards into their
lesson plans, if practical considerations hinder
their perspectives. We already pointed out that
it requires effort in commitment to develop media-enhanced teaching contents for the electronic
whiteboards, which could be considerably taxing
on the teachers initial workload. Also, according
to the studies compiled by BECTA (2003), while
the act of teaching with electronic whiteboards
instills anxiety in the students to learn, it unduly
places the pressure onto the teachers when it
comes to delivering the contents. This is especially
true if the writing environment of the electronic
whiteboards is not sufficiently supported with the
necessary interfacing tools. We have thoroughly
discussed this in the preceding sections.
Good electronic whiteboard contents are generally those that can display themselves immediately in highly visual terms, which can be most
effectively understood by a group of individuals
(BECTA, 2003; Clyde, 2004). The ability to eas-

ily modify and change these content materials

is all about the personalization of the electronic
whiteboard environment. In relation to this, and in
order to achieve good content presentation, there
needs to be good software applications to manage and control the content data with enhanced
features and functions. Currently, makers of the
electronic whiteboards ship out accompanying
software that is embedded into their systems for
use in classroom environments. Examples include
those that are listed on Smart Technologies Inc.s
Web site such as notebook whiteboarding software
and SMART Board tools. These intuitive programs
support numerous operations that are categorized
into many menu items under several options that
teachers can use while delivering their lessons.
So much so, that they tend to become more of a
distraction to the teachers rather than an assistance
to them in front of a live audience.
Writing and feeling in control of the board
environment. Now, we are particularly interested
in the part where the previous support interfacing
software allows teachers to write on the electronic whiteboard freely as if it were a normal
blackboard. The situation described previously
may not necessarily pose as a problem if the
screen-size of the electronic whiteboard is small.
However, when these programs are maximized
to full screen, then teachers will bump into the
problem of hand-and-eye coordination. This is
a cross between uncomfortable peripheral views
for the eyes, and awkward bodily positions to access the menus without disturbing the flow of the
lesson, or blocking the audience. This problem for
a single large screen worsens when we cascade
the electronic boards in the learning environment
in series next to each other as described in one of
the previous case studies (cf. Figure 5).
Without investing in additional tracking
devices such as motion-detectors or cameras,
the large interconnected, high-resolution, wallmounted, multiple displays are by themselves
receptively sensitive to the teachers touch.
Properly configured, this electronic whiteboard

1451

Human Computer Interaction for Computer-Based Classroom Teaching

environment offers an extended desktop metaphor

quite unlike the ones that we presently know
and have grown used to. Further incorporating
the pen-input technology into this setup accords
teachers a higher level of personal presentiment
than a conventional monitor, in a way that it allows them to directly touch the active-screens
and see immediate reactions through the digital
pens. The electronic whiteboards are meant to be
an active presentation medium for a congregation
of audience in a large room. While there is always
a desktop console that controls the electronic
whiteboards, through the mouse and keyboard,
freeing up this mode of interfacing lets teachers
communicate directly with the screen using just
a digital pen.
Judging by the dominant metaphor that what
we have today is a mismatch for the computer
environment we are dealing with tomorrow, it is
difficult to place this huge screen scenario (alone)
into any pronounced categories of user interface
debates on concepts of adaptivity and adaptability
(Shneiderman & Maes, 1997). For instance, the
sphere of influence of the hand-and-eye coordination needs to be enlarged, and perhaps include the
body, to follow the claims of direct manipulation
of adaptable interfaces. Furthermore, while the
audience have it easy watching from a distance,
the teacher does not: standing so close to one
part of the multiple huge screens often leads to
the interface widgets he or she may require to be
out of reach, or worse still, because the teacher
cannot see those widgets, he or she may assume
that such actions or commands represented by
those widgets do not exist within the board application. We point out here that this may affect
the flow of the delivery of the lesson contents. In
this case, we may be left to rely on the adaptivity
of the interface for the electronic whiteboards to
proactively assist the instructor while at the same
time ensuring that they still feel in control of the
whole process.

1452

Interacting with Electronic

Whiteboards Through Gestures
In order to solve the problems previously described, we now describe an interaction concept
that realizes a modeless interactivity by deriving
gestures from digital ink traces. We define modeless interactivity for the digital ink environment on
the electronic whiteboards as the seamless integration of both the writing and gesture-command
modes on a common, unified platform. On top
of this, they further encompass the conventional
on-screen widgets such as the fixed pull-down
and pop-up menus as interaction modes that
can be minimized on their usage or done away
with completely when not needed in the most
cumbersome of circumstances. They go by the
notion that instructors should not always need
to go between different sections of the boards in
order to effect a menu action.
Deriving gestures from digital ink traces. It
is possible to convincingly anticipate, based on
the lead and lag times obtained, that the latest
ink component is more of an instance of a trace,
rather than a gesture. Advanced recognition
techniques have been introduced to the digital
ink domain, and while a number of works have
gone into handwriting and sketch recognition
(Rubin, 1991), there are among these works that
use the same techniques to attempt to recognize
hand-drawn pen gestures.
Hence, the writing mode as well as the gesture mode can help out with the current interface
issues. Gesturing to the system refers to the
actual gesticulated actions resulting from the pen
movements that represent a command, and not
merely the tap-and-drag-the-icon movements.
The prospects of gesturing commands in ink-domain applications are encouraging. So much so
that many authors think that this is the epitome
of the final interface for pen-based applications.
Although it is agreeable to a certain extent, that
this mode of ubiquitous interfacing may form the
fundamentals of future (more enhanced) hardware

Human Computer Interaction for Computer-Based Classroom Teaching

devices, we should not anticipate this to be the

trend in this current context. The perceptual user
interface (PUI) guidelines should be adopted
and used to create the first steps of the bridging
between fully packed screens of user interface
widgets to that of a ubiquitous one (Shneiderman
& Maes, 1997).
Deriving interfaces from gestures. For the
same technique that a right-mouse-click can
bring up a popup menu on documents that contain
menu-items generally related to the application
it serves, pen gestures have the relative capability
to have this metaphor extended. They make it
possible to provide more specific menu-commands that are currently in context. By this, we
mean the placing of directed (or pointedly explicit)
context-specific menu-items that should obviously take precedence on the front-most tier of the
multi-cascading popup interface by virtue of what
is currently being displayed on screen. The rest
of the more general menu-items, however, are
neatly tucked away in the lower-priority cascades
to ensure that they are still accessible, as is required
by the specifications of perceptual user interfaces.
For example, right-clicking anywhere on a Word
document (or most Windows applications for that
matter) brings up menu-items generally affiliated
to the Edit-menu on the applications main menubar. The advantage here is that users do not have
to skit their mouse over a distance to access the
main menus. This same advantage is magnified
many times over when working in front of a series
of large wall-mounted, hi-res digital screens where
the pen is the primary input modal.
The extension we mentioned previously for
this technique will involve additional information
we can gather from the background environment,
and is designed mainly for the convenience of the
wall-mounted SMART electronic board users. We
discussed this thoroughly in our previous work that
deals with digital whiteboard-applications for classroom scenarios, where the sole objective is to assist
lecturers in delivering their lessons comfortably
and not worry about the support technology.

We incorporated an active program module

called the Context-Assembler, running in the
background of the main program application,
that keeps track of all items that are visible on
the digital boards writing area. These include
all digital ink-entities that constitute drawings,
writings, illustrations, and so forth, and other
screen-related widgets like images, fonts, and
the writing canvas itself. These items are ordered and cross-referenced with each other in an
efficient way in the Context-Assembler to give
optimal background support information known
as context-information. That is, if we ask the
Context-Assembler about one particular trace that
we wrote earlier on the writing canvas, it will tell us
when and where the trace was created and located,
which other items are its immediate neighbors,
which are the other items it is overlapping with,
which group(s) of paragraph it is connected to,
the probability of the trace being a gesture, and
all other items that are similar to it in terms of
appearance, size, color, and style. Based on this
context-information that can be retrieved off any
items in the Context-Assembler, we designed a
context-oriented popup menu interface.
The mechanism that determines what menuitems receive higher priorities over others, in order
to appear on the front-most tier of the cascading
popup menu interface, is an application-defined
mapping function that uses the context-information as its input parameters. For example, in
Figure 7, the latest ink-element is queried for
its context-information. Among other things,
the Context-Assembler replies by telling that
the ink-element in question is on top of another
ink-element created at time (t-m), is on top of
an image known as slide p28, is on top of the
writing area known as p9, stretches horizontally
across these three previous itemsfrom left to
rightand has a very high probability of being
a right-arrow gesture. Putting these into the
mapping function returns a list of thematically
categorized menu-items that interprets the last
ink-element as a command gesture. The top few

1453

Human Computer Interaction for Computer-Based Classroom Teaching

Figure 7. The latest ink-element in (a) is interpreted as a gesture that can have several interpretations.
The current context of the board (writing area p9, slide p28, and ink@time(t-m)) is determined and
assembled appropriately in (b). The output of the mapping function is then rendered onto the screen as
support interface on-demand (c).

high-priority menu-items in the list will suggest

that the command gesture advance to the next
slide, or to the next page, or to next slide
and page, or to move the overlapped ink-elements identified in the context-information to a
new location.
To complete the visualization of the on-demand assistance, we bank-in on the last known
coordinate position of that ink-element in-context.
Popping up dynamic-content pie menus and/or
other static command buttons, with respect to
the list of menu-items returned by the mapping
function, in that vicinity seems most appropriate
and serves as a convenience to the user. As said
before, pie menus proved an advantage for users
of digital screens as earlier research pointed out
(Callahan et al., 1988). Our command button
(toolbar) interface is logged to the bottom of the
screens, appearing on-demand with only an array
of anticipated widgets. The combined lengths of

1454

the number of widgets will not exceed the arm

length of the users to ensure that all widgets are
within reach and within the users line of sight.

Conclusion and
Future Trends
Freehand writing is a natural way of exchanging viewpoints and communicating visual ideas
between teachers and students in a classroom.
Motivated by this, we identified the value of penbased computing for the usage of ICT in classroom
teaching. By presenting an historical overview of
the usage of pen-based input devices in classrooms
and lecture halls and two case studies with currently available technology, we demonstrated that
todays hardware equipment can be used to greatly
improve the teaching experience. Using them, we
are able to build a ubiquitous computing environ-

Human Computer Interaction for Computer-Based Classroom Teaching

ment in the classroom that offers teachers full

interaction with the respective technology using
a pen while at the same time not interfering with
the interaction between teacher and students.
As a trend of moving toward natural environments in the digital world, such a setup encourages
freehand writings to become the bridging tool that
encompasses portable and synchronizable standards of manipulating ink data between mediums
and documents. Hence, we introduced the digital
ink as a new first-class data type serving two
purposes, first, the representation and preservation of information through freehand writing
and annotations on the board. In this context
we discussed issues about the personalization of
digital ink as well as its archival, later access and
streaming replay. Subsequently, we described the
second purpose of digital inkthe operation and
control of the classrooms ICT via pen-gestures.
By interpreting ink traces as gestures, users can
interact with the ubiquitous computing environment in a more interactive and intuitive way than
with regular interface designs using traditional
graphical user interfaces developed and optimized
for desktop computing. This has also been proven
to be the case for multi-user interaction on the
horizontal table-interface: (cf. Figure 8) recognizing a gesture in its respective interpretation also
tells us the side of the table it originated from,
thus simultaneously identifying the user seated
at that corner without having to install additional
sensing devices.
For the future, we expect further improvements
in the respective hardware and the development
of new devices that might further improve the
usage of ICT in the classroom. Although current
devices and hardware configurations provide a
real alternative to the traditional blackboard and
overhead projector equipment, as we described
in the hardware-related part of this chapter, there
is always room for improvement in terms of size,
quality, and production costs. At the same time,
new technology keeps evolving. Interesting developments with great potential for classroom

usage include, for example, the so called electronic

paper/electronic ink approaches (see Jacobson,
Comiskey, Anderson, & Hasan, 2005 and Xerox
Parc, 2005).
Wireless networks are more and more installed
at universities, and although not directly related
to input devices, this infrastructure will have
great influence on the ICT usage in classrooms
and thus on the HCI. For example, it is not clear
how to seamlessly integrate different devices with
different sizes into the overall communication
and learning process. Lecturers write on large
electronic whiteboards, while students connect
mobile devices, such as laptops, Tablet PCs, or
even very small PDAs and SmartPhones, to the
wireless LAN. How to represent and exchange
information between such devices that differ in
size and processing power is an open research
problem (MacIntyre & Feiner, 1996) where convincing solutions are still missing.
In addition, there exists an obvious trend
toward computer-supported group work in classrooms. Instead of purely instructional presentations and lectures, which have been the focus
of this chapter, other ways of teaching, such as
seminars and group work, can benefit from the
usage of ICT as well. Important research in this
area has been done, for example, by the Interactive Workspaces Project at Stanford University
(Johanson, Fox, & Winograd, 2002). In our own
research lab, we are currently exploiting the usage
of an interactive table where a data projector is
mounted on the ceiling, projecting the content of
a computer screen onto a horizontally mounted
electronic whiteboard (cp. Figure 8). While the
respective hardware to build such a table is already
available, many open questions and problems remain regarding the design of the software side of
the user interface, such as dealing with different
orientations when people sit on different sides of
the table, enabling people to reach remote objects
on the virtual tables surface (i.e. the projected
computer screen), and so forth. Such horizontally
mounted interactive computing systems are an ac-

1455

Human Computer Interaction for Computer-Based Classroom Teaching

Figure 8. An electronic whiteboard mounted horizontally in order to create an interactive table for
group-work

tive research area, and we believe that classroom

teaching can greatly benefit from the outcomes
of the respective projects.

Note
Some of the notions for tools and devices described in this article are registered Trademarks
of the respective companies or organizations. We
kindly ask the reader to refer to the given references. All features and characteristics of these
systems have been described to the best of our
knowledge. However, we do like to mention that
specific characteristics and technical specification
frequently change, and therefore discrepancies
between the descriptions in this article and the
actual systems are possible.

References
BECTA. (2003). What the research says about
interactive whiteboards. Becta ICT Research.
Callahan, J., Hopkins, D., Weiser, M., & Shneiderman, B. (1988). An empirical comparison of pie
vs. linear menus. In Proceedings of the SIGCHI
Conference on Human Factors in Computing
Systems (pp. 95-100).

1456

Chong, N. S. T., & Sakauchi, M. (2000). Back

to the basics: A first class chalkboard and more.
In Proceedings of the 2000 ACM Symposium on
Applied Computing (pp. 131-136).
Clyde, L. A. (2004). Electronic whiteboards. In
Teacher Librarian Toolkit: Information Technology 32(2). Retrieved May 31, 2005, from http://
www.teacherlibrarian.com/ tltoolkit/info_tech/
info_tech_32_2.html
CorelTM. (2005). Corel grafigo 2Create, annotate, collaborate. Corel Corporation. Retrieved May 31, 2005, from http://www.corel.
com/grafigo
Cyber SIGN, Inc. (2005). Biometric signature
verificationtechnology overview. Retrieved
May 31, 2005, from http://www.cybersign.com/
techoverview.htm
Glover, D., & Miller, D. (2001). Running with
technology: The pedagogic impact of the largescale introduction of interactive whiteboards in
one secondary school. Journal of Information
Technology for Teacher Education, 10(3), 257276.
GRAHL Software Design. (2005). PDF annotatorAnnotate, edit and comment PDF files.
Retrieved May 31, 2005, from http://www.ograhl.
com/en/pdfannotator

Human Computer Interaction for Computer-Based Classroom Teaching

Griess, F. (2000). Online signature verification.

Unpublished masters thesis, Michigan State
University.
Grudin, J. (2001). Partitioning digital worlds: Focal and peripheral awareness in multiple monitor
use. In Proceedings of the SIGCHI Conference
on Human Factors in Computing Systems (pp.
458-465).
Hangai, S., Yamanaka, S., & Hamamoto, T. (2000).
Writer verification using altitude and direction
of pen movement. In Proceedings of the 15th
International Conference on Pattern Recognition
(Vol. 3, pp. 483-486). IEEE.
Heer, J., Newberger, A., Beckmann, C., & Hong,
J. I. (2003). Liquid: Context-aware distributed
queries. In Proceedings of UbiComp 2003 (pp.
140-148).
imc AG. (2003). LecturnityCreate your content
live and on the fly. Information Multimedia Communication AG. Retrieved May 31, 2005, from
http://www.lecturnity.de/
Jacobson, J. M., Comiskey, B., Anderson, P., &
Hasan, L. (2005). Electronic paper books and
electronic books. MIT Media Lab. Retrieved May
31, 2005, from http://www.media.mit.edu/micromedia/elecpaper.html
Johanson, B., Fox, A., & Winograd, T. (2002).
The interactive workspaces project: Experiences
with ubiquitous computing rooms. IEEE Pervasive
Computing Magazine, 1(2).
Klassen, R. V. (1991). Drawing antialiased cubic
spline curves. ACM Transactions on Graphics,
10(1), 92-108.

Multimedia (Part 1) (pp. 477-487). ACM Press.

Mller, R., & Ottmann, T. (2000). The authoring
on the fly system for automated recording and
replay of (tele)presentations. Special issue on multimedia authoring and presentation techniques.
Multimedia Systems, 8(3), 158-176.
Pen&Internet LLC. (2005). riteMailProduct
overview. Pen&Internet LLC. Retrieved May
31, 2005, from http://www.ritemail.net
Phelps, T. A., & Wilensky, R. (1997). Multivalent
annotations. In Proceedings of the First European
Conference on Research and Advanced Technology for Digital Libraries.
Prante, T., Streitz, N. A., & Tandler, P. (2004).
Roomware: Computers disappear and interaction
evolves. IEEE Computer, December, 47-54.
Rubine, D. (1991). Specifying gestures by example.
In Proceedings of the 18th Annual Conference on
Computer Graphics and Interactive Techniques
(pp. 329-337).
Russel, G., Chee, Y.- M., Seni, G., Yaeger, L.,
Tremblay, C., Franke, K., Madhvanath, S., &
Froumentin, M. (2004). Ink markup language.
W3C Working Draft. Retrieved May 31, 2005,
from http://www.w3.org/TR/InkML
Shneiderman, B., & Maes, P. (1997). Direct
manipulation vs. interface agents. Interactions,
4(6), 42-61.
Smart Technologies Inc. (2005). Retrieved May
31, 2005, from http://www2.smarttech.com
Wacom. (2005). ProcreatePainter classic. Retrieved May 31, 2005, from http://www.wacomasia.com/products/intuos2/soft/painter.html

MacIntyre, B., & Feiner, S. (1996). Future multimedia user interfaces. Multimedia systems, 4(5),
250-268.

Weiser, M. (1991). The computer for the 21st century. Scientific American, 265(10), 71-72.

Mukhopadhyay, S., & Smith, S. (1999). Passive

capture and structuring of lectures. In Proceedings of the 7th ACM International Conference on

Xerox Parc. (2005). Project Gyricon. Retrieved

May 31, 2005, from http://www2.parc.com/hsl/
projects/gyricon/

1457

Human Computer Interaction for Computer-Based Classroom Teaching

Zupancic, B., & Horz, H. (2002). Lecture recording and its use in a traditional university course.

In Proceedings of the 7th Annual Conference on

Innovation and Technology in Computer Science
Education (pp. 24-28). ACM Press.

This work was previously published in Enhancing Learning Through Human Computer Interaction, edited by E. McKay, pp. 2142, copyright 2007 by Information Science Reference, formerly known as Idea Group Reference (an imprint of IGI Global).

1458

1459

Chapter 4.3

Wikis as an Exemplary Model of

Open Source Learning
Robert Fitzgerald
University of Canberra, Australia

ABSTRACT
In their simplest form, Wikis are Web pages that
allow people to collaboratively create and edit
documents online. Key principles of simplicity,
robustness, and accessibility underlie the wiki
publication system. It is the open and free spirit
of Wikis fundamental to open source software
(OSS) that offers new contexts for learning and
knowledge creation with technology. This chapter
will briefly consider the role of technology in
learning before discussing Wikis and their development. The emerging literature on the application of Wikis to education will be reviewed and
discussed. It will be argued that Wikis embody an
exemplary model of open source learning that has
the potential to transform the use of information
communication technologies in education.

INTRODUCTION
Wikis are an instance of what is known as a read/
write technology. They allow groups of users,
many of whom are anonymous, to create, view,

and edit Web pages. In many cases, these pages

are online, but there are instances of Wikis used
as personal notebooks (e.g., Tiddlywiki, http://
www.tiddlywiki.com/). All wiki systems use a
simplified html markup language, but as their use
spreads, so does the appeal of more user-friendly
java-based WYSIWYG editors. It will be argued
that the simplicity, accessibility, and openness
of wikis support a model of collaboration and
knowledge building that represents an exemplary
model of learning with technology. This model is
congruent with many of the key principles embodied in free and open source software (FOSS)
and sociocultural theories of learning (Lave &
Wenger, 1991; Vygotsky, 1978; Wenger, 1998).
Many Internet-based communities and groups
are already finding ways to embrace these forms
of learning as a part of their ongoing process of
community capacity building. In contrast, formal
places of learning such as schools and universities
have been slow to explore the potential of this
technology. This chapter will briefly consider the
role of technology in learning before discussing
Wikis and their development. The chapter argues
that FOSS and Wikis in particular offer education

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Wikis as an Exemplary Model of Open Source Learning

far more than just low-cost software or even sound

principles of practice; they open up a space for
new models of learning and knowledge creation
with technology. The emerging literature on the
application of Wikis to education will be reviewed
before considering Wikis as an exemplary model
of open source learning.

TECHNOLOGY AND LEARNING

While the application of computing technology
to teaching and learning has at least a 30-year
history, there is a large body of literature that
suggests education is still struggling to use technology effectively (Cuban, 2001; Healy, 1998;
Oppenheimer, 2003; Postman, 1993; Stoll, 1999).
Results from large international studies show that
the dominant use of technology tends to focus
on skills and involves learners as content users
and not content creators (Kozma, 2003). Part of
the problem is that formal places of learning by
their very nature are highly structured contexts.
The role of context is important because there is
a direct relationship between form and quality of
the pedagogy and the form and quality of the resultant learning. Different teaching approaches and
learning contexts result in different outcomes for
students. In Boalers (1997) study of mathematics
classrooms, she showed that teacher-centered and
rule-based teaching approaches not only produce
low levels of student engagement but work to effectively limit the scope of the learning outcomes.
There is a strong suggestion from her work that
routine-style classrooms generate routine knowledge and that this is neither of the quality nor
quantity required for real-world mathematical
problem solving. Her key finding is that context
matters. The Russian neuropsychologist Alexandra Luria understood this relationship well when
he argued that cognition is a function of context.
Cognitive processes are not independent and
unchanging abilities they are processes occurring in concrete, practical activities and are

1460

formed within the limits of this activity (Luria,

1971, p. 266). In effect, Luria was saying that
cognition is plastic, a finding that has subsequently
been confirmed by contemporary neuroscience
(Goldberg, 1990, 2001). The activities and tasks
we set for learners not only determine the type
and quality of knowledge that is produced but,
more importantly, set the parameters for the development of their cognitive processes. Therefore,
from a philosophical and practical design point of
view, the contexts or settings of learning should
be as open and free as possible.
There is also a growing body of literature
suggesting that young people learn in different
ways to past students and, therefore, require (and
even demand) different teaching approaches (Gee,
2003; Oblinger, 2004; Prensky, 2001). Chris Dede
(2005) has written extensively in this area, and in
his assessment, these learners seek to co-design
their learning experiences and prefer communal
learning over individual learning. Anyone who
has recently studied in schools or universities
will know that despite institutional rhetoric to
the contrary, these new modes of teaching and
learning are not widespread. The pedagogical
challenge is to use technology in ways that build
upon learners existing experiences and foster the
creation of what von Krogh, Ichijo, and Nonaka
(2000) refer to as communities of knowledge. In
education, there is widespread recognition of the
need to explore more collaborative approaches to
learning (Jonassen, Peck & Wilson, 1999; Kozma,
2003; Laurillard, 2002; Sefton-Green, 2004;
Somekh, 2004). Wikis offer one such tool, which
is already a part of many learners everyday lives
as are a wide variety of other social software such
as blogs and social networking applications like
MySpace (http://www.myspace.com/).

BACKGROUND TO WIKIS
The founding developer of the World Wide Web
(WWW), Sir Tim Berners Lee, first conceived of

Wikis as an Exemplary Model of Open Source Learning

the Internet as a way for people to both read and

write data. The reality of nonintuitive operating
systems, html-based coding, clunky file transfer
tools, and security restrictions guaranteed that
while we could all read the Web, very few of us
were able to easily write and publish material.
To address some of these issues, Ward Cunningham developed the first Web-based Wiki
(http://en.wikipedia.org/wiki/WikiWikiWeb) in
1995 as part of the Portland Pattern Repository
(PPR). The origin of wikis, however, goes back
much earlier to the 1980s and Cunninghams
work on hypertext using Hypercard (eWEEK,
2006). In his original formulation (see Table 1),
Cunningham developed wiki software to adhere to
a number of core design principles (http://c2.com/
cgi/wiki?WikiDesignPrinciples).
These basic principles result in a Web publication system that is simple, robust, and accessible.
The system is designed to position users differently from traditional publishing modelsfrom
a Web reader to a Web author. This can be seen
particularly in the universal principle in which
any writer is automatically both a space organizer
and an editor.

FREE AND OPEN LEARNING

Free and open are key principles underpinning
FOSS. This chapter argues that these also underpin learning with Wiki technology. For a piece of
software to be open source, it must also adhere to
four freedoms summarized in Table 2 (http://www.
gnu.org/philosophy/free-sw.html).
While these freedoms relate to OSS, they are
enacted within a community of software developers in which community learning and knowledge
construction is central. In this sense, open can
be used to refer to something that is visible and
without barriers. In relation to FOSS, it also refers
to a principle of practice that embodies an attitude
of generosity. This is perhaps best thought about
in terms of an individual (or group) that makes
himself or herself available to others and is free
and willing to think about new ideas. The notions
of free and open have much to offer our conceptions of learning and the underpinning processes
of collaboration and knowledge construction
with Wikis.

Table 1. Wiki core design principles

Design Principle

Core functionality

Open

Should a page be found to be incomplete or poorly organized, any reader can edit it as they see fit

Incremental

Pages can cite other pages, including pages that have not been written yet

Organic

The structure and text content of the site are open to editing and evolution

Mundane

A small number of (irregular) text conventions will provide access to the most useful page markup

Universal

The mechanisms of editing and organizing are the same as those of writing so that any writer is automatically
an editor and organizer

Overt

The formatted (and printed) output will suggest the input required to reproduce it

Unified

Page names will be drawn from a flat space, so that no additional context is required to interpret them

Precise

Pages will be titled with sufficient precision to avoid most name clashes, typically by forming noun phrases

Tolerant

Interpretable (even if undesirable) behaviour is preferred to error messages

Observable

Activity within the site can be watched and reviewed by any other visitor to the site

Convergent

Duplication can be discouraged or removed by finding and citing similar or related content

1461

Wikis as an Exemplary Model of Open Source Learning

Table 2. Four freedoms of open source software

Freedom

Description

The freedom to run the program, for any purpose

The freedom to study how the program works, and adapt it to your needs Access to the source code is a
precondition for this

The freedom to redistribute copies so you can help your neighbour

The freedom to improve the program, and release your improvements to the public, so that the whole
community benefits. Access to the source code is a precondition for this

EXAMPLES OF WIKI PROJECTS

The use of Wikis had a significant development
when Jim Wales launched Wikipedia (http://
en.wikipedia.org) and the Wikimedia Foundation
(http://wikimediafoundation.org) in 2001. For the
first time, this project made Wiki technology and
tools freely and readily available to the public
while also offering a series of projects to which
they could contribute. The most recognized Wiki
is the free-content encyclopedia Wikipedia (http://
en.wikipedia.org/wiki/), which is maintained by a
group of volunteers from around the world and is
generally regarded as a rich information resource
(LeLoup & Ponerio, 2006; Lih, 2004).
Jim Wales explained:
Wikimedias mission is to give the worlds knowledge to every single person on the planet in their
own language. As part of that mission, Wikipedia is
first and foremost an effort to create and distribute
a free encyclopedia of the highest possible quality.
Asking whether the community comes before or
after this goal is really asking the wrong question:
the entire purpose of the community is this goal.
(Wales, 2005)
Wikipedia currently comprises 1,315,437
English language articles and nearly two million registered accounts (http://en.wikipedia.
org/wiki/Special:Statistics). Roughly 25% of
these articles have been translated into other
languages (nine at last count). Since July 2002,

1462

Wikipedians have made more than 70 million

edits. Web traffic statistics rank Wikipedia as the
17th most popular Web site (out of 250 million)
on the Internet with more than 2.9 billion page
views in the month of August 2006 (http://www.
alexa.com). Currently there is a number of related Wikimedia Foundation projects that are in
development, including Wikipedia (http://www.
wikipedia.org/); Wiktionary (http://wiktionary.
org/); Wikibooks; Wikinews (http://www.wikinews.org/); Wikiquote (http://www.wikiquote.
org/); Wikicommons (http://commons.wikimedia.org); Wikisource (http://wikisource.org/);
Wikispecies (http://species.wikimedia.org) and
the soon to be officially established Wikiversity
(http://en.wikibooks.org/wiki/Wikiversity). Each
of these projects takes the basic wiki model and
extends it into a specific area with a specific goal.
For example, Wikibooks (http://en.wikibooks.org)
is a project designed to produce open-content
textbook modules to create global curricula. To
date 21,019 book modules have been developed
for more than 1,000 books. Some of these books
are available in PDF format.
The large-scale application of wikis to community knowledge building can also be seen in
Wikia (http://www.wikia.com) and Wikispaces
(http://www.wikispaces.com/). The aim of these
initiatives is to provide individuals and communities with a Web site they can use to create
open content around their areas of interest. For
example, establishing a Wikia site requires the
topic to appeal to a large number of people and

Wikis as an Exemplary Model of Open Source Learning

that its content will have some longevity. Computer game players have been particularly active,
creating communities around their games. One
example of how this is used can be seen in the
ways that players of the massively multiplayer
online game Runescape (http://www.runescape.
com/) have built encyclopedic knowledge about
all aspects of the game (http://www.wikia.com/
wiki/Runescape).

LITERATURE REVIEW ON WIKIS

Many of the applications of Wikis are entirely
congruent with the so-called Web2.0 and social
software models that attempt to offer simple and
robust technologies to non-expert users so they are
able to create content and build communities. The
last few years have seen a proliferation of simple
tools for authoring Web content, particularly in the
area of personal blogging. Wikis are still without
peer when it comes to the large-scale collaborative
authoring of Web content (Lamb, 2004; Wagner,
2004). A literature based on the application of
Wikis is beginning to emerge, although to date
it is primarily conceptual and descriptive. Generally, there is widespread agreement that Wikis
represent an innovative and potentially powerful
tool for collaborative content creation and sharing
(Bold, 2006; Engstrom & Jewett, 2005; GodwinJones, 2003; Lamb, 2004; Wegner, 2004). There
have been few studies that have tested empirical
propositions, although many of these have been
exploratory in nature.
Wikis have been studied in language learning
(Godwin-Jones, 2003; LeLoup & Ponerio, 2006;
Wang et al., 2005; Wei et al., 2005) as tools in
higher education (Augar, Raitman, & Zhou, 2004;
Bold, 2006) to promote forms of participatory
journalism (Lih, 2004), as a tool for story-telling
in primary schools (Dsilets & Paquet, 2005), and
examined for their potential role in increasing
citizen participation in e-government initiatives
(Wagner, Cheung, & Ip, 2006).

In summary, the empirical work finds the

following:

The effective use of Wikis appears dependent on a clear goal matched to a group of
committed uses (Godwin-Jones, 2003).
Highly structured environments that rely
on top-down approaches (as opposed to
bottom-up) limit the potential of Wikis as a
tool for learning (Engstrom & Jewett, 2005;
Wagner, 2004).
Wikis such as Wikipedia are a rich source
of information that can promote content
creation, sharing, and discussion (LeLoup
& Ponerio, 2006, Lih, 2004).
It is important to augment students, Wiki
work with strategies to promote deep and
critical thinking to ensure high quality work
emerges (Engstrom & Jewett, 2005).
Wikis support a short edit-review cycle that
ensures the rapid development of content
(Lih, 2004).
Employing the user as organizer and editor (many eyeballs) is a highly effective
strategy for ensuring quality (Lih, 2004).

There have been widespread calls for more

research on Wikis (Lih, 2004; Wagner, 2004; Wei
et al., 2005). It will be particularly important to
develop research methods that are sensitive to
both the quality of content produced in concert
with how this content emerges within a community of learners. There is encouraging work on
the development of metrics to assess the quality
of Wiki (see Wikipedia) articles based on edit
histories (Lih, 2004), but we also need to examine
and assess the quality of the articles. Wiki edits
are easily quantified, but what they relate to is not
always clear. For example, a recent study found an
inverse relationship between the quantity of Wiki
edits and final exam scores (Wang et al., 2005).
The authors advised caution against interpreting
these findings as evidence that wikis are counterproductive to learning, but it does highlight the

1463

Wikis as an Exemplary Model of Open Source Learning

need for more nuanced and in-depth empirical

studies on Wikis.
Wikis have generated considerable interest in
education because they appear to support more
collaborative models of teaching and learning. It
is fair to say that there is considerable anecdotal
evidence that Wikis can and should play a key role
in e-learning in support of a more conversational
and dialogic approach to knowledge creation and
sharing.

AN EXEMPLARY MODEL OF
OPEN source LEARNING
Wikis offer a different model of creating, editing,
and sharing knowledge that is consistent with the
educational push toward what have become known
as sociocultural or constructivist approaches to
learning. A founding thinker in these areas, Lev
Vygotsky (1978), contended that learners neither
receive knowledge nor simply discover it. They
learn in social contexts in interaction with both
humans and tools. A key concept for Vygotsky was
the zone of proximal development (ZPD) in which
he said all learning takes place. In Vygotskys basic
model, it is adults who scaffold young learners,
helping to extend their thinking and learning.
However, as the technological tools develop and
evolve, we are beginning to see ways that both
humans and their tools can scaffold learning. The
technological spaces that make up wikis enable
new forms of sociotechnological ZPDs that support both individual and community knowledge
creation.
This focus on community and the power of joint
construction is taken up in The Wisdom of Crowds
(Surowiecki, 2004). Surowiecki argues that the
collective knowledge of large groups is often
unrecognized and almost always undervalued by
society. He explains that many everyday activities, from voting in elections and the operation of
the stock market to the way Google locates Web
pages, depend on the collective input and knowl-

1464

edge of large groups. Of course, not all crowds

are smart, but Surowiecki believes that under
the right conditions, crowds can act more wisely
than an expert individual. To achieve the best results, crowds must be able to support diversity of
opinion, relative independence in an individuals
thinking, a model of decentralization that allows
individuals to draw on their local knowledge and
aggregation; that is, embody a process whereby
individual knowledge can be combined into an
integrative whole. The ways technology might be
used to support the development of smart crowds
is a relatively unexplored area; however, applications such as Wikis, blogs, and multiplayer games
certainly show how large groups of people can
productively interact online. While there is a huge
qualitative difference between group interaction
and wisdom, there is a relationship to be explored
that highlights the importance of developing largescale social technologies such as Wikis.
The previous discussion suggested that collaborative knowledge creation should be an
important feature of formal learning; however, to
date, and particularly with reference to the use of
technology, it has not been. Some of the uses of
technology by young people in their everyday lives
seem to get closer to this goal. The way they work
in their communities around multiplayer games
to talk to each other and build knowledge is one
example. The next section considers the potential
that Wikis offer to achieve this goal.
One of the strengths of Wiki software is its
capability to document and record aspects of the
knowledge creation process. From an educational
point of view, this can provide valuable insights
to the knowledge construction process. In most
Wikis, an article features a number of views: the
article page, a discussion page, article editor, and
history. A rich edit history features full revision
history permitting comparison between current
and last entry. Edits can further be identified via
flags and commenting, helping others understand
the changes that have been made. To help assure
quality, edits also appear on a recent-changes page.

Wikis as an Exemplary Model of Open Source Learning

Rules can be set up to show pages that have been

changed since the last visit. A list of contributions
by users offers various analyses, often with full
history and comparison tools. In combination,
these tools open up the possibility of exploring
the relationship (and tension) between individual
and group constructions.
Surowiecki said:
Any crowdwhether it be a market, a corporation, or an intelligence agencyneeds to find
the right balance between the two imperatives:
making individual knowledge globally and collectively useful (as we know it can be), while still
allowing it to remain resolutely specific and local.
(Surowiecki, 2004, p. 140)
Wikis allow both individual contributions and
the evolving group product to sit alongside each
other. The examples of Wikis outlined here seem
able to create new forms of sociotechnological
ZPDs for learners. These zones support both
individual and community knowledge creation
in ways that are consistent with the notion of
communities of practice (Lave & Wenger, 1991;
Wenger, 1998).
Wikis appear well suited to building knowledge in which the representation of balanced
opinion is valued. While there is no guarantee that
this prevents wikiwars, it does seem that strong
opinion is better suited to other spaces such as
blogs. It is clear from the work of the Wikimedia
Foundation that there are no hard and fast rules
to using Wikis. What is apparent, though, is that
many of the more successful projects embody a
spirit of community characterized by openness
and freedom.

CONCLUSION
In the end, the success of innovations in learning such as Wikis will be seen in the increased
capacity of individuals and their communities to

create and apply new knowledge. Incorporating

tools that not only facilitate but also document the
effective management of information and creation
of knowledge is now essential for an innovative
and productive 21st-century society. Wikis are
significant and innovative because they attempt
to position learners as knowledge creators rather
than simply content users. They also represent the
application of new collaborative technologies in
ways that are free and open. In terms of education,
the Wiki model locates the challenge of improving
information, data management, and knowledge
creation processes within a community model. It
also builds on strategies for increasing the capacity
of all community members to expand their ways
of thinking creatively and working collaboratively.
A key feature of this approach is that knowledge
and practice are shared in a spirit of generosity.
The extensive use of Wikis in education begs the
question: Are we really prepared to engage in this
form of open source learning?

REFERENCES
Augar, N., Raitman, R., & Zhou, W. (2004). Teaching and learning online with wikis. In R. Atkinson,
C. McBeath, D. Jonas-Dwyer, & R. Phillips (Eds.),
Beyond the comfort zone: Proceedings of the 21st
ASCILITE Conference (pp. 95-104). Retrieved June
17, 2006, from http://www.ascilite.org.au/conferences/perth04/procs/augar.html
Boaler, J. (1997). Experiencing school mathematics:
Teaching styles, sex and setting. Buckingham, UK:
Open University Press.
Bold, M. (2006). Use of Wikis in graduate course
work. Journal of Interactive Learning Research,
17(1), 5-14.
Cuban, L. (2001). Oversold and underused: Computers in the classroom. Cambridge, MA: Harvard
University Press.

1465

Wikis as an Exemplary Model of Open Source Learning

Dede, C. (2005). Planning for neomillennial learning

styles. In D. Oblinger & J. Oblinger (Eds.), Educating
the net generation (pp. 224-249), Retrieved July 4,
2006, from http://www.educause.edu/educatingthenetgen
Dsilets, A., & Paquet, S. (2005). Wiki as a tool for
Web-based collaborative story telling in primary
school: A case study. In P. Kommers & G. Richards
(Eds.), Proceedings of World Conference on Educational Multimedia, Hypermedia and Telecommunications 2005 (pp. 770-777). Chesapeake, VA: AACE.
Retrieved June 17, 2006, from http://iit-iti.nrc-cnrc.
gc.ca/iit-publications-iti/docs/NRC-48234.pdf
Engstrom, M. E., & Jewett, D. (2005). Collaborative
learning the wiki way. TechTrends, 49(6), 12-16.
eWEEK.com. (2006). Father of Wiki speaks out
on community and collaborative development.
Retrieved July 14, 2006, from http://www.eweek.
com/article2/0,1895,1939982,00.asp
Godwin-Jones, R. (2003). Blogs and Wikis: Environments for on-line collaboration. Language Learning
& Technology, 7(2), 12-16. Retrieved June 20, 2006,
from http://llt.msu.edu/vol7num2/emerging/
Goldberg, M. E. (1990). Contemporary neuropsychology and the legacy of Luria. Hillsdale, NJ:
Lawrence Erlbaum.
Goldberg, M. E. (2001). The executive brain:
Frontal lobes and the civilized mind. New York:
Oxford Press.
Healy, J. (1998). Failure to connect: How computers affect our childrens minds. New York: Simon
& Schuster.

spaces: Wikis, ready or not. Educause, 36-48.

Laurillard, D. (2002). Rethinking university teaching.
London: Routledge.
Lave, J., & Wenger, E. (1991). Situated learning:
Legitimate peripheral participation. New York:
Cambridge University Press.
LeLoup, J. W., & Ponerio, R. (2006). Wikipedia: A
multilingual treasure trove. Language Learning &
Technology, 10(2), 12-16. Retrieved June 20, 2006,
from http://llt.msu.edu/vol10num2/net/
Lih, A. (2004). Wikipedia as participatory journalism: Reliable sources? Metrics for evaluating
collaborative media as a news resource. In Proceedings of the 5th International Symposium on Online
Journalism. Retrieved July 7, 2006, from http://jmsc.
hku.hk/faculty/alih/publications/utaustin-2004wikipedia-rc2.pdf
Luria, A. R. (1971). Towards the problem of the
historical nature of psychological processes. International Journal of Psychology, 6, 259-272.
Oppenheimer, T. (2003). The flickering mind: The
false promise of technology in the classroom and
how learning can be saved. New York: Random
House.
Postman, N. (1993). Technopoly: The surrender of
culture to technology. New York: Vintage Books.
Stoll, C. (1999). High tech heretic: Why computers
dont belong in the classroom and other reflections
by a computer contrarian. New York: Doubleday.
Surowiecki, J. (2004). The wisdom of crowds. New
York: Doubleday.

Jonassen, D., Peck, K. L., & Wilson, B. G. (1999).

Learning with technology: A constructivist perspective. Upper Saddle River, NJ: Prentice Hall.

von Krogh, G., Ichijo, K., & Nonaka, I. (2000).

Enabling knowledge creation. New York: Oxford
University Press.

Kozma, R.B. (2003). Technology and classroom

practices: An international study. Journal of Research on Technology in Education, 36(1), 1-14.

Vygotsky, L. (1978). Mind in society. Cambridge,

MA: Harvard University Press.

Lamb, B. (2004, September/October). Wide open

1466

Wagner, C. (2004). Wiki: A technology for con-

Wikis as an Exemplary Model of Open Source Learning

versational knowledge management and group

collaboration. Communications of the Association
for Information Systems, 13, 265-289.
Wagner,C. Cheung, S. K., & Ip, K. F. (2006). Building
semantic Webs for e-government with wiki technology. Electronic Government, 3, 36-55.
Wales, J. (2005). Letter from the founder. Retrieved
July 14, 2006, from http://wikimediafoundation.
org/wiki/Founder_letter
Wang, H. C., Lu, C. H., Yang, J. Y., Hu, H. W.,
Chiou, G. F., Chiang, Y. T., & Hsu, W. L. (2005).
An empirical exploration of using wiki in an English as a second language course. In Proceedings
of 5th IEEE International Conference on Advanced
Learning Technologies (pp. 155-157). Retrieved
July 17, 2006, from http://www.iis.sinica.edu.tw/
IASL/webpdf/paper-2005-An_Empirical_Exploration_on_Using_Wiki_in_an_English_as_Second_Language_Course.pdf
Wei, C., Maust, B., Barrick, J., Cuddihy, E., &
Spyridakis, J. H. (2005). Wikis for supporting
distributed collaborative writing. In Proceedings
of the Society for Technical Communication 52nd
Annual Conference. Retrieved July 7, 2006, from
http://www.uwtc.washington.edu/research/pubs/jspyridakis/STC_Wiki_2005_STC_Attribution.pdf
Wenger, E. (1998). Communities of practice: Learning, meaning, and identity. New York: Cambridge
University Press.

KEY TERMS
Constructivist: An approach based on the
work of Lev Vygotsky, who contended that learners neither receive knowledge nor simply discover

it. They learn in social contexts in interaction with

both humans and tools.
Free and Open Source Software (FOSS): A
term first described by Richard Stallman referring
to a software development process in which the
software source code is made freely available for
subsequent modification and development.
Hypercard: A hypermedia program developed by Apple Computer in the 1980s.
Open Source Learning: A model of learning
inspired by the key principles or freedoms embodied
in the FOSS movement.
Web2.0: A term coined by Tim OReilly (http://
tim.oreilly.com/) referring to a range of secondgeneration Web publishing and social networking
technologies.
Wiki: A form of read/write technology that allows groups of users, many of whom are anonymous,
to create, view, and edit Web pages.
Wikia: A project to provide communities with
Wiki-type Web sites (see http://www.wikia.com).
Wikimedia Foundation: An international
nonprofit organization run by Jim Wales, using wiki
technology to promote free and open large-scale
collaborative content creation projects (http://wikimediafoundation.org).
Wikipedia: A free-content encyclopedia (http://
en.wikipedia.org/wiki/).
Zone of Proximal Development (ZPD): The
difference between what learners can do by themselves and with the assistance of more capable
adults or peers.

This work was previously published in Handbook of Research on Open Source Software: Technological, Economic, and Social
Perspectives, edited by K. St.Amant and B. Still, pp. 681-689, copyright 2007 by Information Science Reference, formerly known
as Idea Group Reference (an imprint of IGI Global).

1467

1468

Chapter 4.4

Re-Examining the
Socioeconomic Factors
Affecting Technology Use in
Mathematics Classroom
Practices
Emiel Owens
Texas Southern University, USA
Holim Song
Texas Southern University, USA
Terry T. Kidd
University of Texas School of Public Health, USA

Abstract
Over the past 15 years a considerable amount
of research has been devoted to study of the
socio
economic
aspects that affect the use of
technology in the mathematics classroom. With
the call for curricular and instructional reform,
educational institutions have embarked on the
process to reform their educational practices to aid
the urban student in their quest to obtain quality
mathematics and science based education with the
integration of technology. The study performed

was to reexamine the socio

economic
disparities
in the mathematics classroom as it relates to
implementing technology interventions to support
quality teaching and active student learning. This
article is to provide empirical evidence of whether
these disparities continue to exist and their effects on student achievement in the mathematics
classroom. The results of this study showed an
overall positive relationship regarding the use of
technology interventions within the mathematics
classroom with levels of student achievement,

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Re-Examining the Socioeconomic Factors Affecting Technology Use in Mathematics Classroom

showing a clear signs of continued disparities

within mathematics classroom.

Introduction
The introduction of microcomputers into classrooms during the 1980s was heralded by many as
the dawn of a new era in American education. Proponents argued that technology had the potential
to fundamentally transform the nature of teaching
and learning (Papert, 1980; U.S. Congress, Office
of Technology Assessment, 1988). However, over
time, it has become apparent that it is far easier
to acquire hardware, software, and Internet access (Becker, 1991; Dividing Lines, 2001) than
it is to capture the potential of technology in
significantly meaningful outcomes (Cuban, 2001).
Likewise, educators concerned about the chronic
underachievement of students often fall prey to
the allure of technology as a tool for reversing the
historical influences of poverty, discrimination,
inequity, chronic underachievement, and lack of
opportunity. However, 25 years after the introduction of the computer into the classroom, many of
the expectations associated with technology in
education remain unrealized to some but to other,
technology has proven to be an effective tool in
the efforts to provide students with opportunities
for quality teaching and active student learning
and engagement.
Educational institutions have called for instructional and curriculum reform that includes
active engagement of students, quality assessments, and the increased and innovative use
of technology applications to promote quality
teaching and active student learning (U.S. Department of Education, 2001). This is true in the
field of mathematics where organizations such as
the National Council of Teachers of Mathematics (1989, 2000), Mathematical Science Board
(MSEB, 1991), and Mathematics Association of
America (1991) have stressed that technology is
essential in teaching and learning mathematics.

The underlying assumption of these organizations

and math educators alike is that technology will
enable students to explore mathematics with a
greater depth and will allow them to study topics
that were previously impractical (Garofalo, Drier,
Harper, Timmerman, & Shockey, 2000). However,
in order for technology to have greatest impact
on our educational system, all students must have
access to technology. For that reason technology
has the potential to narrow the achievement gap,
if equally distributed or widen the gap if only
accessible to selected groups in the educational
system (Kulik, 2002; Waxman, Connell, & Gray,
2002).

Review of Literature
Over the past 15 years a considerable amount of
research has been devoted to sociocultural disparity in technology availability and use in the
mathematics (Becker, 2000; Garofalo et al., 2000;
Means et al., 2001; Manoucherhri, 1999; National
Center for Educational Statistics, 2004; Owens,
1993; Owens & Waxman, 1994, 1995; U.S. Department of Education, 2001; Huang & Waxman,
1996). Past studies conducted by Becker (2001) and
Coley, Cradler, and Engel (1997) found students
from higher income families have been found to
use computers in school and in their homes more
frequently than students from lower-income families. Students of color from urban schools have
also been found to have less access to computers
compared to Anglo-suburban students (Owens &
Waxman, 1993, 1994). More recently, lower SES
schools are only half as likely to have high speed
Internet compared to high SES schools (Advanced
Telecommunications, 1997). Consistent with this
idea of access are the issues within the digital
divide itself. Within the past decade, a growing
body of evidence supports the ever-widening
technological gap among members of society, in
particular children and the elderly (NTIA, 1995,
1997, 1999), with a important emphasis on urban

1469

Re-Examining the Socioeconomic Factors Affecting Technology Use in Mathematics Classroom

schools with the inner cities. The groups identified

who lack access to information and technological
resources include people of color, specifically
African American and Hispanic Americans,
those who are poor and of the working class, individuals of low-income, those who possess less
than a high school level of education, children
of single parents, and residents of inner-cities
(NTIA, 1995, 1997, 1999, 2002, 2004). Access
is key in this issue.
Despite the constraints on school funding in
most states, schools have devoted an increasing
percentage of their annual budgets to technology.
The majority of the efforts of the educational com
munity over the past decade to acquire hardware,
software, and Internet access have been successful
(Dividing Lines, 2001; Education Week, 2004).
However, clear evidence of a digital divide, parallel
to historical disparities, continues to distinguish
urban schools from their affluent counterparts
(Chen & Thielemann, 2001; Guttentag & Eilers,
2004; National Center for Education Statistics,
2004). Historical measures of digital equity have
been based on the ratio of the number of computers
divided by the number of students. A more recent
measure involves determining levels of Internet
access. Another dimension of this problem relates
to questions about differences in home access to
technology, therefore impacting urban student
achievement.
A second source of disparity in technology
use is how technology is used. Previous studies
conducted by Becker (2001) and Finneran (2000)
found that low SES schools are more likely to use
technology for drill and practice, whereas high
SES school uses technology in innovative teaching
strategies. This idea is consistent with the ideas
of curriculum reform and reconceptualization put
forth by Pinar (2002). Pinar (2002) suggests that
the instruments of computer technology are used
to drill and kill students into passing standardized
test, not actually being integrated effectively into
classroom instruction or pedagogical practice
that promote quality teaching and active student

1470

learning. He further explains that the current use

of computer technology in urban schools generally
serve to turn its users (students) into disembodied
and alienated subjects. Furthermore, as explained
in Becker (2001) high SES students are more likely
to use technology for school assignment, use emails, and use educational programs.
Finally, educational reform programs such
as the No Child Left Behind Act of 2001 has
focused on accountability and high-stakes testing. The renewed importance of test scores will
fundamentally alter the attention and activities
of low-achieving schools as they seek to achieve
the required benchmark scores (Johnson, 2002).
As a result, it was anticipated that previous technology buying sprees will be redirected. One
approach to this redirection is that there is new
emphasis placed on technology initiatives that
enhance schools assessment and early intervention efforts (Education Week, 2003) that feature
learning activities with instructional objectives
and record keeping.
A third source of disparity in technology use
deals with the nature of technology adoption and
organizational change. A thorough analysis of
major research related to technology and teacher
motivation, adoption and usage uncovered important factors that are involved in determining their
willingness to use such approaches in the teaching
and learning process (Braak, 2001; Chiero, 1997;
Fabry & Higgs, 1997; Solomon & Wiederhorn,
2000). The literature indicates that teachers willingness to adopt and implement learning technologies for the teaching and learning process varies,
but all share the same denominatorproper
internal and external motivation (Braak, 2001;
Chiero, 1997; Fabry & Higgs, 1997; Mooji &
Smoot, 2001; Solomon & Wiederhorn, 2000). The
top reasons teachers choose to use technology for
the teaching and learning process dealt with the
notion of continuous training and development,
proper technology support from technology
personnel, encouragement from school administration, and an organizational structure that sup-

Re-Examining the Socioeconomic Factors Affecting Technology Use in Mathematics Classroom

ported teachers using technology (Grank, Zhao,

& Borman, 2004). While these are the key issues
ascertained from the research study, the ideas of
change management, technology adoption and
innovation, self-efficacy and motivation, support,
and the computing experience are relevant to this
discussion. The American Council of Education
(1999) report noted that the quest to optimize
the use of technology in schools could not be
fulfilled by the mere supply of more hardware and
software. The report suggested that what needs
to be addressed first are ways identify, motivate,
and equip teachers with the skills necessary for
effective use of educational related multimedia
technologies in their classroom practices.
In contrast to the indicators that profile the
acquisition of technology, less information is
typically available to reflect progress toward
implementing technology applications that enhance the teaching and learning process (NTIA,
1995, 1997, 1999, 2000, 2002; CEO Forum, 1997).
It is rather a task of ease to promote and profile
computer-to-student ratios in school reform experiments with little regard to how technology
is used in the urban classroom (Brainbridge,
Lasley, & Sundre, 2004; Guttentag & Eilers,
2004). The use of technology in the classroom
has moved through definable periods involving
programming, computer-assisted instruction,
problem-solving environments, personal productivity, Web-based instruction, and hypermedia.
Problems associated with limited opportunities
for teacher professional development to learn
about new innovations (Lonergan, 2001), as well
as limited funds for new hardware and software,
often result in the routine use of student learning
activities that have been abandoned by high-performing schools (Burnett, 1994). As a result, it is
essential that questions be raised about the nature
of technology learning activities in urban schools
when expectations are held that technology is
used to enhance academic achievement. While
looking into the use of technology in the urban
school environment, it has been noted that the use

of these tools does not reflect innovative uses in

integration into the curriculum for active student
learning, engagement or to enhance the teaching
process. Typically urban schools have antiquated
resources and in adequate technological support
for the use of technology in classroom.
Although these studies have established a pattern for technology disparities in the past, most of
these studies rely on data collected in the mid to
late 1990s. Moreover, the last national report on
the status of technology use entitled Teachers
Tools for the 21st Century: A Report on Teachers Use of Technology was published in 2000,
however, it used data from the 1999 FRSS survey
(National Center for Educational Statistics, 1999).
Therefore, updated studies are needed that examine current data to determine where we are in the
quest to narrow the achievement gap with the aid
of technology. The purpose of the present study
is to reexamine technology use and to provide
evidence of whether or not disparity issues still
exist using the latest national survey produced
by the National Center for Educational Statistics
(2002). Our focus in this report is on the technology practices in the mathematics classroom.

Methods
The data for the present study was drawn from
the base year of the Educational Longitudinal
Survey of 2002-2004 (ELS: 02). The ELS:02 is a
national survey administered by National Center
for Educational Statistics to provide an overall
picture of the present educational experience
of high school students in the U.S. The survey
has four subcomponents completed by students,
teachers, parents, and school administrators.
The design used to collect the data was a twostage, stratified national probability sample. The
survey included 1,052 public and private school
representing about 12,000 students across the
country. For the present study, only the student
survey data was used.

1471

Re-Examining the Socioeconomic Factors Affecting Technology Use in Mathematics Classroom

The student survey provided items used to

measure how often technology is being used in
mathematics classrooms along with how this
technology is being used as reported by students.
The items used to measure how often technology
is being used in the classroom were the following: (a) how often do you use calculators in your
mathematics classroom?; (b) how often do you use
graphing calculators in your mathematics classroom?; and (c) how often do you use computers
in your mathematics classroom?. The outcome
measures for these items were a five-point Likert scale, ranging from never to everyday or
almost.
To measure how technology is being used in
mathematics classrooms the following items were
used: (a) how often do you use a computer to review
your math work in your mathematics classroom?;
(b) how often do you use a computer to solve math
problems in your mathematics classroom?; (c)
how often do you use a computer for graphing of
math problems in your mathematics classroom?;
(d) how often do you use a computer to practice
math drills in your mathematics classroom?; (e)
how often do you use a computer to analyze data
in your mathematics classroom?; (f) how often
do you use a computer to apply learning in your
mathematics classroom?; (g) how often does your
math teacher use a computer for one-on-one instructions in your mathematics classroom?; and (h)
how often does your math teacher use computer to
show new topics in your mathematics classroom?.
The outcome measures for these items were also
a five-point Likert scale, ranging from never to
everyday or almost. The survey also provided a
measure of students socioeconomic status. This
measure was developed as a composite variable
constructed from five equally weighted, standardized components: fathers education, mothers
education, family income, fathers occupation,
and mothers occupation. The distribution of
these SES values were divided into four quartiles,
ranging from lowest to highest. To determine if
an association exists between technology use

1472

and students socioeconomic status a Pearsons

chi-square was used.

Results
Table 1 reports the results of the frequency of
calculator and computer use in mathematics
classrooms. The results indicated that students are
using more calculators in their math classrooms
compared to computers. 58% of the students reported that they had used calculators every day
in their math classroom compared to 8% that
indicated they were using computers on a daily
basis. 30% of the students reported using the
graphing calculator on a daily basis. One-third
of these students reported they never use the
graphing calculator in their classroom. 61% of
the students indicated they never use computers
in their math classroom. Finally, 7.4% indicated
they were using computers on a daily basis.
Table 2, compares calculator and computer use
across socioeconomic levels. The results indicate
that a significant association (p<.001) exists between calculator use and socioeconomic levels.
In this case, the lowest SES group reported using
calculators on a daily basis the least. On the other
hand students in the highest SES group reported
using calculators on a daily basis more often. 48%
of the lowest SES group reported using calculators
on a daily basis compared to 68% of the high SES
Group. There was also a significant association (p
< .001) between daily use of graphing calculators
and SES group membership. 21% of the students
classified in the lowest SES reported using the
graphing calculator on a daily basis compared
to about twice as many students (42%) classified
in the highest SES group. The final comparison
looked at computer usage across SES levels. The
results also indicated a significant association
exists (p < .001) between computer usage and
SES levels. In this case, students classified in
the lowest SES group were more likely to use
computers computer compared to those students

Re-Examining the Socioeconomic Factors Affecting Technology Use in Mathematics Classroom

Table 1. How often technology is being used in mathematics classrooms overall frequency
N=11,618
How often do you use calculators in your math class

How often do you use graphing calculator in your math class

How often do you use computers in your math class

6.2

Never

12.0

Rarely

7.7

Less than once a week

18.0

Once or twice a week

58.0

Everyday or almost

33.1

Never

19.6

Rarely

3.3

Less than once a week

2.2

Once or twice a week

8.8

Everyday or almost

60.7

Never

2.2

Rarely

7.7

Less than once a week

7.8

Once or twice a week

7.4

Everyday or almost

Table 2. Socioeconomic status and how often technology is being used in mathematics classrooms percentage
of use
N= 11,618

How often do you use calculators in your math

class

n=2627

n=2820

n=2896

n=3275

Lowest

Second

Third

Highest

8.83%

6.67

5.91

4.12

Never

18.12

12.91

11.08

7.18

Rarely

6.78

6.24

5.24

4.82

Less than once

a week

19.57

18.94

17.68

16.24

Once or twice
a week

46.71

55.25

60.08

67.63

Everyday or
almost

37.65

37.30

32.98

25.83

Never

22.95

20.47

20.30

15.66

Rarely

16.16

6.24

6.46

5.56

Less than once

a week

31.31

10.96

11.15

11.42

Once or twice
a week

94.94

25.00

29.11

41.53

Everyday or
almost

Chi. Sq = 341.23, p < .001***

How often do you use graphing calculator in
your math class

continued on following page

1473

Re-Examining the Socioeconomic Factors Affecting Technology Use in Mathematics Classroom

Table 2. continued
Chi. Sq = 375.88, p < .001***
How often do you use computers in your math
class

53.73

61.06

62.81

64.09

Never

21.74

19.50

18.85

20.92

Rarely

13.13

5.00

6.00

5.80

Less than once

a week

53.53

6.24

5.28

4.09

Once or twice
a week

9.86

8.19

7.04

5.10

Everyday or
almost

Chi. Sq = 139.11, p < .001***

in the high SES group. The percentage of daily

usage of computers for students in the lowest SES
group compared to the highest SES group was
10% and 5% respectively.
Table 3 reports the frequency of how computers
are being used in math classrooms. The overall
results indicate that the daily use of computers
is very low. The highest percentage of daily use
for computer use was to solve math problems
(16.3%). The next two popular uses for computers
in mathematics classrooms on a daily basis were
for to apply learning and to practice math drill.
The percentages were 14% and 13% respectively.
However, over one-third of the students (34%)
reported they never used the computer to practice
math drills. The least daily use for computers was
to provide one-on-one instruction. The reported
percentage was 7%. On the other hand, 52% of
the students indicated that their teacher never uses
the computer for one-on-one instructions. Over
50% of the students reported they never used
computers to review math work. Similarly, 52%
of the students indicated that their teacher never
uses the computer for one-on-one instructions.
About 41% of the students indicated that there
teacher never uses the computer to show them
new topics. Over one-third of the student (35%)
reported that they never used the computer to
analyze data. Finally, about 30% of the students

1474

reported that they never used the computer for

graphing math problems.
Table 4, compares computer use across socioeconomic levels. The results indicated that the
higher computer use was significantly associated
with low SES classification. Moreover, as SES
Levels increased computer usage decreased. The
most frequent use of computers was to solve math
problems. The results indicate that about 21% of
the low SES students compared to about 10% of
the high SES students reported they used computers to solve math problems everyday or almost
everyday. Students classified as second and third
SES levels reported about 18% and 15% of the
times they used computers to solve math problems
everyday or almost everyday respectively. The
next most frequent use of computers was to apply
learning. Students classified as low SES indicated
that used computers for this purposes everyday or
almost everyday about 17% of the times compare
to about 9% for the high SES students. The second and third level SES students indicated they
used computers everyday or almost everyday to
apply learning about 15% and 13% of the times
respectively. Similarly, about 17% of the lowest
SES group reported using computers to practice
math drills compared to about 10% of the high SES
students everyday or almost everyday. Furthermore, about 41% of the high SES students reported

Re-Examining the Socioeconomic Factors Affecting Technology Use in Mathematics Classroom

Table 3. How computers are used in mathematics classrooms overall frequency

N= 11,618
How often do you use computer to review math work

How often do you use computer to solve math problems

How often do you use the computer for graphing in math

How often do you use the computer to practice math drills

How often do you use the computer to do analyze data

How often do you use the computer to apply learning

How often does your math teacher uses computer for one-on-one
instructions

52.4

Never

22.3

Rarely

8.6

Less than once a week

7.6

Once or twice a week

9.1

Everyday or almost

28.6

Never

6.9

Rarely

14.7

Less than once a week

13.5

Once or twice a week

16.3

Everyday or almost

30.6

Never

29.4

Rarely

18.0

Less than once a week

12.2

Once or twice a week

9.9

Everyday or almost

33.6

Never

24.8

Rarely

15.3

Less than once a week

13.3

Once or twice a week

13.0

Everyday or almost

35.1

Never

26.4

Rarely

3.3

Less than once a week

12.1

Once or twice a week

9.2

Everyday or almost

28.0

Never

26.5

Rarely

18.1

Less than once a week

13.6

Once or twice a week

13.8

Everyday or almost

51.6

Never

22.9

Rarely

8.8

Less than once a week

6.6

Once or twice a week

9.9

Everyday or almost

continued on following page

1475

Re-Examining the Socioeconomic Factors Affecting Technology Use in Mathematics Classroom

Table 3. continued
How often does your math teacher use the computer to show new
topics

41.1

Never

24.3

Rarely

9.9

Less than once a week

1.1

Once or twice a week

10.7

Everyday or almost

Table 4. Socioeconomic status and how computers are used in mathematics classrooms percentage of
use

How often do you use computer to review your math

work

n=2627

n=2820

n=2896

n=3275

Lowest

Second

Third

Highest

38.76%

50.19

56.81

65.61

Never

27.91

23.74

20.42

16.32

Rarely

11.16

7.34

8.72

6.84

Less than once a week

9.73

9.27

5.75

5.44

Once or twice a week

12.44

9.46

8.30

5.80

Everyday or almost

21.36

26.19

30.89

36.74

Never

29.15

23.81

25.61

28.37

Rarely

15.15

14.97

15.07

13.64

Less than once a week

Chi. Sq = 96.72, p < .001***

How often do you use a computer to solve math problems

12.99

16.67

13.18

11.47

Once or twice a week

21.36

18.37

15.25

9.77

Everyday or almost

Chi. Sq = 70.91, p < .001***

How often do you use a computer

30.16

31.90

31.80

28.96

Never

to graph math problems

27.74

28.77

28.75

32.19

Rarely

16.29

16.43

18.30

21.04

Less than once a week

12.74

13.50

11.93

10.43

Once or twice a week

13.07

9.39

9.15

7.37

Everyday or almost

27.41%

32.21

34.61

40.73

Never

26.67

21.17

25.05

25.67

Rarely

15.56

16.37

16.25

13.15

Less than once a week

13.04

17.26

13.00

10.30

Once or twice a week

17.33

12.99

11.09

10.14

Everyday or almost

32.95

33.14

36.46

38.14

Never

24.75

26.98

24.02

29.56

Rarely

17.54

15.68

19.65

16.61

Less than once a week

Chi. Sq = 20.38, p < .060

How often do you use a computer to practice math drills

Chi. Sq = 50.60, p < .001***

How often uses computer to analyze data

continued on following page

1476

Re-Examining the Socioeconomic Factors Affecting Technology Use in Mathematics Classroom

Table 4. continued
13.77

13.49

10.04

10.58

Once or twice a week

10.98

10.71

9.83

5.11

Everyday or almost

26.48

27.85

25.53

31.77

Never

24.41

26.40

28.60

27.10

Rarely

17.16

17.00

19.77

18.55

Less than once a week

Chi. Sq = 27.66, p < .006**

How often do you use a computer to apply learning

14.50

13.56

12.86

13.39

Once or twice a week

17.46

15.19

13.24

9.19

Everyday or almost

43.32

49.61

51.08

63.07

Never

26.41

24.22

22.41

18.38

Rarely

9.66

11.13

11.42

7.60

Less than once a week

9.82

8.40

9.48

6.89

Once or twice a week

10.79

6.64

5.60

4.06

Everyday or almost

41.01

41.71

Chi. Sq = 26.05, p < .011*

How often does your math teacher use a computer for
one-on-one instructions

Chi. Sq = 61.59, p < .001***

How often does your math teacher use a computer to
show new topics

38.16
42.99
Never

24.52

25.50

23.68

23.47

Rarely

12.33

13.30

14.85

15.12

Less than once a week

8.77

10.93

11.47

9.61

Once or twice a week

13.37

8.56

11.84

8.82

Everyday or almost

Chi. Sq = 16.96, p < .151

never using computers to practice math drills

compared to about 27% of the low SES students.
About 13% of the second level SES students and
11% of the third level reported using computers
everyday or almost everyday to practice math
drills. When asked if their math teacher used
computers to show them new topics about 13%
of the low SES students and 9% of the students
in the high SES group indicated this happened
everyday or almost everyday. Twelve percent of
the third level SES group and 9% of the second
level SES students reported their teacher used
computers to show them new topics everyday or
almost everyday. Equally, about 13% of the low

SES students reported using computers everyday

or almost everyday to graph math problems. About
7% of the high SES students and 9% of the second and third level SES students reported using
computers everyday or almost everyday to graph
math problems. 12% of the low SES students and
6% of high SES students reported that they use
computers to review their math work everyday
or almost everyday. Also, about 66% of the high
SES students reported that they never use computers to review their math work compared to 39%
of the low SES students. 57% of the third level
SES students and 50% of the second level SES

1477

Re-Examining the Socioeconomic Factors Affecting Technology Use in Mathematics Classroom

students reported that they never use computers

to review their math work
Twice as many low SES students reported that
their teacher used the computer to provide oneon-one instructions everyday or almost everyday
compared to students in the high SES group. The
reported percentages for the two levels were 11%
and 4% respectively. Additionally, 63% of the high
SES students compared to 43% of the low SES
students indicated that their teacher never used
computers for one-on-one instructions. About
50% of the second and third level SES students
indicated that their teacher never used computers
for one-on-one instructions. Finally, about 11%
of the lowest and second level of SES students
reported using computers to analyze data everyday
or almost everyday compared to only 5% of the
students in the high SES group. About 10% of the
third level SES students reported using computers
to analyze data everyday or almost everyday

Discussion
The use of technology in the math and science
classroom has been a main focus in improving
student learning outcomes. Technology not only
can provide visual learning in the classroom,
it also opens the door to improve higher level
thinking skills. The results of the present study
indicate that 10th graders use more calculators on
a daily basis compared to computers. Moreover,
calculator use far outweighs the use of computers
in todays math curriculum. This is also true for
the use of the graphing calculator.
The results of the present study suggests that
there are important differences in the use of
technology in 10th grade mathematics classrooms
associated with levels of SES status. Students from
low SES families are less likely to use calculators
on a daily basis compared to students from high
SES families. This also includes the use of the
graphing calculator on a daily basis. Low SES
students also reported that they were more likely

1478

to use computers on a daily basis compared to high

SES students. This may dispel previous findings
that low SES students have a less opportunity to
use computers compared to high SES students.
However, the findings do raise a new issue of
disparity in calculator use, which there may be
need for further exploration.
Another important finding from this study
describes students overall use of computers. The
results from the present study suggest that overall
10th grade students do not use computers often in
their mathematics class. This suggests that schools
across the country need to do a much better job
in integrating technology in the secondary school
curriculum. Depending on its use, technology can
be very useful in the mathematics classroom. In
fact, operating a computer can be simple compared
to designing a scientific experiment and solving
challenging math problems, that is, as long as
the student and the teacher have the tools. For
those who do not use the tools, technology can
become a diversion or a simple device for entertainment. Access to technological resources in the
classroom is only part of the solution (Feldman,
2001). According to Feldman, research leads us
to surmise that teachers who feel more prepared
to use technology are more likely to use it in instructional activities (Feldman, 2001). If we want
to help students from low SES backgrounds, our
emphasis should be on providing nourishing support, specifically, providing well-trained teachers
and a rigorous curriculum that integrates the use
of technology.

Implementing Technology in Urban

Schools
One of the major concerns for urban teachers when
integrating educational technology in the classroom is the identification of appropriate principles
that help achieve high student learning outcomes.
Recent research synthesis efforts by the Center
for Applied Research in Educational Technology
(2006), the International Society for Technology

Re-Examining the Socioeconomic Factors Affecting Technology Use in Mathematics Classroom

in Education (2003), and Roblyer (2006) provide

principles for appropriate ways to use education
technology in urban schools as supported by the
education technology research.
Resent research in educational technology
(Boster, Meyer, Roberto, & Inge, 2002; Bracewell
& Laferriere, 1996; Coley, Cradler, & Engel, 1997;
Cradler & Cradler, 1999; Koedinger & Anderson,
1999; Kulik, 2003; White & Frederiksen, 1998)
has shown that the effective use of educational
technology occurs when the application directly
(a) supports the curriculum objectives being assessed; (b) provides opportunities for student col
laboration and project/inquiry based learning; (c)
adjusts for student ability and prior experience, and
provides feedback to the student and teacher about
student performance; (d) is integrated throughout
the lesson; (e) provides opportunities for students
to design and implement projects that extend the
curriculum content being assessed; and (f) is used
in environments where the organization leaderships supports technological innovation.
Some examples of the strategies that have
proved successful in influencing student academic
performance include students working in pairs
on lessons at the computer assisted instruction
through social interactions and teamwork (Bracewell & Laferriere, 1996); digital video clips, audio,
and graphics to supplement instruction (Boster,
Meyer, Roberto & Inge, 2002); mathematics curricula focusing on mathematical analysis of realworld situations supported by computer assisted
instructional software program (Koedinger &
Anderson, 1999); multimedia creation involving
research skills to locate content resources, capability to apply learning to real world situations,
organizational skills, and interest in content
(Cradler & Cradler, 1999); software programs that
allow students to be aware where they are in the
inquiry process and to reflect upon their own and
other students inquiries (White & Frederiksen,
1998); word processing software that utilizes
writing prompts (Kulik, 2003); and online feedback among peers who know one another allows

learners to feel more comfortable with and adept

at critiquing and editing written work (Coley,
Cradler, & Engel, 1997).

Conclusion
As educators and educational technology professionals demand for research-based evidence about
the effectiveness of specific instructional practices has created renewed interest in educational
research (Edyburn, Higgins, & Boone, 2005).
As a result, there is an urgent need for research
that provides evidence about the effectiveness
of various educational technology interventions
applied to specific subject domains. This is especially needed in mathematics, where we are still
trying to get a handle on the achievement gap of
students in urban and suburban communities.
Research is a critical piece to the puzzle to fully
understanding the impact of technology on student
achievement. As a result more research is needed
on a large scale that will focus and assess some
of the following questions not addressed in the
present study:

What instructional design models will help

instructors design effective strategies for
student learning success specifically in the
mathematics classroom?
What type of professional development
experiences do educators need in order to
design successful learning experiences for
urban students for mathematics?
How can educators locate innovative and
effective interventions for enabling urban
students to achieve high academic standards
through the use of educational technology
in mathematics?
What is the impact of race, ethnicity, culture, and language as it relates to the use of
technology in urban environments?
What works, for whom, and which conditions
does technology impact student achieve-

1479

Re-Examining the Socioeconomic Factors Affecting Technology Use in Mathematics Classroom

ment in urban environments relating to

mathematic instructions?
What new technologies are needed for
routine classroom activities for urban learners?
How can effective strategies be developed
to enhance the teaching and student learning process to reflect active learning in an
inquiry/project-based environment?

As we look to the future, technology is often viewed as an enticing means of closing the
achievement gap. It is seen as a magic bullet to
solve all instructional and learning related issues
in an educational environment. However, this
is not the reality. Statistics on the digital divide
have shown are that the use of technology is often
based on simple computer-to-student ratios that
have little relevance in describing the quality
of the technology experience of the use of the
intervention in the classroom. Recent advances
in educational technology have the potential to
significantly enhance the learning and achievement for all students in the urban environment.
However, these contributions hold for diverse urban learners suggests unlimited potential for their
application in urban schools. Finally, the current
accountability environment demands significant
attention to questions of efficacy, which must be
addressed in the context of using technology to
enhance student achievement.
The current literature has implies that innovative approaches used in teaching with technology
leaves students with a more effective learning
environment that promotes quality teaching and
active student learning. Consequently education
planners and policy makers must think beyond providing more hardware, software, and connecting
schools to the Internet, but instead thinking about
keeping urban schools and teachers well-informed
and trained in the effective use of technology for
educational purposes. One of these investments is
meaningless without the other. High-speed con-

1480

nections, complete digital services, and modern

computers are basic to every professional workplace and are essential to student learning in the
21st century. However, technology will fail to
meet its educational promise if we neglect to equip
teachers with the skills they need to understand
and use it and transmit this knowledge and skills
to the urban learner.
Adhering to these procedures, educators will
be able to grow as practitioners in the field and
use educational technology to support quality teaching and active student learning. The
research examined in this discussion provides
teachers with a relevant framework to understand
the factors that affect the urban learner and the
power of technology in the teaching and learning
process can offer the urban learner, thus leading
our educational system to fulfill the promise of
providing quality teaching and student learning
for a more consistent and dynamic educational
learning environment that continues to support
the ideals and concepts of the great American
education system.

References
Advanced Telecommunications in U.S. Public
Elementary and Secondary Schools, Fall 1996,
U.S. Department of Education, National Center for
Education Statistics, February 1997 (Heaviside,
Riggins and Farris, 1997).
American Council on Education (1999). To touch
the future: Transforming the way teachers are
taught. An action agenda for college and university presidents. Retrieved July 23, 2007, from
http://www.acenet.edu/resources/presnetiteachered-rpt. pdf
Becker, H. J. (2000). Whos wired and whos not:
Childrens access to and use of computer technology. The Future of Children and Computer
Technology, 2(10), 44-75.

Re-Examining the Socioeconomic Factors Affecting Technology Use in Mathematics Classroom

Becker, H. J. (1991). How computers are used

in United States schools: Basic data from 1989
LE.A. computers in education survey. Journal of
Educational Computing Research, 7, 385-406.

CEO Forum. (1997). From pillars to progress.

Washington, DC: Author. Retrieved July 23,
2007, from http://www.ceoforum.org/downloads/
97report.pdf

Boster, F. J., Meyer, G. S., Roberto, A. J., & lnge,

C. C. (2002). A report on the effect of the united
streaming application on educational performance. Farmville, VA: Longwood University.

Coley, Cradler, & Engel, P. (1997). Computers

and classrooms: The status of technology in
U.S. schools. Princeton, NJ: Educational Testing
Service, Policy Information Center.

Braak, I. V. (2001). Factors influencing the use of

computer mediated communication by teachers
in secondary shools. Computers & Education,
36(1), 41-57.
Bracewell, R., & Laferriere, T. (1996). The contribution of new technologies to learning and
teaching in elementary and secondary schools
(Documentary review). Retrieved July 23, 2007,
from http://wwwfse.ulaval.calfac/tactlfrlhtmlJapport/ impact96.html
Brainbridge, W. L., Lasley, T. J., & Sundre, S.
M. (2004). Policy initiatives to improve urban
schools: An agenda. Retrieved July 23, 2007,
from http://www.schoolmatch. comlarticleslSESjUNE03 htm
Burnett, G. (1994). Technology as a tool for urban classrooms. (ERIC Document Reproduction
Service ED 368 809).
Chen, L. I., & Thielemann, J. (2001). Understanding the Digital DivideThe increasing
technological disparity in America; Implications
for educators. In D. Willis, & Price, J. (Eds.),
Technology and teacher education annual2001
(pp. 2685-2690). Charlottesville, VA: Association
for Advancement of Computing in Education.
Chiero, R. I. (1997). Teachers perspectives on factors that affect computer use. Journal of Research
on Computing in Education, 30(2), 133-145.
Center for Applied Research in Educational
Technology (2006). How can technology influence student academic performance? Retrieved
march 18, 2006 from http://caret.iste.org/index.
cfm?fuseaction=evidence&answerID=1

Cradler, R., & Cradler (1999). Just in time: Technology innovation challenge grant year evaluation report for Blackfoot School District No. 55.
San Mateo, CA: Educational Support Systems.
Cuban, L. (2001). Oversold and underused:
Computers in the classroom. Cambridge, MA:
Harvard University Press.
Dividing Lines (2001). Technology counts 2001:
The new divides: Looking beneath the numbers to
reveal digital inequities. Retrieved July 23, 2007,
from http://counts.edweek.orglsreportsltc01ItcOlarticle.cfm ?slug=35divideintro.h2 O
Edyburn, D., Higgins, K., & Boone, R. (2005).
Handbook of special education technology research and practice. Whitefish Bay, WI: Knowledge by Design, Inc.
Fabry, D. L. & Higgs, I. R. (1997). Barriers to the
effective use of technology in education: Current status. Journal of Educational Computing
Research, 17(4), 385-395.
Feldman, S. (2001). The haves and have nots of
the digital divide. The School Administrator,
1(3), 1-4.
Finneran, K. (2000). Let them eat pixels. Issues
in Science and Technology, 1(3), 1-4
Frank, K. A., Zhao, Y., & Borman, K. (2004).
Social capital and the diffusion of innovations
within organizations: The case of computer
technology in schools. Sociology of Education,
77(2), 148-171.

1481

Re-Examining the Socioeconomic Factors Affecting Technology Use in Mathematics Classroom

Garofalo, J., Drier, H., Harper, S., Timmerman,

M. A., Shockey, T. (2000). Promoting appropriate uses of technology in mathematics teacher
preparation. Contemporary Issues in Technology
and Teacher Education, 1(1), 66-88.
Guttentag, S., & Eilers, S. (2004). Roofs or RAM?
Technology in urban schools. Retrieved July
23, 2007, from http://www.horizonmag.com/4/
roofram.htm
Huang, S.Y.L., & Waxman, H.C. (1996). Classroom observations of middle school students
technology use in mathematics. School Science
and Mathematics, 96(1), 28-34.
International Society for Technology in Education
(2004). Retrieved July 23, 2007, from http://www.
iste.org/standardsl
Johnson, R.S. (2002). Using data to close the
achievement gap: How to measure equity in our
schools. Thousand Oaks, CA: Corwin Press.
Internet Use in America. Full Report. Retrieved
March 10, 2006, from http://www.ntia.doc.gov/
ntiahome/digitaldivide/
Koedinger, K., & Anderson, J. (1999). PUMP
algebra project: AI and high school math. Pittsburgh, PA: Carnegie Mellon University, Human
Computer Interaction Institute. Retrieved July
23, 2007, from http://act.psy.cmu.edu/awpt/awpthome. html
Kulik, J. A. (2002). School mathematics and
science programs benefit from instructional
technology (InfoBrief, NSF 03-301). Washington,
DC : National Science Foundation. Retrieved
January 21, 2007, from http://dwbrr.unl.edu/iTech/
TEAC859/Read/KulikTech.pdf
Lonergan, J. M. (2001). Preparing urban teachers
to use technology for instruction. (ERIC Document Reproduction Service ED 460 190).
Mathematics Sciences Education Board. (1991).
Counting on you: Actions supporting mathematics

1482

teaching standards. Washington, DC: National

Academy Press.
Mathematical Association of America. (1991). A
call for change: Recommendations for the mathematical preparation of teachers of mathematics.
Washington, DC: Author.
Means, B., Penuel, W. R., &Padilla, C. (2001).
The connected school: Technology and learning
in high school. San Francisco: Jossey-Bass.
Mooij, T. & Smeet, E. (2001). Modeling and supporting ICT implementation in secondary schools.
Computers & Education, 36(3), 265-281.
National Center for Education Statistics. (2004).
Technology in schools: Suggestions, tools, and
guidelines for assessing technology in elementary
and secondary education. Retrieved July 23, 2007,
from http://nces.ed.govlpubs2003/tech_schools/
index.asp
National Council of Teachers of Mathematics.
(1989). Curriculum and evaluation standards
for school mathematics. Reston, VA: National
Council of Teachers of Mathematics.
National Council of Teachers of Mathematics.
(2000). Principles and standards for school
mathematics. Reston, VA: National Council of
Teachers of Mathematics.
National Telecommunication and Information Administration (1995). Falling through the net. Full
report. Retrieved July 23, 2007, from http://www.
ntia.doc.gov/ntiahome/fallingthru.html
National Telecommunication and Information
Administration (1997). Falling through the net
II: New data on the digital divide. Full report.
Retrieved July 23, 2007, from http://www.ntia.
doc.gov/ntiahome/net2/falling.html
National Telecommunication and Information
Administration (1999). Falling through the net
III: Defining the digital divide. Full report. Re-

Re-Examining the Socioeconomic Factors Affecting Technology Use in Mathematics Classroom

trieved July 23, 2007, from http://www.ntia.doc.

gov/ntiahome/digitaldivide/

classrooms across urban, suburban and rural high

schools. The High School Journal, 79(1) 41-48.

National Telecommunication and Information

Administration (2000). Falling through the net:
Toward digital inclusion. Full report. Retrieved
July 23, 2007, from http://www.ntia.doc.gov/ntiahome/digitaldivide/

Papert, S. (1980). Mindstorms: Children, computers, and powerful ideas. New York: Basic
Books.

National Telecommunication and Information

Administration (2002). A NATION ONLINE:
National Telecommunication and Information
Administration (2004). A nation online: Entering
the broadband age. Full report. Retrieved July
23, 2007, from http://www.ntia.doc.gov/ntiahome/
digitaldivide/
Owens, E.W. (1993). The quantity and quality of
technology use in urban, suburban, and rural
eighth grade mathematics classrooms. In D.
Carey, R. Carey, D. A. Willis, & J. Willis (Eds.),
Technology and teacher education annual (pp.
122-126). Charlottesville, VA: Association for the
Advancement of Computing in Education.
Owens, E.W. & Waxman, H. C. (1994). The
quantity and quality of technology use in urban,
suburban, and rural tenth grade mathematics
and science classrooms. In D. Carey, R. Carey,
D. A. Willis, & J. Willis (Eds.), Technology and
teacher education annual (pp. 147-154). Charlottesville, VA: Association for the Advancement of
Computing in Education.
Owens, E.W. & Waxman, H. C. (1995). Differences among urban, suburban, and rural schools
on technology access and use in eighth-grade
mathematics classrooms. Journal of Educational
Technology Systems, 24(1),83-92.
Owens, E.W. & Waxman, H. C. (1995). Investigating technology use in science and mathematics

Pinar, W. (2004). What is Curriculum Theory?

New York: Lawrence Erlbaum Associates, Inc.
Roblyer, M. D. (2006). Integrating educational
technology into teaching (4th ed.). Upper Saddle
River, New Jersey: Pearson Prentice Hall.
Solomon, L. & Wiederhorn, L. (2000). Progress
of technology in the schools: 1999 report On
27 states. Milken exchange on education and
technology. Santa Monica, CA: Milken Family
Foundation.
U.S. Department of Education.(2001). Enhancing education through technology act of 2001.
Sec. 2402. Purposes and Goals. Retrieved April
10, 2007, from http://www.ed.gov/legislation/
ESEA02/pg34.html
U.S. Congress, Office of Technology Assessment.
(1988). Power on: New tools for teaching and
learning. Washington, DC: U.S. Congress, Office
of Technology Assessment.
Waxman, H. C., Connell, M. L., & Gray, J. (2002).
A quantitative synthesis of recent research on the
effects of teaching and learning with technology on student outcomes. Naperville, IL : North
Central Regional Educational Laboratory. Retrieved August 21, 2006, from http://www.ncrel.
org/tech/effects/effects.pdf
White, B. Y. & Frederiksen, J. R. (1998). Inquiry,
modeling, and metacognition: Making science
accessible to all students. Cognition and Instruction, 16(1), 3-188.

This work was previously published in International Journal of Web-Based Learning and Teaching Technologies, Vol. 2, Issue
4, edited by L. Esnault, pp. 72-87, copyright 2007 by IGI Publishing, formerly known as Idea Group Publishing (an imprint
of IGI Global).

1483

1484

Chapter 4.5

Let Them Blog:

Using Weblogs to Advance Literacy in the

K-12 Classroom
David A. Huffaker
Northwestern University, USA

Abstract
This chapter introduces the use of blogs as an
educational technology in the K-12 classroom. It
argues that blogs can be used to promote verbal,
visual, and digital literacy through storytelling and
collaboration, and offers several examples of how
educators are already using blogs in school. This
chapter also reviews issues such as online privacy
and context-setting, and ends with recommendations for educators interested in implementing
blogs with current curricula.

INTRODUCTION
As Internet technologies continue to bloom,
understanding the behaviors of its users remain
paramount for educational settings. For teachers,
parents, school administrators, and policymakers,
learning what types of activities and applications
students are using on the Internet is only the
surface. Understanding how they are using these

applications can provide innovative strategies for

learning environments.
Previously, many scholars have explored how
Internet users communicate and present themselves online, using computer-mediated communication venues such as e-mail, chat rooms, instant
messaging, newsgroups, multi-user domain
(MUDs), and personal home pages to examine
communication patterns, online identity construction, and even gender differences (Crystal, 2001;
Dring, 2002; Greenfield & Subrahmanyam,
2003; Herring, 2000; Lee, 2003; Turkle, 1995;
Witmer & Katzman, 1997).
Internet technologies continue to evolve, and
it is important for scholars to examine the latest
CMC arenas in comparison with past research in
hopes of finding new ways to find creative learning solutions and enhance pedagogical method
in educational technology. Weblogs, commonly
referred to as blogs, represent one of the latest
advances in CMC.
A blog can be simply defined as an online
journal. Made up of reversed chronological entries

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Let Them Blog

infused with text, images, or multimedia, blogs

embody a place where individual expression and
online community development coexist. Not only
do the authors, or bloggers, post thoughts and
feelings on a Web page for the world to view,
but blog readers can comment, creating a dialogue between the blogger and the community
he inhabits. Furthermore, bloggers link to other
bloggers, creating an interwoven and perhaps
interdependent online community of writers and
readers. Blog popularity continues to resonate
throughout the media, with many scholars suggesting an evolution in individual self-expression,
education and research, online journalism, and
knowledge management (Alterman, 2003; Blood,
2003; Herring, Scheidt, Bonus, & Wright, 2004;
Lasica, 2003; Moore, 2003; Mortenson & Walker,
2002; Oravec, 2002; Pollard, 2003b; Schroeder,
2003).
In a recent survey, Perseus Development
Corporation found that among the four million
published Weblogs, almost 53% are created by
children and adolescents between ages 13-19
(Henning, 2003). With such a strong population of
young bloggers, understanding its potential uses
within a classroom remains an exciting prospect
for educators and parents. Can blogs be used to
enhance learning? In what ways can they be used
in the classroom?
This chapter hypothesizes that blogs can be
effective educational tools in the following ways:
(1) they promote verbal and visual literacy through
dialogue and storytelling, (2) they allow opportunities for collaborative learning, and (3) they are
accessible and equitable to a variety of age groups
and developmental stages in education.
In order to evaluate this hypothesis, this chapter
will proceed as follows. First, it will provide a
thorough explanation of what blogs are, fundamental blog features, how they are used, and the
demographics of the blog population. Second,
this chapter will define verbal, visual, and digital
literacy, and their importance in learning. Third,
it will explain how blogs foster literacy through

storytelling and peer collaboration. Fourth, this

chapter will describe examples where blogs are
used in K-12 classrooms, with an emphasis on the
previous concepts of storytelling, peer collaboration, and literacy. Finally, this chapter will provide
specific recommendations for educators and
school administrators interested in implementing
blogs in their schools and classrooms.

WHAT IS A BLOG?
Blogs are personal journals written as a reversed
chronological chain of text, images, or multimedia, which can be viewed in a Web page and are
made publicly accessible on the Web (Huffaker,
2004a; Winer, 2003). As depicted in Figure 1,
blogs typically contain text in the form of a blog
post, offer the ability for readers to comment or
provide feedback, contain archives to past blog
posts, and link to other blogs and bloggers.1
Blogs are inherently different from personal
home Web pages. First, bloggers post entries
through manual software, such as a Web browser,
or automatic software, which is downloaded off
the Internet and used to instantly publish content
to the Web. Therefore, bloggers do not need to
understand HTML or other Web programming

Figure 1. An example blog

1485

Let Them Blog

languages to maintain their blogs. Second, the

resulting blog page resembles a personal diary with
entries sorted by time and date, a much stricter
format than personal Web pages of the past.

Blog Features

David Winer, a fellow at Harvard Law School,

considered one of the more visible writers and
developers of Weblogs, describes some of the
more important features of blogs:

Weblog Posts: Weblog posts include a

subject title and a body message. Posts can
be relatively short in length such as one or
two paragraphs, or they could be a long,
thoughtful expos, encompassing a dozen
paragraphs. They can comprise a variety
of media objects, including text, pictures,
graphics, or multimedia, and even popular
file formats such as Microsoft Office documents or Adobe PDFs (Winer, 2003). These
posts receive a timestamp to denote time
and date of the post. Figure 2 demonstrates
the format of a typical blog post.
Comments: Readers of a Weblog have
an opportunity to respond to a blog post
through a comment or feedback link. These
comments create a thread, as many readers
can comment on a single post (Winer, 2003).
They contain a timestamp and are viewable
to the public. Bloggers can also respond back

Figure 2. Blog post with comment link

1486

to reader comments. Figure 2 highlights the

comment section. Blog readers can click
Post a Comment to add a new comment
or 1 Comment to read previous ones.
Archives: The front page of a blog contains
only a certain amount of posts, sometimes
two or three and sometimes twenty. For
authors who have maintained their blogs for
longer periods of time, they can store past
blog posts in an accessible, often searchable
archive. As depicted in Figure 3, archives
can be organized by month, by week, or
even by number of posts.
Templates: Another useful feature for Web
authors is a set of presentation tools that
allow pages to be built from preexisting
templates. Blog authors can choose from
a variety of graphical layouts, typography,
and color schemes. This allows complete
customization and a feeling of personalization for bloggers without any sophisticated
technical expertise (Winer, 2003). However,
some bloggers like to tinker with a Web
programming language, to add third-party
applications or other bells-and-whistles to
their blogs, and the capability to reprogram
a blog is available. Figure 4 represents the
types of templates Blogger.com, one of the
most popular blog-hosted sites, offers.

Let Them Blog

Figure 3. An example of blog archives

What is the Blogosphere?

Although some studies suggest the majority of
blogs are still highly personalized venues for
self-expression (Herring et al., 2004), many
blogs contain links to other bloggers, creating an
online community that is often referred to as the
blogosphere. These blog communities typically
share a common purpose or responsibility (Carl,
2003), for example, a group of friends spread
out across the world may use blogs as a means
to communicate with each other, or a support
group might encourage each others therapeutic
development, or a group of amateur journalists
might be spreading news on situations of political
strife or violent conflict in an area that a global
news agency may be unaware of.

The blogosphere also represents the total

population of bloggers, an Internet culture that is
continuing to grow. Several reports and Web sites
that gather information on blog statistics, which
estimate between two and four million blogs,
demonstrate the vastness of the blogosphere; many
expect this growth to continue (Henning, 2003;
Herring et al., 2004; Kumar, Novak, Raghavan,
& Tomkins, 2003). Certainly, the acquisition of
population blog software or services within online commercial giants such as AOL, Yahoo, and
Google suggest that the world may be on the cusp
of a blog saturation in Internet culture (Munarriz,
2003; Olsen, 2004; Shirpy, 2003).

Figure 4. Example of blog templates

1487

Let Them Blog

Size of the Blogosphere

Blogs continue to be created and abandoned each
day, so the exact population figures on the size of
the blog population remains in transit. The population size is captured using software that indexes
blogs or comes directly from sites that host blogs
such as LiveJournal or Blogger. It must also be
considered how many blogs are active or updated
within the last three months, because many blogs
become abandoned (Henning, 2003).

1488

Blogcount (http://dijest.com/bc/), a Web

site dedicated to understanding how vast
the Weblog community is, collects scholarly
and industrial reports that discuss technical
and demographic issues, from frequency of
blog posts to mapping the blog community.
By June 2003, Blogcount estimated there
were 2.4 to 2.9 million active Weblogs
(Greenspan, 2003).
BlogCensus (http://www.blogcensus. net/)
uses a software program to search and locate
Weblogs and categorize them by language
and authoring tool. As of April 2004, the
BlogCensus2 has indexed 1.83 million blogs
and estimates at least 1.21 million of these
are active.
LiveJournal (http://www.livejournal. com/)
publishes daily statistics direct from its
servers, citing that there are over 2.9 million
livejournals, with 1.4 million active, as of
April 2004 (specific data can be found at
http://www.livejournal.com/stats/stats.txt).
Perseus Development Corporation, which
offers enterprise-level surveys regarding
software and technology, recently published
a white paper that estimates there are 4.12
million blogs, but at least 2.7 million have
been temporarily or permanently abandoned
(Henning, 2003).

Age and Gender Demographics of the

Blogosphere
Most of these surveys suggest that a significant
portion of the total blog population is made up of
teenagers, and almost evenly partitioned between
genders. The largest age distribution of bloggers
typically ranges between 13 and 20, which assumes most bloggers are either in early secondary
school or beginning college or university. There
are some subtle discrepancies in gender within
the studies listed below; however, the margins are
not far enough apart to suggest severe differences
in gender use of blogs.

Perseus Development Corporation, for

instance, finds blogs are dominated by the
youth, with 51.5% of all blogs being developed and maintained by ages 13-19. They
also find 56% of the total bloggers are female
and 44% are male (Henning, 2003).
A recent academic study of 203 randomly
selected Weblogs revealed 54.2% male authors and 45.8% female authors, as well as
40.4% of blog authors being under age 20
(Herring et al., 2004).
Another academic study of 358 randomly
selected blogs found 52% male and 48%
female bloggers, and 39% of bloggers are
less than 20 years old. However, they also
found there are more females than males
in the teen category (Herring, Kouper,
Scheidt, & Wright, 2004).
BlogCensus randomly sampled 490,000
blogs to find 39.8% male and 36.3% female,
with the rest of the blogs unidentifiable in
terms of gender (see http://www.blogcensus.
net/weblog/).
Finally, Jupiter Research found that blogging
is split evenly among genders (see http://
www.jup.com/bin/home.pl) (Greenspan,
2003).

Let Them Blog

Use of Blogs
Understanding the features of the blogs helps
distinguish them from other Internet applications,
and grasping the size of the blogosphere signifies
the popularity of blogs in Internet culture. The
next question involves the content of blogs. What
are bloggers writing about? The answer not only
provides a context for online community interaction, but possible application for educational
technology. These can be divided into five areas:
(a) personal blogs, (b) community blogs; (c) journalism blogs; (d) education and research blogs;
and (e) knowledge blogs.

Personal Blogs
The most popular use of blogs are similar to personal Web sites authored by individuals, which
include chronological posts as well as links to
other Web sites or Weblogs (Lamshed, Berry,
& Armstrong, 2002). Despite the popular notion
that Weblogs lean toward external events, or
remain highly interlinked with the blogosphere,
the majority of Weblogs are still individualistic
self-expressions written by one author (Herring
et al., 2004).

Community Blogs
Virtual communities develop through the use of
a blog (Lamshed et al., 2002). Examples might
include a community support group, a site for
parents to ask questions and exchange answers, a
research community sharing resources and data,
or a mirror of an offline community, like a softball team or neighborhood newsletter. Although
personal blogs may dominate the blogosphere,
the ability for individuals to choose their level of
community participation may be another reason
for blog popularity, as it allows the blog author to
explore individual needs while benefiting from
community interactions (Asyikin, 2003). The
linkages with other Web sites, people, and ideas

even form micro-communities with varying levels

of involvement.

Journalism Blogs
The idea of alternative forms of journalism manifesting through Weblogs has received increasing
attention in media and scholarship (Alterman,
2003; Blood, 2003; Lasica, 2003). Where is Raed?
(http://dear_raed.blogspot. com), for instance, is
a blog by an Iraqi that discusses what is happening in Iraq since September 2003. He discloses
a candid view of the U.S. occupation, but also
introduces readers to fellow Iraqi bloggers. For
most, the global news agency is the only link to
international affairshaving personal, subjective
commentary within a foreign world provides a
unique view to outsiders.
A different, but equally unique log is J-Log,
which provides community critiques and commentary on current journalism and news. The
community not only shares interesting news
items, but also poses questions such as, Is this
news fair and balanced? Perhaps J-Log (http://
www.mallasch.com/journalism/) and individual
reports such as Raed demonstrate new forms of
online journalism; critiques, however, as to the
viability of these news sources remain an issue,
including the resources and references and even
the subjectivity amidst objective journalistic
philosophy.
A link to http://blogdex.net/, an MIT Laboratory experiment that captures the fastest-spreading
ideas in the blog community, typically results in
news headlines as the most contagious information.

Education and Research Blogs

Blogs have been heralded as opportunities to promote literacy in learning by allowing students to
publish their own writing, whether it is a journal
or story, or even comments on class readings
(Kennedy, 2003). For more advanced students,

1489

Let Them Blog

blogs present the same opportunities: writing

and thinking with Weblogs, archiving and analyzing past knowledge, and developing a social
network to collaborate and critique (Mortenson
& Walker, 2002).
Blogs allow educators and students to interact
in the same common space and format (Wrede,
2003). Lamshed et al. (2002) find that students
believe blogs are easy to use and navigate, and
enthusiastic about learning features such as storing and managing information, communicating,
reviewing work before posting, and keeping on
track, which refers to managing time or monitoring progress (Lamshed et al., 2002). If students
are eager to adopt blogs in the classroom, teachers
may have an opportunity to also keep on track
with students, as communication and interactions
are visible and accessible anytime-anywhere.
Several sites explore the use of blogs in education. Weblogg-Ed (http://www.weblogg-ed.com/),
maintained by Will Richardson, collects ideas
about Weblogging in school settings and facilitates
dialogue between teachers. Similarly, Edublog
(http://edublog.com/) is an initiative to develop
and study blog-based tools for the classroom.

Knowledge Blogs
Similar to education and research, blogs provide
opportunities for organizations to manage and
share content, as well as communicate across the
network. Dave Pollard, Chief Knowledge Officer
at Ernst and Young, Inc. and popular writer on the
role of blogs in the business, suggests that blogs
can be used to store and codify knowledge into a
virtual file cabinet. But unlike other content management tools, blogs allow authors to publish in a
personal and voluntary way, creating a democratic,
peer-to-peer network (Pollard, 2003b). Pollard also
suggests companies can increase profitability by
designing information architecture to embrace the
use of Weblogs (Pollard, 2003a, 2003c).

1490

VERBAL, VISUAL, AND

DIGITAL LITERACY
Literacy has always been a focus of learning,
especially considering that the foundations of
education are grounded in reading and writing.
In fact, reading and writing, often referred to as
verbal literacy, serves as a benchmark for success
in education. Reading and writing are not only
important in language arts or humanities; they
serve as prerequisites for all academic disciplines,
including science and mathematics (Cassell,
2004). Verbal literacy is developed even before
children enter school as parents read stories to
their children, helping them to understand the
relationship between words and pictures, as well as
helping to develop narrative structure (Bransford,
Brown, & Cocking, 1999; Huffaker, 2004b).
Scholars have recognized how communication
also takes place in the form of images and symbols,
what is referred to as visual literacy (Gee, 2003).
As James Paul Glee (2003) suggests, texts are
becoming increasingly multimodal, containing
both text and images, and can be recognized in
everything from the advertisement to the high
school textbook (Gee, 2003). Our society is filled
with these multimodal symbols, and their coherence is intrinsic for operating in the modern world.
Therefore, reading and writing should not only
include words but also images, and the development of both verbal and visual literacy is essential
for success inside and outside school walls.
The use of technology represents a third type
of literacy, equally ubiquitous and important
as the other forms. Digital literacy, sometimes
referred to as technological fluency, embodies
the idea of using technology comfortably, as one
would a natural language (Cavallo, 2000). As
users of computers and other digital technology
become more fluent, they learn to communicate
and express themselves explicitly and eloquently
using these tools; in effect, technology becomes
innate. Furthermore, just as reading and writing
are widely hailed as the building blocks for success

Let Them Blog

in society, digital literacy becomes necessary for

success in the technological world we inhabit.
This is what makes educational technology so
excitingit encourages and advances all three
types of literacy: verbal; visual; and digital.
Blogs, for example, utilize both textual reading
and writing, but also involve the use of graphics in
the forms of emoticons, or graphical expressions
of emotions, images, and multimedia. Blogs also
encourage digital literacy, which is grasped from
navigating a graphical user interface, and using
computers and the Internet in order to publish
content.
Some CMC contexts such as e-mail and instant
messaging often utilize short pieces of dialog,
informal language, and even altered words from
spoken language (Crystal, 2001). Educators might
complain that using these forms in a classroom
might reduce literacy skills, as language development is often paired with precision and formality.
Blogs might combat this issue, as its medium
involves longer written passages and can be
contextualized by educators to promote formal
language skills.

STORYTELLING AND COLLABORATION

Storytelling is a natural part of adolescent development, and children understand the fundamentals
of storytelling in their first three years of life
(Bransford et al., 1999). Even as a baby, a parent
introduces the child to storytelling via bedtime
readings or oral tales. By the time a child reaches
age four, he can recall many types of stories,
whether fictional or autobiographical (Bransford
et al., 1999). The stories develop into more mature
narrative as the child gets older, thus storytelling
provides a way for childrens language and reading
skills to develop (Bransford et al., 1999).
Children advance literacy skills through the
practice of telling stories to adults and peers
alike (Ryokai, Vaucelle, & Cassell, 2003). With

adults, children advance language skills because

their partners have even more advanced language
skills, and children will adapt (Ryokai et al., 2003;
Vygotsky, 1980). With peers of similar age groups,
children feel more comfortable to collaborate,
and enjoy learning together and building on each
others knowledge (Ryokai et al., 2003). Because
peer relations remain quite important to children (Bullock, 1998), effective group dynamics
and learning also results in pro-social behavior
through a sense of ownership, compromise, and
shared responsibility (Calvert, 1999).
Storytelling does not end in childhood, but
continues throughout adolescence and even adulthood. Stories help children and adults alike share
experiences and feelings in an engaging and even
entertaining way (Denning, 2001). From simple
sandbox sketches to dinner party yarns, the importance of stories as a catalyst for conversation
and dialogue is clear. Furthermore, storytelling
helps create connections between people, to engage people, to captivate them. This captivation
occurs on both endsit is just as fun for the teller
of the story as the listener. Storytelling fused with
peer collaboration is an excellent way to improve
language abilities and advance literacy for children
and adolescents (Ryokai et al., 2003).
Blogs have the potential to foster both storytelling and collaboration. First, blogs serve as a
natural venue for personalized self-expressions.
Like diary entries, blogs take form as stories, anecdotes, or vignettes, similar to the types of oral
and written stories people encounter everyday.
This includes past and present activities, feelings
about oneself or other people, or even hyperbolic
or fictitious tales.
Secondly, blogs have technical features that
offer the potential to create a dialogue between
blog author and blog reader, whether it is a reader
response, a question-and-answer sequence, or
even general brainstorming. In some cases, blog
authors discuss candid feelings and readers respond with encouraging statements, providing
therapeutic connections. These comments form

1491

Let Them Blog

a chain between the author and readers, and in

essence, an online community. Communities are
also built as bloggers link to each other, creating
a group of storytellers that provide individualistic
expressions, as well as interactions with each
other.

EXAMPLES OF BLOGS
IN PRACTICE
Blogs are just beginning to infiltrate classrooms,
as educators and school administrators consider
blogs as a useful tool for communicating between
teachers, schools, students, and parents, and as a
way to showcase the work of students (Richardson,
2004). These practices are celebrated on the Internet through communities of educators interested
in blogging and education. Will Richardsons
Weblogg-ed: Using Weblogs and RSS in Education Web site (http://www.weblogg-ed.com/), for
instance, is a useful source of information. His
site focuses on best practices, offers a quick start
guide for teachers, and links to other bloggers
concerned with blogs in education. Edblogger
Praxis (http://educational.blogs. com/edbloggerpraxis/) is another important Web site which
unites educators who blog about their experiences
or pedagogical philosophies.
This section will look at examples of blogs in
practice, separating them into high school (grades
9-12), elementary and middle schools (grades 48), and primary school (grades K-3) in order to
contextualize blog use by age and developmental
stage, and to provide useful models for educators
and school administrators interested in viewing
how blogs work at different levels of the school
system.

High School Blogs

Will Richardson is a teacher and Supervisor of
Instructional Technology and Communications
at Hunterdon Central Regional High School in

1492

Flemington, New Jersey. He not only initiates

school-wide policies for blog use in the classroom,
but has found success in using blogs for his own
literature and journalism classes. In the literature
class, students use blogs to comment and critique
on reading assignments. In the journalism class,
the students collect news stories to post in their
blogs; they also comment on each others stories
while serving in an editorial role. Therefore, in
language arts classes, blogs can be used to generate a discussion using critical analysis, or as a
collaborative tool where students comment and
edit each others work. For Richardson, blogs help
students become more aware of their writing, as
well as their audience (Kennedy, 2003). Figure
5 represents how blogs can be contextualized to
provide assignments for students and allow sections for critique and reflection. Each student has
an individual blog, which is linked to a homepage
where the teacher assigns work and posts relevant
news and other information.
Blogs also help create a community of practice
among participating students. They can collaborate with each other and build knowledge. These
types of discussions, where ideas are synthesized
and news ideas created, may be intrinsic to
building critical thinking skills. They may also
feel that they are part of a team, and that each
individual has a responsibility to contribute in
order to achieve success for the group. Again, effective group behavior involves shared ownership,
compromise, and responsibility, which creates
pro-social conduct (Calvert, 1999).
Blogs are accessible to the general public, and
Richardson found an important side effect as students began blogging. When the class discussed
a new novel, the author of the book accessed the
blog and posted an impressive response to the
questions students were asking (Richardson,
2004). This may have a profound affect on how
students view their own work. For instance, this
may give students the impression that their ideas
and discussion are important in the real world,
providing a sense of empowerment and self-ef-

Let Them Blog

Figure 5. Example of high school blog

ficacy. Literature and journalism classes and

experts are not the only beneficiaries; political
leaders, scientists, artists, and philosophers can
also directly participate in an educational blog,
making students feel their academic work remains
valuable, an important consideration for motivating children to learn.

Elementary and
Middle School Blogs
Blogs can be utilized in many of the same ways
in other grade levels. The Institut St-Joseph in
Quebec City, Canada, for instance, uses blogs
among fifth and sixth graders in order to practice
reading and writing. Implemented by the school
principal, Mario Asselin, Institut St-Joseph
bloggers use a software program to write about
anything and everything that is school related.
Similar to Richardsons realization, the fact that
blog posts are being read by anyone in the world
has an acute effect on students. They felt empowered as their blogs received comments from total
strangers and even Canadian celebrities (Asselin,
2004). Similar to Will Richardsons work, Figure

6 portrays a homepage blog where projects are

assigned and school- or project-related links are
provided. Students can comment on homework
assignments to the entire class, and still post to
their own blog space.
At first, some parents and other readers complained that the language of the student blogs
was too informal to be considered good writing,
and even contained misspelling and grammar
errors. Critics complained that blogs were teaching improper language skills, so Asselin (2004)
discussed these challenges with the students. The
students came up with a system where each blog
post would be reviewed by students and a graphic
would be posted alongside the text, which served
as a stamp of approval. Suddenly, student writing improved dramatically, as no student wanted
mistakes after insisting their quality of writing
was excellent (Asselin, 2004).
Similarly, Anne Davis Newsquest involves
fourth through sixth graders who practice writing
by commenting on news and events. Students felt
that blogging helped them write better stories,
extend their vocabulary, and even feel grown
up knowing their voice is on the Internet. After
a period of working with blogs, these students

1493

Let Them Blog

began to collaborate with Will Richardsons high

school journalists, forming the Georgia-NJ Connection blog. Now, older journalists and younger
journalists can share ideas, comments, and critiques. Again, this resonates with Vyogtskys
notions that children will adapt to more advanced
language skills observed in adults (Ryokai et al.,
2003; Vygotsky, 1980). Figure 7 demonstrates how
blog information can be scaled down to reach a
younger audience.

Primary School
Mrs. Dudiak, who teaches second grade at Oakdale
Elementary School in Maryland, uses blogs to
create writing assignments for her students. She
might use a picture such as a waterfall and ask
students to write a description, a story, or poetry.
She also asks them to discuss favorite books, what
types of books and stories they like, as well as
depictions of books into a movie in our head;
students reply in the comment section of the blog.
This is an excellent example of how blogs can be
placed in specific contexts to meet the goals of
classroom curriculum. Figure 8 exemplifies how
blogs can be simplified to reach even the young-

est students. Changing colors, enlarging texts,

and contextualizing material to reach the needs
of primary students represent easy changes for
educators to make.
Lewis Elementary School in Portland, Oregon,
uses a blog to showcase K-5 student work, as
well as to post important information and news
for parents. For instance, photo galleries of artwork are displayed alongside weekly classroom
notes from each teacher. This demonstrates how
blogs cannot only be effective in practicing and
advancing writing and language skills, but also
in communicating between teachers, students,
and parents. Blogs may give parents more direct
exposure to their childrens school life, and thus
more opportunities to have an impact on their
childrens learning. Little Miami Schools (http://
www.littlemiami schools.com) in Morrow, Ohio,
also uses a blog to communicate school goals and
news, as well as spotlighting the successes of its
students and teachers.
In sum, blogs can be used at a variety of age
and grade levels to promote reading and writing skills, as well as to provide communication
links between students, parents, teachers, and
schools. Students can use blogs on their own,

Figure 6. Example of middle school blog (http://cyberportfolio.ixmedia.com/carriere/)

1494

Let Them Blog

Figure 7. Example of elementary school blog (http://www.schoolblogs.com/NewsQuest/)

finding stories or news they find interesting, as

demonstrated with the classes of Mr. Richardson
and Mr. Asselin, or they can have specific assignments designed by the teacher, as seen in the case
of Mrs. Dudiaks class. Furthermore, schools can
use blogs to bridge the links between students,
teachers, and parents, making demonstration of
work and student progress available, as well as
school news and schedules. Figure 9 demonstrates
how blogs can be used as a central communication
hub for teachers, schools, parents, and students
to share information.
The next section will address some of the
issues and considerations for educators interested in implementing blogs in their schools and
classrooms.

ISSUES AND CONSIDERATIONS

School- and District-Wide
Considerations
This chapter has mainly approached blog implementation from the educator perspective. How-

ever, it is important to examine the schools and

districts that encompass the individual classroom.
Not only do school and district administrators
remain an intrinsic part to choosing educational
technology initiatives, but their choices can
have a positive or negative effect on how these
choices become applied. For instance, schools
and districts have to work within a budget when
making choices for technology. Second, they have
to choose which hardware and software standards
will be equitable across the community. Finally,
they have to implement technology within certain
ethical, legal, and security issues.
For each of these considerations, blogs still
stand as a viable solution. First, most blog software
is free and open-sourced, so administrators can
host blogs within local technology infrastructure. Second, blogs are delivered via the Web, so
standards have already been establishedblogs
can be implemented anywhere a computer and
Internet accessibility is present. Finally, blogs
have the option of being available to the public or
of being private, an important consideration that
will be discussed in the next section.

1495

Let Them Blog

Figure 8. Example of primary school blog (http://mrsd.tblog.com/)

Figure 9. Example of blog as communication hub for teachers, schools, and parents (http://lewiselementary.org/)

In sum, the importance of school- and district-wide influence on educational technology

implementation cannot be ignored. However, for
the same reasons blogs can be a useful tool and
practical solution for educators; administrators
should also be impressed.

Issues of Online Privacy

A key concern for any educator and school system
involves student privacy. Cyberstalking or sexual
predation is a serious concern for adolescents using

1496

the Internet (Earley, 1999; Gilbert, 1995). Many

of the blog hosts offer free Web space available to
the public on the Internet; information on a blog
is publicly accessible. If the content of blogs or
identities of students need to remain anonymous
or private to the world, blog software and hosts
offer other options.
For instance, free blog-hosts such as LiveJournal offer a Friends Only option, which involves
a username and password for entrance. For even
more security, blog software such as MoveableType, which can be implemented on local school

Let Them Blog

Web servers, can remain completely in-house, that

is, no one outside of school or classroom computers would even have access to the blogs.
While privacy is an important consideration for
educators and school administrators, the counterargument to keeping blogs private would be that
making blogs available to the world might generate
feedback that will empower bloggers and further
develop communication and collaboration. For
example, if a classroom is discussing a science
project and a professional scientist participates
in the discussion, it would add significant value
to the learning experience.

Setting Context
One of the primary challenges for using blogs,
as with many technologies, in the classroom is
the importance of setting a context for learning
to occur. While discovery and creativity does
abound when adolescents are allowed freedom to
explore these CMC venues, some structure needs
to be in place to facilitate a learning outcome.
For instance, if writing quality is a concern for
a teacher, then contextualizing the language to
focus on clear and succinct writing skills is a
must. Similarly, students have to be encouraged
to use the blog on a steady basis and focus on the
classroom material.
In sum, contextualizing blogs to the learning
experience will serve to produce more exciting
and educational blog experiences in the classroom, experiences that parents, administrators,
and students alike can observe and reflect upon.
The next section will provide some recommendations for educators interested in implementing
blogs in the classroom to promote literacy and
learning.

RECOMMENDATIONS
Use free blog software for easy implementation
in the classroom. There are several popular Web

sites that provide templates, maintenance, and

hosting of blogs. Using free blog software makes
it an easier decision for educators and school
administrators concerned with school budgets
and technology requirements. With a computer
and an Internet connection, blogs can be set up
in minutes. For educators who want to keep blogs
inaccessible to the public, most blog hosts offer
the opportunity to password-protect blogs. The
most popular blog-hosts are:

LiveJournal: http://www.livejournal.com
BlogSpot: http://www.blogspot.com
Xanga: http://www.xanga.com
MoveableType: http://www.moveable type.
org
T-Blog: http://www.tblog.com

For those who want to implement the blogs on

local school servers, there are open source applications such as MoveableType and phpBB which
can be altered and used in any expected capacity,
although they have technical requirements (documentation are available on these sites).
Open-source applications such as MoveableType and phpBB allow complete customization of
the interface and blog application. So educators
should feel that blogs can be completely tailored
to suit the needs of students and the affordances
and constraints of school district policy. The examples of blogs used in this chapter are presented
to demonstrate current applications and serve as a
baseline for new ideas and implementation. Educators should feel complete flexibility in redesigning
blog software for learning experiences.
Fuse emergent learning with curriculumbased assignments. If given the opportunity,
students might begin to navigate their own learning paths, a feature that is recommended by the
new science of learning as directed by the U.S.
National Research Council (Huffaker & Calvert,
2003). Surely, letting students do as they want
with blogs will reveal some interesting trends
that could be manipulated into important learning

1497

Let Them Blog

opportunities. However, it may be a better idea to

create a hybrid form where students construct their
own learning, as well as assignments that teachers
feel aid the curriculum. Therefore, students have
some flexibility in exploring the medium of blogs,
while teachers provide direction and structure in
the educational practice.
Encourage collaboration. Blogs have great
potential as a collaborative environment. As
students read and write, they may comment and
critique each others work. They may even edit
each others work. Assignments can be created
where students have to work as a team to complete the goals, such as a story train, or chain
of story parts devised by individual students, or
literature critiques of different parts of a book, or
even a mix of different perceptions on the same
experience.
Develop a system for rating student work
before it is published. One way to maintain
the caliber of writing that is expected in formal
writing is to create a way to proof student work
before it goes into the public. Most blogs contain
a preview area before publication takes place.
The teacher does not necessarily need to be the
final editorstudents can edit each others work
and provide the final stamp of approval before
the world sees the writing. This may be more
difficult at early ages, but the metaphor provides
accountability for students, which the Institut
St-Joseph found dramatically improved writing
quality (Asselin, 2004).
Create interactions between students and
the outside world. One way to make students feel
their work is valuable is to demonstrate how their
words are being read outside the classroom. For
instance, if parents or even total strangers provide
feedback on student work, children and adolescents might become more aware of their writing,
as well as their audience (Richardson, 2004).
Perhaps even more empowering are experts such
as authors, poets, artists, scientists, engineers, or
other professionals who may comment on student
work or add to the discussion. These linkages are

1498

not only useful for learning and achievement,

but are not difficult tasks to achieve when using
blogs, especially considering their accessibility
by the public sphere.
Link with other educational blogs. As exemplified in the Georgia-NJ Connection, students
using blogs may connect with other classrooms
despite geographic constraints, in order to expand
the community of practice. Even regardless of age,
students can share with other students, providing
an online learning community that may become
self-motivating and self-sustaining. Blogs provide
an easy way to do this with built-in options for
linking to other blog communities.
Get parents involved in reading and participating with blogs. As previously mentioned,
getting parents to provide feedback to student
work might be an excellent way for students to
feel their work is valuable and even appreciated.
Yet this has a two-fold benefitparents may feel
they have more exposure to, and thus more interaction with, their childrens learning. Therefore,
both students and parents benefit.
Furthermore, blogs can provide links between
schools, classrooms, teachers, students, and
parents. Parents who visit a classroom blog may
encounter the latest third-grade arts and crafts,
see the latest school news, or see what is on for
lunch that day. This helps bridge communication
between all parties responsible for educating
children and adolescents.
Creating linkages between all the classrooms
in the school creates an interwoven community
where everyone feels connected and accessible.
This community of practice may have a paramount
effect on increasing the efficiency and success of
learning in K-12 educational settings.

CONCLUSION
This chapter has sought to demonstrate how blogs
can be a useful tool for promoting literacy in the
K-12 classroom. Literacy most noticeably takes

Let Them Blog

form in reading and writing, but visual literacy as

well as digital literacy may be just as important
for success in education and beyond school walls
(Cassell, 2004; Cavallo, 2000; Gee, 2003). As an
Internet technology, blogs involve the reading
and writing of texts, but also the creation and
interpretation of images and multimedia, as well
as the need to navigate a graphical user interface,
all of which serve to advance skills in verbal,
visual, and digital literacy.
Storytelling, a natural part of childrens
language development, serves as a catalyst for
promoting literacy (Bransford et al., 1999; Huffaker, 2004b). Children and adults alike use stories
to express thoughts, feelings, and experiences
in an engaging and entertaining way (Denning,
2001), and storytelling helps create collaborations
between the teller and listener. Collaborations,
whether between children and adults or among
peers, are also important aspects of learning,
producing ways to adapt or scaffold learning
(Ryokai et al., 2003; Vygotsky, 1980). Blogs,
which resemble personal journals, create a perfect environment for sharing stories and other
forms of writings, and provide ways for others
to participate through feedback or critique on
student work.
Currently, blogs in practice highlight that
blogs can be used in a broad range of age groups
and developmental stages in education. Second
graders and high school teenagers alike can use
blogs to practice reading and writing, and help
develop language skills. Blogs can also be used in
a variety of disciplines, from literature and journalism to arts and sciences. Anywhere language
is used to promote discussion, blogs serve as an
arena where thoughts can be published and made
available to everyone for additional feedback. Not
only can students use the blogs, teachers, parents,
and experts can collaborate alongside them, which
may help develop a sense of empowerment and
self-efficacy for children and adolescents.
Finally, this chapter has recommended several
concepts for educators interested in implementing

blogs in the classroom. These concepts embrace

collaboration, online community building, interlinking parents with schools, student accountability, and ways to monitor student progress. Because
blogs are readily available for free at a selection
of blog hosts, implementation into the class does
not require a high level of technical expertise
or an extravagant school budget. Therefore, the
benefits of advancing language skills or bridging
communication between teachers, parents, and
schools far outweigh the costs.
Let them blog is a simple ideablogs provide a computer-mediated communication context
where students can practice and advance different types of literacy, use storytelling to express
themselves, and collaborate with peers and adults
as they complete assignments. Letting them blog
is even simplerblogs are easy to use, free, and
extremely accessible for anyone with a computer
and Internet access. Teachers may find blogs to
be one of the easiest educational technology applications to implement, and since adolescents are
already using blogs outside the classroom, it may
be an excellent time to connect the two.

ACKNOWLEDGMENTS
The research of this chapter was funded in part by
a grant to the Childrens Digital Media Center at
Georgetown University (http://cdmc.georgetown.
edu) by the National Science Foundation (Grant
#0126014). Special thanks to Sandra Calvert for
all her mentorship and support.

REFERENCES
Alterman, E. (2003). Determining the value of
blogs. Nieman Reports, 57(3).
Asselin, M. (2004). Weblogs at the Institut StJoseph. Proceedings of the International Confer-

1499

Let Them Blog

ence of Educational Multimedia, Quebec City,

Canada.
Asyikin, N. (2003). Blogging lifeAn inquiry
into the role of Weblogs in community building.
National University of Singapore.
Blood, R. (2003). Weblogs and journalism: Do
they connect? Nieman Reports, 57(3).
Bransford, J. D., Brown, A. L., & Cocking, R. R.
(Eds.). (1999). How people learn: Brain, mind,
experience, and school (expanded ed.). Washington, DC: National Academy Press.
Bullock, J. R. (1998). Loneliness in young children. ERIC Digest. Availble from http://eric.
ed.gov/ERICDocs/data/ericdocs2/content_storage_01/0000000b/80/2a/2b/28.pdf
Calvert, S. L. (1999). Childrens journeys through
the information age. Boston: McGraw-Hill College.
Carl, C. (2003). Bloggers and their blogs: A depiction of the users and usage of Weblogs on the
World Wide Web. Washington, DC: Georgetown
University.
Cassell, J. (2004). Towards a model of technology
and literacy development: Story listening systems.
Journal of Applied Developmental Psychology,
25, 75-105.
Cavallo, D. (2000). Emergent design and learning
environments: Building on indigenous knowledge.
IBM Systems Journal, 39(3&4), 768-779.
Crystal, D. (2001). Language and the Internet.
Cambridge: Cambridge University Press.
Denning, S. (2001). The springboard: How storytelling ignites action in knowledge-era organizations. Boston: Butterworth Heinemann.
Dring, N. (2002). Personal home pages on the
Web: A review of research. Journal of ComputerMediated Communication, 7(3).

1500

Earley, M. (1999). 1999 report on cyberstalking:

A new challenge for law enforcement and industry. Washington, DC: United States Department
of Justice.
Gee, J. P. (2003). What video games have to
teach us about learning and literacy. New York:
Palgrave Macmillan.
Gilbert, P. (1995). On space, sex and stalkers.
Women and Performance, 17.
Greenfield, P. M., & Subrahmanyam, K. (2003).
Online discourse in a teen chatroom: New
codes and new modes of coherence in a visual
medium. Applied Developmental Psychology,
24, 713-738.
Greenspan, R. (2003). Blogging by the numbers.
CyberAtlas, (June 23).
Henning, J. (2003). The blogging iceberg: Of
4.12 million Weblogs, most little seen and quickly
abandoned. Braintree, MA: Perseus Development
Corporation.
Herring, S. C. (2000). Gender differences in CMC:
Findings and implications. Computer Professionals for Social Responsibility Journal, 18(1).
Herring, S. C., Kouper, I., Scheidt, L. A., &
Wright, E. L. (2004). Women and children last:
The discursive construction of Weblogs. In L.
Gurak, S. Antonijevic, L. Johnson, C. Ratliff, &
J. Reyman (Eds.), Into the blogosphere: Rhetoric,
community, and culture of Weblogs. Los Alamitos,
CA: IEEE Press.
Herring, S. C., Scheidt, L. A., Bonus, S., & Wright,
E. (2004). Bridging the gap: A genre analysis of
Weblogs. Proceedings of the 37th Hawaii International Conference on System Sciences.
Huffaker, D. (2004a). Gender similarities and
differences in online identity and language use
among teenager bloggers. Washington, DC:
Georgetown University.

Let Them Blog

Huffaker, D. (2004b). Spinning yarns around

a digital fire: Storytelling and dialogue among
youth on the Internet. Information Technology
in Childhood Education Annual, 1.
Huffaker, D. A., & Calvert, S. L. (2003). The new
science of learning: Active learning, metacognition, and transfer of knowledge in e-learning
applications. Journal of Educational Computing
Research, 29(3), 325-334.
Kennedy, K. (2003). Writing with Weblogs.
Technology and Learning Magazine, 23(February 13).
Kumar, R., Novak, J., Raghavan, P., & Tomkins,
A. (2003). On the bursty evolution of blogspace.
Proceedings of the 12th Annual World Wide Web
Conference, Budapest, Hungary.
Lamshed, R., Berry, M., & Armstrong, L. (2002).
Blogs: Personal e-learning spaces. Australia:
Binary Blue.
Lasica, J. D. (2003). Blogging as a form of journalism. In R. Blood (Ed.), Weve got blog: How
Weblogs are changing our culture (pp. 163-170).
Cambridge: Perseus Publishing.
Lee, C. (2003). How does instant messaging affect
interaction between the genders? The Mercury
Project for Instant Messaging Studies at Stanford
University, USA.
Moore, C. (2003). Blogs refine enterprise focus.
InfoWorld, (January 10).
Mortenson, T., & Walker, J. (2002). Blogging
thoughts: Personal publication as an online research tool. In A. Morrison (Ed.), Researching
ICTs in context (pp. 249-279). Oslo: Intermedia
Report.
Munarriz, R. A. (2003). Googles big day. Fool.
com, (September 11).
Olsen, S. (2004). Yahoo tests syndication technology for MyYahoo. CNET News.com, (January
12).

Oravec, J. A. (2002). Bookmarking the world:

Weblog applications in education. Journal of
Adolescent & Adult Literacy, 45(7).
Pollard, D. (2003a). Blogs in business: Finding the
right niche. Retrieved January 30, 2004.
Pollard, D. (2003b). Blogs in business: The Weblog
as filing cabinet. Retrieved January 27, 2004.
Pollard, D. (2003c). A Weblog-based content
architecture for business. Retrieved January 27,
2004.
Richardson, W. (2004). Blogging and RSSthe
whats it? and how to of powerful new Web
tools for educators. Information Today, Inc., 11
(January/February).
Ryokai, K., Vaucelle, C., & Cassell, J. (2003).
Virtual peers as partners in storytelling and
literacy learning. Journal of Computer Assisted
Learning, 19(2), 195-208.
Schroeder, R. (2003). Blogging online learning
news and research. Journal of Asynchronous
Learning Networks, 7(2), 56-60.
Shirpy, C. (2003). AOL, Weblogs, and community.
Corante, (July 5).
Turkle, S. (1995). Life on the screen. New York:
Simon and Schuster.
Vygotsky, L.S. (1980). Mind in society: The
development of higher psychological processes.
Cambridge, MA: Harvard University Press.
Winer, D. (2003). What makes a Weblog a Weblog?
Cambridge, MA: Harvard Law School.
Witmer, D. F., & Katzman, S.L. (1997). On-line
smiles: Does gender make a difference in the use
of graphic accents? Journal of Computer-Mediated Communication, 2(4).
Wrede, O. (2003). Weblogs and discourse. Proceedings of the Blogtalk Conference, Vienna,
Austria.

1501

Let Them Blog

ENDNOTES
1

Except where noted, all examples of blogs

were created by the author in order to avoid
any privacy or copyright issues.

Blogcensus was developed by the National

Institute for Technology and Liberal Education (NITLE), a non-profit organization
funded by the Andrew Mellon Foundation.

This work was previously published in Handbook of Research on Literacy in Technology at the K-12 Level, edited by L. Tan
and R. Subramaniam, pp. 337-356, copyright 2006 by Information Science Reference, formerly known as Idea Group Reference (an imprint of IGI Global).

1502

1503

Chapter 4.6

Ubiquitous Computing
Applications in Education
Kostas Kolomvatsos
National & Kapodistrian University of Athens, Greece

Abstract

Inside Chapter

With the development of technology, new roads

have been opened in education. An interesting
idea is to use computers in teaching and learning procedure. Students will have the opportunity to gain access to information resources in
a timeless and limitless way. Teachers will be
able to transform their classes in a student-centered environment, avoiding the drawbacks of
the traditional teacher-centered model. In this
direction, ubiquitous computing has significant
advantages. Ubiquitous means that computational
devices are distributed into the physical world,
giving us boundless access to communication
and information channels. Now, knowledge can
be built based on collaboration, communication,
experimentation, and on students experiences.
Research has shown positive impacts on learning.
This chapter deals with issues directly connected
to ubiquitous computing, such as its features,
types of devices used, and pedagogical goals.
The advantages and disadvantages of ubiquitous
environments are fully examined and some initiatives are referred.

With this effort we try to cover the subject of

ubiquitous or pervasive computing in education.
We present important issues related to it. The first
is the features of this technology. It is important
to see and understand them. We also describe
the connection means in a pervasive environment. Devices are basic elements of such systems
because all educational activities are based on
them. Devices vary from those with small screens
to those with larger screens. Another separation
may be given based on their computational capabilities. We also deal with the pedagogical goals
that must be implemented. It is a crucial part of
such efforts because the desired result is to assimilate and efficiently teach the students. We
describe the advantages and disadvantages of the
referred technology. As we will see, ubiquitous
computing offers a lot of interesting advantages,
but on the other hand, there are open issues that
must be taken into consideration. The last part
is devoted to the description of some initiatives
that take or took place in universities and schools.
We describe some platforms that may be used to

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Ubiquitous Computing Applications in Education

construct learning environments, most of which

originated in the USA. Of course, there are common characteristics between these attempts, but
their number is large enough to force us to describe
only a few. These were randomly selected.

Introduction
New technologies have brought many changes in
teaching, and of course in learning. Traditional
classrooms are being transformed in order to
utilize the advantages of the technology.
Ubiquitous computing (also known as Pervasive, Ambient, 1 to 1, or one to one)
is about distributed computing devices in the
environment, with which users are able to gain
access to information resources. These devices can
be wearable computers, or sensors and computers embedded in everyday objects. On the other
hand, ubiquitous computing involves the necessary infrastructures needed to support pervasive
computing applications.
Ubiquitous computing integrates technology
into the environment, giving the opportunity to
users to utilize it anytime and anywhere. It differs from traditional systems where the user is
bonded to a computer in a specific place. Now
it is possible for a user to utilize the technology
without the restriction of place or time.
Ubiquitous computing may provide significant
advantages in the application domain of education.
It can offer continuous access to a wide range of
software, or the Internet, to all students, as well
as teachers. As we will see below, the main targets of using pervasive techniques in education
are efficiency in teaching and learning, equality
between all students as to access to technology,
regardless of their economical state, increased
student engagement with their lessons, and different approaches according to the students needs
(Bonifaz & Zucker, 2004).
This chapter is organized as follows. The next
section gives information about the examined

1504

technology, and the third section describes its basic

features. In the fourth section, a full description
of the means that are being used in order to help
a user connect and utilize ubiquitous facilities is
given. The fifth section describes the pedagogical
goals of the use of pervasive computing, and in the
sixth section, the advantages and disadvantages
of the emerged technology are given. In the seventh section, we outline some initiatives in this
research area, and in the eighth section, we give
a specific case study. Finally, our conclusions are
depicted in the last section.

Background
Ubiquitous computing environments are different
from what one traditionally finds in most school
settings. It offers to all students and teachers
continuous access to a wide range of software,
electronic documents, the Internet, and other
digital resources for teaching and learning. These
initiatives goals include increasing economic
competitiveness, reducing inequities in access
to computers and information between students
from wealthy and poor families, and raising
student achievement through specific interventions. Other reasons cited for supporting laptop
initiatives include improving classroom culture,
increasing students engagement, making it easier
to differentiate instruction according to students
needs, and solidifying home-school connections
(Bonifaz & Zucker, 2004).
The UK government and Scottish executives
have listed a number of priorities for ubiquitous
education for the 14+ age range. This list is discussed in Smith (2002) and Sutherland (2002).
According to authors, the priorities posed are:

Widening participation of students

Increasing the diversity of students in education
Quality and standards through the use of
ubiquitous computing

Ubiquitous Computing Applications in Education

Employability in students life

Accessibility to various resources through
the emerged technology
The globalization of learning and the emergence of alternative HE (Higher Education)
suppliers
Provision of education in more accessible
forms on a part-time basis
Employers wanting more workplace learning
Professionalisation of teaching
Student awareness of quality teaching and
pastoral care
Increasing IT literacy of younger students
and higher students expectations
Staff handling larger groups

Similar priorities have been posed in other

countries recognizing the need to change the
traditional educational model that has important
disadvantages. To this target, the International
Society for Technology in Education (ISTE)
published a list (ISTE, 2002) with the most commonly referred conditions to create a learning
environment for the use of technology. These
factors are:

Shared vision: This means that the commitment to technology is systemic and continual. Also, there is a proactive leadership
and administrative support for the entire
system.
Access: Teachers must have limitless access to current technologies, software and
telecommunications networks.
Skilled educators: The educators that are
called to instruct students, who use the
technology in their tasks, must be skilled
and familiar with the technology and with
its use in the teaching procedure. Hence,
learning will be more efficient.
Professional development: Educators
must have consistent access to professional
development in support of technology use

in teaching and learning.

Technical assistance: Educators and students must have limitless technical support
for maintaining and using the technology.
Usually a technical helpdesk is established
in order to meet these needs.
Content standards and curriculum resources: Teachers must be knowledgeable in
their subject and up to date with the content
standards and teaching methods in their
courses.
Student-centered teaching: Ubiquitous
access to technology drives the transformation of the traditional teacher-centered
system to student-centered. Educators
methodologies must comply with these new
requirements.
Assessment: There must be continuous
assessment of the effectiveness of technology for learning. All the problems and the
progress of each program must be recorded
and encountered in the programs implementation.
Community support: The community
and every schools partners must provide
expertise, support, and resources.
Support Policies: There must be policies
for financing, accessing in networks, and
rewards structures that must be in place to
support technology in learning.
As we can see, there is an intention to use
ubiquitous computing to help students learn
in a more productive way. Ubiquitous access to technology by teachers and students
has the capacity to transform not only the
physical learning environment, but also the
learning process, the role of students, and
the role of teachers (Fullan, 1999).

Results of some 1 to 1 computing initiatives

have shown:

Improved writing skills.

Increased student motivation and interest

1505

Ubiquitous Computing Applications in Education

in school.
Students that are more engaged in their
learning.
Teachers increasingly utilizing project-based
and hands-on curriculum and teaching
methods.
Increase of math and science scores in the
eighth grade.
Children born since 1980 process information differently than children born before
1980. They learn best with multisensory
input.

A case study is presented later in this chapter. It

can be very helpful because it depicts valid results
taken from questionnaires and interviews.
There are several research efforts in the field
of ubiquitous computing in education. Especially
with the spread of the wireless and handheld
technology, these efforts become more intensive
due to the advantages it can offer to learning
process. Current research focuses on the impact
of pervasive computing in teachers and students
(Bayon et al., 2002). As for the teachers, the goal
is to make them use effectively the new technology in their lessons design in order to produce
a student-centered class (Becker & Riel, 2000).
This is the basic characteristic of constructivist
theory, where students learn based on their collaboration and experimentation. Students build
their knowledge based on their experiences and
their research.
Regarding students, researchers study the
effect ubiquitous learning has on them in terms
of when and how they use the technology, how
much the technology affects their behaviour, and
how the new attempt is received by parents and
the school community (Diggs, 2004). Interviews
and observations have shown positive effects on
students and this way of learning is accepted
from the entire school community. There is an
enhanced motivation, collaboration, and communication between students. This is also supported
in Vahey, Enyedy, and Gifford (2002).

1506

The research results are very optimistic, but

there must be careful design and systematic observation to achieve the ultimate goal, which is
more productive student learning.

Features
The use of ubiquitous computing in education has
characteristics very important in learning. An
educational policy can be based on these features
in order to achieve a high level of learning. These
features are:

Information Access: Students have access

to their documents, and also to various
information sources from everywhere. All
students may search the information needed
in order to complete a task. Of course, the
initiative is on them. They pose questions
and take the results. The final step is to
combine the results and extract the final
conclusion.
Time and Place Immediacy: This feature is
allocated in the place and time in which the
information can be reached. Whenever and
wherever a student needs to access information is feasible. There are no limits in time
or in place. This has the advantage of easy
and useful access to resources, increasing
the productivity of the work.
Interactivity: Students are able to interact
with teachers or experts with a synchronous
or asynchronous communication. Hence,
they have the opportunity to approach other
peoples knowledge, without the one teachers limitation, of the traditional system.
They may search, find, and pose questions
to specific domain experts, and afterward
combine the answers to effectively build
their knowledge.
Student Activation: A system in which all
students have their own device and work
by themselves demands activation. Every
student has to work and learn, experiment-

Ubiquitous Computing Applications in Education

ing with software and searching to find the

information needed to complete specific
tasks.
Adaptability: Learners can get the right
information at the right time and at the right
place. In this direction, the Intelligent Agent
technology may be very useful. Intelligent
software may learn from the owners habits
or instructions and work as its representative, searching and find information. Even
further, this software can be used to adapt
the information presented to the users based
on their learning style.

Connection Means
When computers first arrived in educational
environments, they brought a lot of challenges
in the form of space for electrical cords, attachments, peripherals, and other entanglements
(McKenzie, 2001). It is obvious that in recent
years there was a blast in wiring and cabling
in school networks.
The ultimate goal of ubiquitous computing
in education is the use of a computer for all
students and teachers. This means that we need
a more flexible way to connect all the devices.
The development of the technology in the domain of wireless radio connections, which gives
a high speed communication channels, can give
us a means to efficiently interconnect devices in
large areas.
We believe that a wireless connection will
be more appropriate and efficient than a wired
network where the users will be bound to specific
spots in the area. Using wireless connection,
students may have access from anywhere. They
may have access in the cafeteria, in the library,
in the class or outside.
In McKenzie (2001) authors provide a list
of advantages of the wireless technology. They
are:

Ease of movement
Relaxed fit
Strategic deployment
Flexibility
Cleanliness
Low profile
Convenience
Simplicity
Speed (especially nowadays)

Synoptically, we can say that the greatest

advantage of wireless technology is that it offers
continuous access to the networks from anywhere
in school. One can see a wireless scenario in
Figure 1. All the devices can connect to access
points and thus can have access to all the network
resources. It can be implemented with either
infrared rays or wireless Ethernet (e.g., 802.11b,
or 802.11g with greater speed). Unfortunately,
infrared rays have a disadvantage because they
do not penetrate opaque objects. Hence, they may
only be used in the classroom. On the other hand,
wireless ethernet offers access to the network at
a high rate, at any time or place.
The devices that can be used for wireless access
vary, from a device with small screen and limited computational capabilities, like Blackberry
RIM, to a device with larger screen and increased
computational capabilities, like a laptop (Deters,
2001). Some pictures of these devices are shown
in Figure 2.
The first kind of device usually allows simple
interactions, like message passing, and so forth,
but on the other hand, its use is very easy, even
for a beginner. It may be used to help beginners
enter into this new technology.
Bigger devices like Palm computers, Pocket
PCs, or laptops offer increased computational
capabilities, giving users the opportunity to utilize more complex software. They have greater
memory size and a more powerful processor,
and they comprise miniaturization of a desktop
machine.

1507

Ubiquitous Computing Applications in Education

Figure 1. A wireless network example

Figure 2. Portable devices: (1) Laptop, (2) small $100 laptop (3) PocketPC, (4) BlackBerry RIM, (5)
handheld laptop, (6) palm computer

1508

Ubiquitous Computing Applications in Education

The most important matter with respect to

devices and their software is compatibility, because students working in groups must exchange
messages, files, or even more programs. Their
devices must be compatible in order to facilitate
the communication between them. Otherwise,
there will be a significant problem that may be
critical to the entire effort.

Pedagogical Goals
Todays generation of students looks at technology as part of their everyday environment. It is
important to understand the difference between
todays life in school in contrast to the past, where
students only occasionally used computers. In
the future, pupils will own a handheld device,
which will be their partner to complete tasks.
This means that these devices must be used in
a correct manner in order to help them in the
learning procedure.
To fully meet the students needs, technology
should be pervasivealways available. In one-toone learning, each student has access to a wireless
laptop to use at school or at home, enabling communication and collaboration among peers and
teachers, and connecting parents to their childrens
learning. Educators are provided digital tools to
create learning plans, manage educational content,
track student progress, and more.
The most important point is that full attention
to learning methodologies, with the help of the
digital world, must be given. Technology is the
mean to learning. A possible mistake will be to
pay attention in technology rather than learning. For this reason, policy makers must grapple
with several issues concerning appropriate and
effective integration of computers into schools
(Eagleton, 1999). In the referred list, one can
distinguish financial, equality, curricular, and
literacy issues.
The ultimate goal of such efforts is to improve
the assimilation ability of the students. To this

target, there are two critical factors. The first

is teachers and their ability to transform their
teaching. Teachers must know how to utilize the
technology in their courses. They must apply
new educational techniques oriented to the use of
technology. This procedure needs careful design.
It is useless to leave children to play for hours
or aimlessly browse the Web. A specific scope
must be posed.
The other factor is students. Research indicates
that constant exposure to multiple, overlapping
sights and sounds has affected the neural pathways of the brains of digital kids (Apple Educationhttp://www.apple.com/education). In fact,
some researchers contend that multisensory input
helps kids learn, retain, and use information
better. However, the use of technology does not
mean that all the students would do their learning with computers. For this reason, it is crucial
what type of technology will be distributed to
students. Students must be driven to collaborate
with their classmates, to use learning software
for their courses, to use their devices for communication, or to research with resources useful to
complete their tasks. They build their knowledge
by experimenting and testing their ideas based on
their experiences. The true value of technology
for learning lies not in learning to use technology,
but in using technology to learn (Educational
Research Service, 2001). A technology-using attitude must be cultivated in students in order to
enhance their achievement and engagement.
Moreover, computers may help in the direction
of the relevancy of schooling (Muir, 2005). Children always wander if what they learn in school
may be useful in their future. Laptop initiative
makes learning to appear more relevant to lifes
requirements. This is because students, with the
help of technology, learn based on projects that
allow students to use their own learning style. A
typical project may be the searching for information in the network and the configuration of
it to cover their needs. Hence, students see the

1509

Ubiquitous Computing Applications in Education

connection between things they learn and their

use in real life.
Furthermore, computers are an object that
can be used in many ways related to the personal
needs of each student. For example, some students
may learn better by searching for information, as
we described, while others may need multimedia
information in order to assimilate the concepts. It
is obvious that in such cases the most appropriate
and efficient way for learning is working in groups,
where all the students can contribute.
Thus, careful design of courses from the
teachers side may offer significant advantages
in learning procedure. Unfortunately, many
classrooms do not work that way. Pencil, paper,
lecture, textbook, review, and test are still the
norm. However, there are important attempts that
show us the road to the ubiquitous computing
revolution. Some of these tries are presented in
Ubiquitous Computing Initiatives.
In summary, ubiquitous computing environments provide portability, flexibility and efficiency
for collaborative learning projects (Sottilo, 2003).
Teachers have the ability to integrate computers
to every aspect of teaching and learning.

Advantages Disadvantages
In this point, it is necessary to describe all the
advantages and disadvantages of the emerged
technology. It is critical to identify these issues
because it is a key to the road of the embodiment
of computers in the learning procedure.
Some advantages are:

1510

Teaching efficiency: With the use of ubiquitous computing, teaching style has to change.
Teachers must adapt their methods to the
new environment. This active environment
gives the opportunity for students to build
their knowledge on their own. Therefore, it
is imperative that teachers change their lessons plan in order to reflect the new situation.

They will have the convenience to adjust

the lessons to each students needs and the
lessons goals. Educators can now implement a broader range of new approaches to
learning that are proven to be more effective.
Also, teachers must encourage and support
their students more vigorously, due to the
different nature of teaching.
Learning efficiency: One of the most important findings in past years is that student
engagement is one of the most critical factors
to learning achievement. It is obvious the
traditional teaching method has significant
drawbacks for its efficiency. A teacher-centered class does not provide the opportunity
for students to learn to build their knowledge
through collaboration or experiments. Ubiquitous computing offers an environment
where students use their devices to access
a class from anywhere, collaborate with
other students, experiment with appropriate software, and pose questions to their
classmates or their teacher. This method
is more productive because the knowledge
is built on students experiences. Students
are more active in the course and their engagement is compulsory in order to achieve
the goals that are posed to them. Hence,
the traditional model is transformed to a
student-centered model, where the teacher
becomes a consultant to their students. The
procedure that we described reflects the
method of constructivist pedagogy.
Access to technology: Today, there are many
computers in schools placed at labs, but students do not have limitless and timeless access to them. Consider the scenario where all
the students have their own device to access
teaching resources either from their class or
from their homes. In this scenario, all the
students are fluent with the use of technology. This is a very attractive idea because
in such a class students are technologically
equal. This equality can be allocated in the

Ubiquitous Computing Applications in Education

use of technology. In past years, there was a

gap in technology access that translated into
inequalities in different fields among people.
This is known as Digital Divide (Digital
Divide.org, 2005). Now, each student has a
device and limitless access to digital information augmenting their productivity. The
new technology gives more opportunities
to students from previously disadvantaged
groups. Also, research has shown that giving
students a personal device can make learning
more efficient (Vahey et al., 2002). Students
become more autonomous and self-directed
in their learning.
Easy to use: Todays children are familiar
with technology, especially with games.
Gameboys, Playstations, and even
computers are used for gaming. Thereby, it
is easy to provide learning devices to students for their use. Of course, instructions
must be given to them for the right use of
the tools in order to avoid problems from
inappropriate use, as we see below. Students
want personal devices that, like a toy, are
easy to operate.
Productivity: With the emerged technology
students can raise their productivity. This is
because they build their knowledge based
on techniques that are proven by research to
give better results. Students become active,
taking initiatives and working with their own
pace. Also, the interaction with classmates
of whom they share common interests plays
an important role in the learning process.
Finally, we must note that in such cases
experimentation is another factor that builds
knowledge. Students have boundless access
to the Internet or to specific software, and
may perform their own tasks with their
devices or search information in various
resources, such as the Internet or digital
databases, and so forth.
Portability: Students, during their time
spent in schools, must be familiarized with

tools, software, and devices that differ due

to the different architecture of each computer. Computers are used mainly in labs
devoted to a specific lesson. They spend
significant time learning the use of each
device and applications that detaches them
from the actual learning activity. With
their personal devices, they can focus on
the learning procedure without worrying
about the differences between various tools
and applications. They can connect to the
network from any place at any time.
Just like all the other domains of technology,
ubiquitous computing has some drawbacks.
These drawbacks must be taken into consideration by educational policy makers, and
by the implementers of this technology, in
order to anticipate and avoid problems that
may appear in the implementation phase.
Cost: Initiatives such as ubiquitous computing in education, where every student will
have their own device in order to perform
tasks, have a cost, detected at two points.
The first is the financial cost. The provision
of a big amount of devices is very costly,
but, of course, their prices are reduced as
the technology evolves, making them affordable. On the other hand, there is cost in
the transformation of the traditional system
into the new. A lot of things must change,
such as the lesson plan, the teaching style,
and the assessments model. This requires
much effort from the teachers and the policy
makers. It is important to note that the new
model requires a careful design of all the
teaching parts because in this environment
students will be able to act freely.
Teachers Expertise: This kind of program
needs teachers expertise with the technology and with the design of the new kinds of
lessons. Teachers must be fluent with this
type of education. There is a fear that they
will not know how to use the technology
and how to design their lessons. Also, there

1511

Ubiquitous Computing Applications in Education

is a possibility that they may not use the

technology in their teaching methodology
in an appropriate way. A solution to these
problems is to further educate teachers in
coping with the new situation.
Inappropriate use of Technology: There is
a fear that students may use the electronic
devices with which they are provided inappropriately. For this reason measures must
be taken. A lot of programs have the mechanisms to control the network access in order
to avoid problems. Firewalls are in the front
line of security. With them the administrators can check the traffic and discover the
cases that are dangerous to students, and
even more, to the system.
Equipment damage: Devices are basic
elements in ubiquitous computing environments. They are means for access in
information sources. Also they are used for
communication purposes. Communication
and collaboration are the keys to succeed
with the pedagogical goals we described in
Pedegogical Goals. Hence, problems will
arise when technical inefficiencies occur. In
such cases, technical support is important
for the programs success; otherwise the
participants faith may be unsettled.

All these advantages and disadvantages must

be taken into consideration from the policy makers, and a specific policy must be set up according
to each cases characteristics.

Ubiquitous Computing Initiatives

This section will present some initiatives and tools
originating from all over the world. Of course, it
is impossible to present all the efforts. A full list
of the schools that apply the laptop initiative is
available at http://etc.usf.edu/L4L/A-Review.html.
Additional information about all programs in each

1512

state of the United States is given at http://www.

k12one2one.org/initiatives.cfm?initID=97. We
present some of them, in order to describe basic
characteristics of ubiquitous computing in education. We must note that most of them originate in
the United States, where first attempts appeared
in the late 90s. Tools for a learning environments
construction are also presented. All these initiatives are being held in elementary schools or
universities. The common characteristic between
them is that the majority use portable devices
(see Connection Means) with wireless access to
the network. On their Web sites, evaluations are
given for every one of the programs in order to
discern their results and identify the advantages
and disadvantages. However, it is commonly the
conviction that the use of a computing environment in which the students complete their tasks,
offers positive remarks in learning procedure,
augmenting the enhancement and engagement
of the pupils.

Ulysses
This is an effort from the University of Laval, in
Quebec, Canada (Mantha, 2001). All the students
and teachers are provided with a laptop. In the academic year 2001-2002, there were approximately
1800 laptops in the school. Students and teachers
invested money to buy the devices. In Ulysses,
the laptop is to a large extent a communication
tool, and thus there is a need to be able to connect
to the network. The connection is established in
the classrooms, in the cafeteria, in hallways, and
so forth. For financial reasons, the system uses
wired connections to the network. There are 17
wired classrooms, of which 11 have the U shape.
There are 6 classrooms with movable tables with
network connections on the periphery of the room,
allowing the grouping of tables for team work. In
every classroom there is a console in which one
can find a computer and multimedia tools. An
intranet was developed to store all the material
needed for courses, like exercises, examples, and

Ubiquitous Computing Applications in Education

so forth. Tools used for communication are e-mail,

forums, course management utilities, and Athena.
Athena is a tool with which the instructor is able
to poll the class in a small amount of time and
get the students opinion about a question. The
answers can be short texts, yes/no or a selection
in a multiple choice question. Also, the instructor
can display the answers on the projection surface
at the front of the class. This capability may be
very useful in the teaching process.
For user support, two helpdesks were set up.
There is a technical helpdesk and a pedagogical.
The first is responsible for hardware and software
problems. Common issues encountered are computer parts replacement, systems configuration,
network connection parameterization, or systems
and applications installation. The second helpdesk
is responsible for answering questions concerning
the use of technology from teachers and students
in a pedagogical setting. There are a lot of means
to access this support center. One can get support from the desk, from the Web site, over the
phone, or via e-mail. This center can help tutors
find and apply new means of uses of technology
in their lessons.

eFusion
This system consists of an initiative of the University of Illinois that originated in the spring of
2002 (Peiper, Warden, Chan, Capitanu, & Kamin,
2005). eFuzion is an interactive classroom learning
environment where all students and teachers have
access to computing devices inside and outside the
class. During the lesson, educators present their
notes and examples with the help of presentation
tools that eFusion provides. Consequently, teachers send their notes to students devices through
a wireless network and students may make their
remarks in presentations and store them for
future access. The system supports interactive
newsgroups and communication tools that give
the opportunity for students to communicate
with their instructors or their assistants and take

answers during a lecture or asynchronously at a

later time.
The most important part of the system is
the facility for incorporation of active learning
exercises into the lectures, giving instructors a
mean to see the comprehension level of the learners in real time. As we understand, it is about a
system that has the ability to facilitate teachers
to diagnose their students needs in a real-time
manner during lectures. This is a critical factor
because, based on results taken from quizzes or
other forms of questions, instructors may change
the teaching methodology in order to achieve a
better comprehension level.

WIL/MA Toolkit
This is a toolkit written in Java and initiated by the
University of Mannheim (http://www.lecturelab.
de/UCE_4.html). It is a client-server application,
where a server provides connection management,
user management, and service management. The
first supports the establishment of a connection
to the users and gives the capability for administrators to monitor the entire communication in
and out from the server. The second is used for
user identification and authentication. Finally, the
service management unit is responsible to inform
users of what services are available and also to
control data flow.
The system is based on wired connections,
but in order to avoid extensive use of cables,
students are able to connect to the network with
wireless LAN. The devices used by the students
are PocketPCs. On the other hand, teachers are
able to publish their material, which can be
presentations, slides, or files, and furthermore,
they can broadcast image and voice to remote
places. During their lectures, teachers can also
use tools like call-in (spontaneous questions),
or quiz and feedback (from the students in real
time). Asynchronous tolls are messaging and
chat/forum channels.

1513

Ubiquitous Computing Applications in Education

University of Texas
In 2002, the University of Texas initiated a requirement for all teacher education students to
obtain a prescribed laptop and software for use
throughout their academic preparation (Resta &
Tothero, 2005). The goal is the preparation of a
new generation of teachers who would be able to
use new tools and practices in their teaching. This
programs vision is to prepare the future teachers
to enhance their future students to learn in technologically rich classrooms. Its official name is
LIFE (Laptop Initiative for Future Educators).
All the students are provided with laptops and
software that meet specific requirements, and
wireless access to network in their classrooms or
throughout the building. The involvement with the
technology creates two critical needs. The first
is the need to train the students to use the new
hardware and software tools. For this reason, in
the start of every semester, workshops are offered
to familiarize new students with the systems and
applications. The second is the need for technical
support. For this, a Students Laptop Helpdesk was
established. It provides equipment, supplies, and
instruction for both students and teachers.

Maine Public Schools

This is one of the first programs that try to embody ubiquitous computing in education and is
considered the single largest 1 to 1 educational
technology program in the world (http://www.
maine.gov/portal/education/educ_tech.html). It
is a project of the Maine Department of Education. Every 7th and 8th grade student and teacher
has a wireless laptop with a rich set of software.
Students have the capability to work with word
processing, spreadsheet creation, movie processing, e-mail clients, Internet browsing, and other
multimedia software. Also, students can take
laptops to their home, where they can have access
to the network. There are now 37.000 laptops in
Maines 239 middle level schools. Additionally,

1514

there is a small group of educators, educational

technologists producing research, documentation, evaluation, and advocacy of 1 to 1 learning.
Evaluations have shown that student engagement
and attendance are up and behaviour referrals are
down (Silvernail, Harris, Lane, Fairman, Gravelle,
Smith et al., 2003). The most prominent result is
that parents can see this positive effect on their
childrens behaviour. The key to the success of
this program is to focus on the initiatives as a
computer enhanced learning procedures and not
merely as technical training. The technology
initiatives must focus on teaching and learning,
and not on technology, in order to be successful
(Schacter, 1995).
Recently, there was an agreement to upgrade
the hardware and software tools. The most important change is the wireless network transition
to Airport Extreme 801.11g, giving a 54 Mbps
speed and improving performance.

Vermont Public Schools

This is an initiative developed by the Vermont Department of Education (http://www.k12one2one.
org/www.oakgrove.k12.vt.us). Vermont teachers
and students have access to ubiquitous technology,
which is available on an as needed basis. The
results of the first year of the program were:

Teachers were more comfortable with

implementing more technology integrated
lessons.
Students were better organized.
Students created better quality work.
Students experimented with various tools.
An impact on student learning because of
the full access to the technology.

All the teachers and students are provided with

laptops, with which they can have access to the
network and to various software tools. They may
process text, databases, spreadsheets, video, mu-

Ubiquitous Computing Applications in Education

sic, or even more images. Finally, Web browsing

and communication tools are available.

New Hampshire Schools

This attempt concerns all the 7th grade students in
six of the states neediest schools (http://www.nhadvantage.com/laptop.htm). Approximately 400
students participated in the program the first year.
All these students are provided with a laptop and
wireless network access. Additionally, cameras,
printers, and videoconferencing tools are available
to students and teachers. Teachers can control the
childrens laptops during classes while students
do their assignments. The analysis of the results
concerning the first year evaluation indicates
many of the most frequently cited benefits of 1
to 1 computing, such as: increased involvement
with technology from both teachers and students,
increased student engagement, achievement, and
motivation, and improved interactions between
teachers and students.

Talbot County Public Schools

This is a laptop initiative in which all pupils hold
a portable device with which they can cope with
school tasks. This project has four goals, which
are:

Increased student achievement

Effective technology for instruction
Increased student engagement in the learning process
Improved educational access for and participation by high-risk students and their
parents

The programs drift derives from classroom

observations, teacher competencies identified
by National Education Technology Standards
(NETS) for teachers, lesson plans that utilize
technology, and examining students work and
performances.

Computers were provided to each ninth

grade student at the beginning of the 2005-06
school year (http://www.tcps.k12.md.us/index.
php?page=1_to_1_laptop_initiative). Teachers
were given extensive training on the computers
and on instructional software.
Students have access to network from any
place, in school or at home. Also, a helpdesk is
available to students for technical and insurance
issues.

Alexandria City Public Schools

The Alexandria City Public Schools High School
Technology Integration Project provides a 1 to 1
computing environment for all students in grades
9-12 (http://www.acps.k12.va.us/tip/index.php).
A laptop is provided to each student, who has
the opportunity to keep the device for the duration of the school year. Students have access to
a wireless network with which they can access
resources and material useful for their tasks. A
set of applications is installed on laptops that is
useful for common tasks like word processing,
as well as a variety of teaching material and
applications. Teachers can leave their material
in a students server, enhancing the electronic
exchange of information. In this server, students
have personal work folders where they can store
their assignments. Printing support is available
for students and teachers. Finally, we must note
that for the programs support, there is a help desk
that is responsible for repairing damages.

Hopkins Public Schools

This pilot program involves approximately 600
students and 30 staff members. The programs
goals are to enhance students achievement
and engagement. Each student owns a laptop
with which to complete his or her assignments.
Teachers can give their files and presentations
in electronic mode (http://weblogs.hopkins.k12.
mn.us/onetoone/). Students having wireless access

1515

Ubiquitous Computing Applications in Education

can take the material and, using software like office, Web browsing tools, database, video, music,
image processing, and communication software,
can cope with their commitments. It is worthwhile
to mention that students cannot have access to the
Internet from their homes except in the cases that
the family has its own Internet provider. Technical
support is provided on a full-time basis.

Other Areas in the United States

There are laptop initiatives in 30 states in the
United States (http://www.k12one2one.org/initiatives.cfm?initID=97). These states are:
Alaska
Arizona
Arkansas
California
Connecticut
Florida
Georgia
Illinois
Indiana
Iowa
Kansas
Maine
Massachusetts
Michigan
Minnesota

Missouri
New Hampshire
New Jersey
New Mexico
New York
North Carolina
Ohio
Oregon
Pennsylvania
South Carolina
Texas
Vermont
Virginia
Washington

These programs either came as district or

school initiatives. Like the previously described
examples, they are programs that offer wireless
access from portable devices to their students
and teachers. The goals and the characteristics
of these programs are the same as we described
in the previous cases.

Kaifu School Hamburg

The project started in 1999 and lasted until 2003,
and it is among the first in Europe (http://www.

1516

hamburgmediaschool.com/pages/p118328_
v1.html). Since then, the school chose to continue
using portable devices in three classes. However,
the entire effort is characterized as positive. All the
students were provided with wireless laptops and
specialized software. Students that continue this
program are specialized in organizing themselves
and in learning to use technology, which was the
goal of the project.

Minervaskolan School
Since 1999, pupils in Minervaskolan have used
laptops with wireless access to the network
(http://www.minervaskolan.se/). The programs
technical specifications are similar to the others
as we described. The most important is that from
the projects evaluation students non-attendance
approaches to zero and scores to national tests
are remarkable. Also, students grades appear to
be to a continuous increment advocating to the
programs success. Now, Minervaskolan is one
of the top schools in Sweden.

Case Study: The Maine

Learning Technology
Initiative Teacher, Student,
and School Perspectives,
Mid-Year Evaluation Report
In this section, we synoptically show the results
taken from the evaluation of the Maine learning technology initiative (Sirvenail et al., 2003).
Three core areas are investigated: Teachers and
teaching, students and learning, and schools and
community. In order to derive useful conclusions
for each area, specific questions were posed. The
goal of these questions is to identify the impact,
the use, and the obstacles of the technology.
Therefore, there are three key questions.

Ubiquitous Computing Applications in Education

How are the laptops being used?

What are the impacts of the laptops on
teachers, students, and schools?
Are there obstacles to full implementation
of the initiative?

We give a short presentation of the evaluation

results with regard to each core area.
The evidence in the area of teachers and teaching indicates that the majority of teachers use
their laptops in lesson development and classroom
instruction. They utilize their devices to prepare
and present their lessons. Of course, they agree
that the use of the merged technology has positive
impacts on their teaching. Now, they are aware
of up-to-date information critical to present a
meaningful course. On the other hand, they see
an obstacle to this effort concerning the technical
support. They feel that technical problems and
the lack of technical support sometimes limit
the use of laptops. This is a critical factor for the
programs success. They also felt that they need
time to learn the new technology and its applications and professional development activities
designed to help them in the teaching process. Of
course, time must be given in order to overcome
the limitations.
Students report that they use their laptops to
complete tasks, to search for information sources,
and to communicate with teachers and classmates.
More importantly, they feel an increase in the
benefits of their work, both in and out of school.
The new technology has increased their interest
in school activities, but there are some problems
in coping with technical difficulties.
Positive impacts of the application of new
technology in learning procedure are recognized
from the school community. The difference in
childrens interest is visible to their parents.
Financial matters are obstacles to this effort, but
through creative solutions many schools find ways
to minimize the overall cost.
As we have seen, there are many positive
impacts in introducing these new technologies

to education. Teaching focuses on students

generating a new environment, where learning
comes from collaboration, experimentation, and
research. At the same time there are obstacles,
especially in technical support, that in time will
be eliminated.

Conclusion
Ubiquitous computing changes the field of education. It offers a timeless access to information
resources and allows learning methods that
are difficult to apply in traditional classrooms.
Research has shown it has positive impacts on
students' learning. Also, a lot of advantages exist
in teaching, and school communities see these
positive impacts of the new technology.
In order to integrate computers in education
smoothly, careful design is necessary. The focus
must be on the pedagogy and not on technology.
This is because the final goal is to accomplish a
high level of learning. For this reason, we speak
about technology enhanced learning, and not
about technical training.
Ubiquitous or pervasive computing has significant advantages over traditional teaching
methods, and we must work on them to reach
the desired results.

References
Apple Computer, Inc. (n.d.). Why 1 to 1 learning.
Retrieved October 11, 2006, from http://www.
apple.com/education
Bayon, V., Rodden, T., Greenhalgh, C., & Benford,
S. (2002, August 26-28). Going back to school:
Putting a pervasive environment into the real
world. Pervasive, 69-84. Zurich, Switzerland.
Becker, H., & Riel, M. (2000). Teacher professional engagement and constructive-compatible
computer usage (Report No. 7). Irvine, CA:

1517

Ubiquitous Computing Applications in Education

Teaching, Learning, and Computing. Retrieved

October 11, 2006, from www.crito.uci.edu/tlc/
findings/report7

The Educational Technology Journal, 10(4).

Retrieved October 11, 2006, from http://www.
fno.org/jan01/wireless.html

Bonifaz, A., & Zucker, A. (2004). Lessons learned

about providing laptops for all students. Retrieved
December 18, 2006 from http://www.neirtec.
org/laptop/LaptopLessonsRprt.pdf

Muir, M. (2005). Laptops for learning: The Maine

Learning Technology Initiative. Joint publication
in the Middle Level newsletters of Association of
Childhood Education International (ACEI) and of
the National Association of Elementary School
Principals (NAESP).

Deters, R. (2001, May 19-23). Pervasive computing devices for education. In Proceedings of the
International Conference on AI and Education
(pp. 49-54), San Antonio, Texas.
Diggs, C. (2004, May). The effect of palm pilots
on student learning in the elementary classroom.
Retrieved October 11, 2006, from http://www.
successlink.org/handheld/PalmInHand.pdf
Digital Divide.org (2005). Digital divide: What
it is and why it matters. Retrieved December 18,
2006 from http://www.digitaldivide.org/dd/digitaldivide.html
Eagleton, M. (1999, April). The benefits and
challenges of a student-designed school Web site.
Retrieved October 11, 2006, from http://www.
readingonline.org
Educational Research Service. (2001). Does technology improve student achievement? Retrieved
October 11, 2006, from http://www.ers.org
Fullan, M. (1999). Change forces: The sequel.
London: Farmer Press.
International Society for Technology in Education (ISTE) (2002). Essential conditions to make
it happen. Retrieved December 18, 2006 from
http://cnets.iste.org/students/s_esscond.html
Mantha, R.W. (2001, May 4-6). Ulysses: Creating a
ubiquitous computing learning environment sharing knowledge and experience in implementing
ICTs in universities. EUA / IAU / IAUP Round
Table, Skagen, Denmark.
McKenzie, J. (2001, January). The unwired
classroom: Wireless computers come of age.

1518

Peiper, C., Warden, D., Chan, E., Capitanu, M.,

Kamin, S. (2005, June 27-29). eFuzion: Development of a pervasive educational system. In Proceedings of the 10th Annual SIGCSE Conference on
Innovation and Technology in Computer Science
Education (pp. 237-240). Caparica, Portugal.
Resta, P., & Tothero, M. (2005). Establishing a
ubiquitous computing environment for teacher
preparation students and faculty. In S. Powers &
K. Janz (Eds.), Ubiquitous and pervasive computing in higher education. Terre Haute, Indiana:
Indiana State University Curriculum Research
and Development Center.
Schacter, J. (1995). The impact of educational
technology on student achievement. The Milken
Exchange on Educational Technology. Retrieved
October 11, 2006, from http://www.mff.org/publications/publications.taf?page=161
Silvernail, D.L., Harris, W.J., Lane, D., Fairman,
J., Gravelle, P., Smith, L., et al. (2003, March). The
Maine learning technology initiative: Teacher,
student, and school perspectives mid-year evaluation report. Maine Education Policy Research
Institute.
Smith, T. (2002). Strategic factors affecting the
uptake, in higher education, of new and emerging technologies for learning and teaching. York:
Technologies Centre. Retrieved October 11, 2006,
from www.techlearn.ac.uk/NewDocs/HEDriversFinal.rtf

Ubiquitous Computing Applications in Education

Sottilo, S. (2003, May-June). Pedagogical advantages of ubiquitous computing in a wireless

environment. The Technology Source Archives.
Retrieved October 11, 2006, from http://ts.mivu.
org/default.asp?show=article&id=1034

Sutherland, A. (2002). Strategic factors affecting the uptake, in further education, of new and
emerging technologies for learning and teaching.
Retrieved October 11, 2006, from http://www.
techlearn.ac.uk/cgi-bin/docspec.pl?l=83

Stiles, M. (2002, September 10-11). Strategic and

pedagogic requirements for virtual learning in the
context of widening participation. In Proceedings
of the Nirtual Learning & Higher Education
Conference, Oxford, UK.

Vahey, P., Enyedy, N., & Gifford, B. (2000). Learning probability through the use of a collaborative,
inquiry-based simulation environment. Journal of
Interactive Learning Research, 11(1), 51-84.

1519

Ubiquitous Computing Applications in Education

Appendix I: Useful URLs

Ubiquitous Computing Information
http://nano.xerox.com/hypertext/weiser/UbiHome.html
Notes on Ubiquitous Computing
http://hci.stanford.edu/cs147/notes/ubicomp.html
Ubiquitous Computing Applications
http://www-lce.eng.cam.ac.uk/~kr241/html /101_ubicomp.html
Ubiquitous Computing Research Activities
http://ubicomp.teco.edu/index2.html
Ubiquitous Computing Evaluation Consortium
http://www.ubiqcomputing.org/ lit_review.html
Ubiquitous Computing Education
http://www.ubqed.info/
IEEE Pervasive Computing - A catalyst for advancing research and practice in ubiquitous
computing
http://www.computer.org/portal/site/pervasive//

This work was previously published in Ubiquitous and Pervasive Knowledge and Learning Management: Semantics, Social
Networking and New Media to Their Full Potential, edited by M. Lytras and A. Naeve, pp. 94-117, copyright 2007 by IGI
Publishing, formerly known as Idea Group Publishing (an imprint of IGI Global).

1520

1521

Chapter 4.7

Unraveling the Taste Fabric of

Social Networks
Hugo Liu
The Media Laboratory, USA
Pattie Maes
The Media Laboratory, USA
Glorianna Davenport
The Media Laboratory, USA

ABSTRACT
Popular online social networks such as Friendster
and MySpace do more than simply reveal the superficial structure of social connectednessthe
rich meanings bottled within social network
profiles themselves imply deeper patterns of
culture and taste. If these latent semantic fabrics
of taste could be harvested formally, the resultant
resource would afford completely novel ways
for representing and reasoning about web users
and people in general. This paper narrates the
theory and technique of such a featthe natural
language text of 100,000 social network profiles
were captured, mapped into a diverse ontology
of music, books, films, foods, etc., and machine
learning was applied to infer a semantic fabric of
taste. Taste fabrics bring us closer to improvisational manipulations of meaning, and afford us

at least three semantic functionsthe creation

of semantically flexible user representations,
cross-domain taste-based recommendation, and
the computation of taste-similarity between peoplewhose use cases are demonstrated within the
context of three applicationsthe InterestMap,
Ambient Semantics, and IdentityMirror. Finally,
we evaluate the quality of the taste fabrics, and
distill from this research reusable methodologies
and techniques of consequence to the semantic
mining and Semantic Web communities.

Introduction
Recently, an online social network phenomenon
has swept over the WebMySpace, Friendster,
Orkut, thefacebook, LinkedInand the signs
say that social networks are here to stay; they

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Unraveling the Taste Fabric of Social Networks

constitute the social Semantic Web. Few could

have imagined ittens of millions of Web users
joining these social network sites, listing openly
their online friends and enlisting offline ones too,
and more often than not, specifying in great detail
and with apparent exhibitionism tidbits about who
they are, what music they listen to, what films they
fancy. Erstwhile, computer scientists were struggling to extract user profiles by scraping personal
homepages, but now, the extraction task is greatly
simplified. Not only do self-described personal
social network profiles avail greater detail about a
users interests than a homepage, but on the three
most popular sites, these interests are distributed
across a greater spectrum of interests such as
books, music, films, television shows, foods,
sports, passions, profession, etc. Furthermore,
the presentation of these user interests is greatly
condensed. Whereas interests are sprinkled
across hard-to-parse natural language text on
personal homepages, the prevailing convention
on social network profiles sees interests given as
punctuation-delimited keywords and keyphrases
(see examples of profiles in Figure 1), sorted by
interest genres.

It could be argued that online social networks

reflectwith a great degree of insightthe social
and cultural order of offline society in general,
though we readily concede that not all social
segments are fairly represented. Notwithstanding, social network profiles are still a goldmine
of information about people and socialization.
Much computational research has aimed to understand and model the surface connectedness and
social clustering of people within online social
network through the application of graph theory
to friend-relationships (Wasserman, 1994; Jensen
& Neville, 2002; McCallum, Corrada-Emmanuel,
& Wang, 2005); ethnographers are finding these
networks new resources for studying social behavior in-the-wild. Online social networks have
also implemented site features that allow persons
to be searched or matched with others on the basis
of shared interest keywords.
Liminal semantics. However, the full depth
of the semantics contained within social network
profiles has been under-explored. This paper
narrates one such deep semantic exploration
of social network profiles. Under the keyword
mediation scheme, a person who likes rock

Figure 1. Examples of social network profile formats, on Orkut (left) and Friendster (right). Note the
similarity of categories between the two.

1522

Unraveling the Taste Fabric of Social Networks

climbing will miss the opportunity to be connected with a friend-of-a-friend (foaf) who likes
wakeboarding because keyword-based search
is vulnerable to the semantic gap problem. We
envision that persons who like rock climbing
and wakeboarding should be matched on the
basis of them both enjoying common ethoi (characteristics) such as sense of adventure, outdoor
sports, and thrill seeking. A critic might at
this point suggest that this could all be achieved
through the semantic mediation of an organizing ontology in which both rock climbing and
wakeboarding are subordinate to the common
governor, outdoor sports. While we agree that
a priori ontologies can mediate, and in fact they
play a part in this papers research, there are subtler
examples where a priori ontologies would always
fail. For example, consider that rock climbing,
yoga, the food sushi, the music of Mozart,
and the books of Ralph Waldo Emerson all have
something in common. But we cannot expect a
priori ontologies to anticipate such ephemeral affinities between these items. The common threads
that weave these items have the qualities of being
liminal (barely perceptible), affective (emotional),
and exhibit shared identity, culture, and taste. In
short, these items are held together by a liminal
semantic force field, and united they constitute
a taste ethos.
What is a taste ethos? A taste ethos is an
ephemeral clustering of interests from the taste
fabric. Later in this paper we will formally explain
and justify inferring a taste fabric from social
network profiles, but for now, it suffices to say
that the taste fabric is an n by n correlation matrix,
for all n interest items mentioned or implied on a
social network (e.g., a book title, a book author,
a musician, a type of food, etc.). Taste fabric
specifies the pairwise affinity between any two
interest items, using a standard machine learning numeric metric known as pointwise mutual
information (PMI) (Church & Hanks, 1990). If a
taste fabric is an oracle which gives us the affinity between interest items as a(xi, xj), and a taste

ethos is some set of interest items x1, x2, xk,

then we can evaluate quantitatively the strength,
or taste-cohesiveness, of this taste ethos. While
some sets of interest items will be weakly cohesive, other sets will demonstrate strong cohesion.
Using morphological opening and thresholding
(Serra, 1982; Haralick, Sternberg, & Zhuang,
1987), standard techniques for object recognition
in the image processing field, we can discover
increasingly larger sets of strong cohesiveness.
The largest and most stable of these we term taste
neighborhoodsthey signify culturally stable
cliques of taste. Visualizing these interconnected
neighborhoods of taste, we see that it resembles
a topological map of taste space!
Taste neighborhoods and taste ethoi, we suggest, are novel and deep mechanisms for tastebased intrapersonal and interpersonal semantic
mediation. Rather than mapping two persons into
interest keyword space, or into a priori ontological space, the approach advocated in this paper
is to map the two persons first into taste-space,
and then to use their shared ethoi and neighborhoods to remark about the taste-similarity of
these persons.
Emergent and implicit semantics. While our
work builds directly upon age-old language modeling techniques in Computational Linguistics, and
graph-based associative reasoning in Artificial
Intelligence (Collins & Loftus, 1975; Fellbaum,
1998; Liu & Singh, 2004), it is also sympathetic
to trends in the Semantic Web literatureaway
from formal semantics, and toward an embrace of
emergent and implicit semantics. In Volume 1 of
this journal, Sheth, Ramakrishnan, and Thomas
(2005) distinguish between formal, implicit, and
powerful (soft) semantics for the Semantic Web
movement. Whereas formal semantics must be
manually specified, implicit semantics can be
readily mined out of the unstructured Web using
statistical approaches. Upon further refinement,
implicit semantic resources can be transformed
into powerful (soft) semantic resources that afford the ability to mediate informal and formal

1523

Unraveling the Taste Fabric of Social Networks

entities. Related to implicit semantics, emergent

semantics is an evolutionary approach to knowledge management (Staab et al., 2002; Aberer et
al., 2004) that advocates semantic organization
to be shaped from the ground-up, a posteriori,
and in accordance with the natural tendencies of
the unstructured datasuch a resource is often
called a folksonomy. We suggest that online social
network profiles give an implicit semantics for
cultural taste-space, and that taste fabrics afford
a semi-formal, soft semantics appropriate for
semantic mediation between informal and formal
entities. Finally, arising out of correlation analysis, topological features of the taste fabricsuch
as taste neighborhoods, identity hubs, and taste
cliquesconstitute an emergent semantics for
taste-space.
Papers organization. The rest of the paper
has the following organization. Section Two lays
out a theoretical foundation for representing and
computing taste, framed within theories in the
psychological and sociological literatures. In
particular, it addresses a central premise of our
taste-mining approachis the collocation of
interest keywords within a single users profile
meaningful; how does that tell us anything about
the fabric of taste? The section titled Weaving
the Taste Fabric narrates the computational architecture of the implementation of taste fabric,
including techniques for ontology-driven natural
language normalization, and taste neighborhood
discovery. The section What Is a Taste Fabric
Good For? describes three semantic functions
of a taste fabricsemantically flexible user
modeling, taste-based recommendation, and interpersonal taste-similaritywithin the context
of three applicationsInterestMap (Liu & Maes,
2005a), Ambient Semantics (Maes et al., 2005),
and IdentityMirror. The following section evaluates the quality of the taste fabric by examining
its efficacy in a recommendation task, and also
entertains an advanced discussion apropos related
work and reusable methodologies distilled from

1524

this research. The final section in the paper is

the conclusion.

Theoretical Background
This section lays a theoretical foundation for
how taste, identity, and social network politics
are approached in this work. For the purposes of
the ensuing theoretical discussion, social network
profiles of concern to this project can be conceptualized as a bag of interest items which a user has
written herself in natural language. In essence, it
is a self-descriptive free-text user representation,
or harkening to Julie Andrews in The Sound of
Music, these are a few of my favorite things. A
central theoretical premise of mining taste fabric
from social network profiles by discovering latent
semantic correlations between interest items is
that the collocation of a users bag of interest
items is meaningful, structured by his identity,
closed within his aesthetics, and informs the total
space of taste. Next, the paper argues that a users
bag of interests gives a true representation of his
identity, and enjoys unified ethos, or, aesthetic
closure. This is followed with a section which plays
devils advocate and betrays some limitations to
our theoretical posture. The section theorizes a
segregation of users profile keywords into two
speciesidentity-level items vs. interest-level
items. This distinction has implications for the
topological structure of the taste fabric.

Authentic Identity and Aesthetic

Closure
In the wake of this consumer-driven contemporary
world, the proverb you are what you eat is as
true as it has ever beenwe are what we consume. Whereas there was a time in the past when
people could be ontologized according to social
class, psychological types, and generationsthe
so-called demographic categoriestodays world

Unraveling the Taste Fabric of Social Networks

is filled with multiplicity, heterogeneity, and diversity. The idea that we now have a much more
fine-grained vocabulary for expressing the self
is what ethnographer Grant McCracken, echoing
Plato, calls plenitude (McCracken, 1997). In a
culture of plenitude, a persons identity can only
be described as the sum total of what she likes
and consumes. Romantic proto-sociologist Georg
Simmel (1908/1971) characterized identity using
the metaphor of our lifes materials as a broken
glassin each shard, which could be our profession, our social status, our church membership, or
the things we like, we see a partial reflection of
our identity. These shards never fully capture our
individuality, but taken together, they do begin to
approach it. Simmels fundamental explanation
of identity is Romantic in its genre. He believed
that the individual, while born into the world as
an unidentified content, becomes over time reified
into identified forms. Over the long run, if the individual has the opportunity to live a sufficiently
diverse set of experiences (to ensure that he does
not get spuriously trapped within some local
maxima), the set of forms that he occupiesthose
shards of glasswill converge upon an authentic
description of his underlying individuality. Simmel believes that the set of shards which we collect over a lifetime sum together to describe our
true self because he believes in authenticity, as
did Plato long before him, and Martin Heidegger
after him, among others.
While Simmel postulated that earnest self-actualization would cause the collection of a persons
shards to converge upon his true individuality,
the post-Freudian psychoanalyst Jacques Lacan
went so far as to deny that there could be any
such true individualhe carried forth the idea
that the ego (self) is always constructed in the
Other (culture and worlds materials). From
Lacans work, a mediated construction theory of
identity was bornthe idea that who we are is
wholly fabricated out of cultural materials such
as language, music, books, film plots, etc. Other
popular renditions of the idea that language (e.g.,

ontologies of music, books, etc.) controls thought

include the Sapir-Whorf hypothesis, and George
Orwells newspeak idea in his novel 1984. Today,
mediated construction theory is carried forth primary by the literature of feminist epistemology,
but it is more or less an accepted idea.
At the end of the day, Simmel and Lacan have
more in common than differences. Csikszentmihalyi and Rochberg-Halton (1981), succeed in the
following reconciliation. Their theory is that the
objects that people keep in their homes, plus the
things that they like and consume, constitute a
symbolic environment which both echoes (Simmel) and reinforces (Lacan) the owners identity.
In our work, we take a persons social network
profile to be this symbolic environment which
gives a true representation of self.
If we accept that a user profile can give a
true representation of self, there remains still the
question of closure. Besides all being liked by a
person, do the interests in his bag of interests have
coherence amongst themselves? If it is the case
that people tend toward a tightly unified ethos, or
aesthetic closure, then all the interests in a persons
bag will be interconnected, interlocked, and share
a common aesthetic rationale. If there is aesthetic
closure, then it will be fair for our approach to
regard every pair of interest co-occurrences on
a profile to be significant. If we know there is
not any closure, and that people are more or less
arbitrary in what interests they choose, then our
approach would be invalid.
Our common sense tells us that people are not
completely arbitrary in what they like or consume,
they hold at least partially coherent systems of
opinions, personalities, ethics, and tastes, so there
should be a pattern behind a persons consumerism. The precise degree of closure, however, is
proportional to at least a persons ethicalness and
perhaps his conscientiousness. In his Ethics (350
B.C.E.), Aristotle implied that a persons possession of ethicalness supports closure because ethics
lends a person enkrasia or continence, and thus
the ability to be consistent. Conscientiousness, a

1525

Unraveling the Taste Fabric of Social Networks

dimension of the Big Five personality theory (John,

1990), and perhaps combined with neuroticism, a
second dimension in the same theory, would lead a
person to seek out consistency of judgment across
his interests. They need not all fall under one genre,
but they should all be of a comparable quality
and enjoy a similarly high echelon of taste. Grant
McCracken (1991) coined the term Diderot Effect
to describe consumers general compulsions for
consistencyfor example, John buys a new lamp
that he really loves more than anything else, but
when he places it in his home, he finds that his
other possessions are not nearly as dear to him,
so he grows unhappy with them and constantly
seeks to upgrade all his possessions such that he
will no longer cherish one much more than the
others. Harkening to the Romantic hermeneutics
of Friedrich Schleiermacher (1809/1998), we might
seek to explain this compulsion for uniformity
as a tendency to express a unified emotion and
intention across all aspects of personhood. Indeed,
McCracken termed this uniformity of liking the
various things we consume, Diderot Unity. Diderot
Unity Theory adds further support to our premise
that for the most part, a persons bag of interests
will have aesthetic closure.

Upper bounds on Theoretical Ideal

From the above discussion, we could conclude
a theoretically ideal situation for our taste-mining approach (1) a users bag of interests is an
authentic and candid representation of what the
user really likes, and (2) none of the interests
are out-of-place and there is strong aesthetic
closure and share taste which binds together all
of the interests in the bag. Here, we raise three
practical problems which would degrade the
theoretically ideal conditions, thus, constituting
an upper bound; however, we suggest that these
would degrade but not destroy our theoretical
premise, resulting in noise to be introduced into
the inference of the taste fabric.

1526

A first corruptive factor is performance. Erving

Goffman (1959) posed socialization as a theatrical performance. A social network is a social
setting much like Goffmans favorite example
of a cocktail party, and in this social setting, the
true self is hidden behind a number of personae
or masks, where the selection of the mask to
wear is constrained by the other types of people
present in that setting. Goffman says that we
pick our mask with the knowledge of those surrounding us, and we give a rousing performance
through this mask. In other words, the socialness
of the social network setting would rouse us to
commit to just one of our personae, and to give a
dramatic performance in line with that persona.
Performance might strength aesthetic closure, but
it could also be so overly reductive that the bag
of interests no longer represent all of the aspects
of the persons true identity.
A second corruptive factor is publicity. In her
ethnographic review of the Friendster social networking site, Danah Boyd (2004) raises concerns
over the quality and truth of profiles in light of
the fact that a personal profile is public, not only
to strangers, but also to ones high school friends,
college friends, professors, ex-girlfriends, and
coworkers alike. Because social networking sites
generally make a profile visible to all these different social circles at once, Boyd suggests that
some users are cowed to the fear of potentially
embarrassing exposurefor example, teacher
exposing to his students, or teenager exposing to
his mother. As a result, users may be cowed into a
lowest-common-denominator behavior, sanitizing
the personal profile of all potentially embarrassing,
incriminating, or offensive content.
Finally, a third corruptive factor also raised
by Boyd, concerns the integrity and timeliness
of social networks themselves. Boyd claims that
Friendster profiles and friend connections are not
frequently updated, leading to stale information
which could distort the taste fabric if we were
interested in looking at the temporal animation
of the fabric. Boyd also writes about a phenom-

Unraveling the Taste Fabric of Social Networks

enon known as Fakestersthe creation of bogus

user profiles such as for celebrities. However, the
scope of Fakesters is arguably limited, and since
Fakesters are chiefly imitations of actual people,
aesthetic closure should still be observed and
learning over Fakester profile examples should not
greatly compromise the integrity of the meaning
implied by the taste fabric.

Identity Keywords vs. Interest

Keywords
While each social network has an idiosyncratic
representation, the common denominator across
all the major web-based social networks we have
examined is the representation of a persons broad
interests (e.g., hobbies, sports, music, books,
television shows, movies, and cuisines) as a set
of keywords and phrases. But in addition, more
than just interests, higher-level features about a
person such as cultural identities (e.g., raver,
extreme sports, goth, dog lover, fashionista) are also articulated via a category of special
interests variously named, interests, hobbies
& interests, or passions.
As shown in the web page layout of the personal
profile display (Figure 1), the special interests
category appears above the more specific interest
categories. We suggest that this placement encourages a different conceptualization for the special
interests categoryas a container for descriptions
more central to ones own self-concept and selfidentification. Of course, syntactic and semantic
requirements are not enforced regarding what
can and cannot be said within any of these profile
entry interfaces, but based on our experiences,
with the exception of those who are intentionally
tongue-and-cheek, the special interests category
is usually populated with descriptors more central
to the self than other categories. For example, a
person may list Nietzsche and Neruda under
the books category, and reading, books, or
literature under the special interests category.

In the profile normalization process, identity

descriptors are inferred from descriptors listed
under the special interests category (e.g., dogs
Dog Lover, reading Book Lover,
deconstruction Intellectual).
Theoretically speaking, it is desirable to have
two different granularities of description for a
person. Identity descriptors are more general and
constitute a far smaller ontology than interest
descriptors, thus, the resulting effect is to create
a taste fabric structured according to a hub-andspoke topology. Identity descriptors serve as hubs
and interest descriptors serve as spokes. The advantages to such an organization are revealed in a
later section on applications, and in the evaluation
of the taste fabric in a recommendation task.
Having established a theoretical premise for
mining taste fabric from social network profiles,
and having argued for identity descriptors to be
separate from interest descriptors, the following
section dives into the architecture and techniques
of the taste fabric implementation.

Weaving the Taste Fabric

The implementation of the taste fabric making
system was completed in approximately 3,000
lines of Python code. As depicted in Figure 2,
the architecture for mining and weaving the taste
fabric from social network profiles can be broken
down into five steps: (1) acquiring the profiles
from social networking sites, (2) segmentation of
the natural language profiles to produce a bag of
descriptors, (3) mapping of natural language fragment descriptors into formal ontology, (4) learning
the correlation matrix, and (5) discovering taste
neighborhoods via morphological opening, and
labeling the network topology. The following
subsections examine each of these phases of processing more closely. A condensed description of
the mining process, sans neighborhood discovery,
can be found in Liu and Maes (2005a).

1527

Unraveling the Taste Fabric of Social Networks

Figure 2. Implemented architecture of the taste fabric maker

Acquiring Profiles from Social

Networking Sites
The present implementation of the taste fabric
sources from a one-time crawl of two Web-based
social network sites, which took place over the
course of six months in 2004. The shortcoming
of this approach is that we were only able to mine
100,000 personal profiles from the sites, and only
approximately 80% of these profiles contained
substantive content because about 20% of users
elected to not make their profile details publicly
visible to our robotic crawler. Also, the one-time
crawls prevents us from being able to engage in
more interesting dynamic tracking of profiles
which would potentially allow us to animate taste
fabrics through time. At press time, we are in discussions with two social network Web sites to gain
research access to their user profiles, which should
allow for the implementation that we discuss in
this paper to be deployed on a larger scale.

1528

At every step of mining, we are careful not to

compromise the anonymity or personal information of social network users. In fact, in the end
product, all traces of individual users, as well as
their idiosyncratic speech, are purged from the
taste fabric. From our 100,000 seed profiles, we
used only the text of the categorical descriptors
and none of the personal information including
names and screen names. We chose two social
networks rather than one, to attempt to compensate
for the demographic and usability biases of each.
One social network has its membership primarily
in the United States, while the other has a fairly
international membership. Both however, have
nearly identical descriptor categories, and both
sites elicit users to specify punctuation-delimited
descriptors rather than sentence-based descriptors. One cost to mining multiple social networks
is that there is bound to be some overlap in their
memberships (by our estimates, this is about
15%), so these twice-profiled members may
have disproportionately greater influence on the
produced fabric.

Unraveling the Taste Fabric of Social Networks

Segmenting Profiles
Once profile texts are acquired, these texts need
to be segmented. First, texts are easily segmented
based on their interest categories. Recall in Figure 1 that texts are distributed across templated
categories, e.g., passions/general interests, books,
music, television shows, movies, sports, foods,
about me. Experience with the target social
network websites tell us that most users type
free-form natural language text into about me,
and natural language fragments for the specific
interest categories. For the passions/general interest category, text is likely to be less structured
than for specific interest categories, but still more
structured than about me. Perhaps this is due to
the following psychologyfor specific interests,
it is clear what the instances would be, e.g., film
names, director names, and film genres for the
films category, yet for the general interest category,
the instance types are more ambiguousso that
field tends to elicit more idiosyncratic speech.
For each profile and category, its particular
style of delimitation is heuristically recognized,
and then applied. Common delimitation strategies
were: comma-separated, semicolon-separated,
stylized character sequence-separated (e.g. item
1 \../ item 2 \../ ), new line separated, commas
with trailing and, and so on. Considering a successful delimitation as a category broken down

into three or more segments, approximately 90%

of specific categories were successfully delimited,
versus about 75% of general categories. We did not
attempt to segment about me. Unsegmentable
categories were discarded.

Ontology-Driven Natural Language

Normalization
After segmentation, descriptors are normalized by
mapping them into a formal ontology of identity
and interest descriptors (Figure 3). Newly segmented profiles are represented as lists containing
casually-stated natural language fragments referring to a variety of things. They refer variously to
authorships like a book author, a musical artist,
or a filmmaker; to genres like romance novels,
hip-hop, comedies, French cuisine; to titles
like a books name, an album or song, a television show, the name of a sport, a type of food; or
to any combination thereof, e.g., Lynchs Twin
Peaks, or Romance novels of Danielle Steele.
To further complicate matters, sometimes only
part of an authors name or a title is given, e.g.,
Bach, James, Miles, LOTR, The Matrix
trilogy. Then of course, the items appearing
under the general interest categories can be quite
literally anything.
Figure 3 presents the ontology of descriptor
instance types for the present taste fabric. At the

Figure 3. Table of instance type ontology and data sources

1529

Unraveling the Taste Fabric of Social Networks

top-level of the ontology are six specific interest

categories plus one general interest category (i.e.,
identities). Also, as shown, there are roughly 25
second-level ontological types. There are a total
of 21,000 recognizable interest descriptors, and
1,000 recognizable identity descriptors, sourcing
from ontologies either scraped or XML-inputted
from The Open Directory Project (dmoz)1, the
Internet Movie Database (imdb)2, TV Tome3, TV
Guide4, Wikipedia5, All Music Guide6, AllRecipes7, and The Cooks Thesaurus8. Figure 3 only
lists the primary sources, and lists them in order
of descending saliency. The diversity and specificity of types ensures the maximal recognition
capability over the free-form natural language
in the profiles.
Ontologizing identity. The ontology of 1,000
identity descriptors required the most intensive
effort to assemble together, as we wanted them
to reflect the types of general interests talked
about in our corpus of profiles; this ontology was
hand-engineered out of a few nomenclatures and
folksonomiesmost prominently Wikipedias
extensive list of subcultures and The Open Directory Projects hierarchy of subcultures and
hobbies. We also generated identity descriptors
in the form (blank) lovers where blank was
replaced with major genres in the rest of our ontology, e.g., book lovers, country music lovers,
etc. Some profiles simply repeat a select subset
of interest descriptors in the identity descriptors
category, so having the (blank) lovers template
would facilitate the system in recognizing these
examples. The mapping from the general interest
category into the identity descriptors ontology is
far more indirect a task than recognizing specific
interests because the general interest category
does not insinuate a particular ontology in its
phrasing. Thus, to facilitate indirect mapping,
each identity descriptor is annotated with a bag
of keywords which were also mined out from
Wikipedia and The Open Directory Projectso
for example, the Book Lover identity descriptor
is associated with, inter alia, books, reading,

1530

novels, and literature. Because we employed

two parallel mechanisms for identity descriptors,
i.e., cultures versus (blank) lovers, we cannot be
completely assured that these do not overlapin
fact, they are known to overlap in a few cases, such
as Book Lovers and Intellectuals or Indie
Rock Music Lovers (genre of music) and Indie
(subculture). Most cases of overlap, however, are
much more justified because the cultural lexicon,
just as natural language, cannot be flattened to
a canon. Perhaps the most debatable choice we
made wasfor the sake of bolstering recognition
ratesup-casting descriptors until they could be
recognized in the identity ontology. For example,
while Rolling Stones is not in the ontology of
identity descriptors, we automatically generalize
it until it is recognized, or all generalizations are
exhaustedin the case of Rolling Stones, it is
up-cast into Classic Rock Music Lovers.
Popularity-driven disambiguation. To assist
in the normalization of interest descriptors, we
gathered aliases for each interest descriptor and
statistics on the popularity of certain items (most
readily available in The Open Directory Project)
that the system uses for disambiguation. For
example, if the natural language fragment says
simply Bach, the system can prefer the more
popular interpretation of JS Bach over CPE
Bach.
Situated semantic recognition. Once a profile has been normalized into the vocabulary of
descriptors, they are relaxed semantically using
a spreading activation (Collins & Loftus, 1975)
strategy over the formal ontology, because more
than simply being flat wordlists, the ontological
instances are cross-annotated with each other to
constitute a fabric of metadata. For example, a
musical genre is associated with its list of artists,
which in turn is associated with lists of albums,
then of songs. A book implies its author, and a band
implies its musical genre. Descriptors generated
through metadata-association are included in the
profile, but at a spreading discount of 0.5 (read:
they only count half as much). This ensures that

Unraveling the Taste Fabric of Social Networks

when an instance is recognized from free-form

natural language, the recognition is situated in
a larger semantic context, thus increasing the
chances that the correlation learning algorithm
will discover latent semantic connections.
In addition to popularity-driven disambiguation of, e.g., Bach into JS Bach, we also
leverage several other disambiguation strategies.
Levenshtein (1965/1966) edit distance is used to
handle close misspellings such as letter deletions,
consecutive key inversions, and qwerty keyboard
near-miss dislocations, e.g., Bahc into Bach.
Semantically empty words such as articles are allowed to be inserted or deleted for fuzzy matching,
e.g. Cardigans into The Cardigans (band).
Using this crafted ontology of 21,000 interest descriptors and 1,000 identity descriptors,
the heuristic normalization process successfully
recognized 68% of all tokens across the 100,000
personal profiles, committing 8% false positives
across a random checked sample of 1,000 mappings. Here, tokens refer to the natural language
fragments outputted by the segmentation process;
a recognition is judged successful if after stripping
away semantically empty words, the token finds
correspondence with an instance in the ontology,
while remaining within the heuristically-specified
tolerances for misspelling and popularity-driven
disambiguation. We suggest that this is a good
result considering the difficulties of working with
free text input, and with an enormous space of
potential interests and identities.

Correlation: Weaving the Raw Fabric

From the normalized profiles now each constituted
by normalized identity and interest descriptors, correlation analysis using classic machine
learning techniques reveals the latent semantic
fabric of interests, which, operationally, means
that the system should learn the overall numeric
strength of the semantic relatedness of every pair
of descriptors, across all profiles. In the recommender systems literature, our choice to focus

on the similarities between descriptors rather

than between user profiles reflects an item-based
recommendation approach such as that taken by
Sarwar et al. (2001).
Technique-wise, the idea of analyzing a corpus
of profiles to discover a stable network topology
for the interrelatedness of interests is similar to
how latent semantic analysis (Landauer, Foltz, &
Laham, 1998) is used to discover the interrelationships between words in the document classification
problem. For our task domain though, we chose to
apply an information-theoretic machine learning
technique called pointwise mutual information
(Church & Hanks, 1990), or PMI, over the corpus
of normalized profiles. For any two descriptors
f1 and f2, their PMI is given in equation (1). The
probability of a descriptor, Pr( f i), is defined here
as the frequency of global occurrences of f i divided
by the summed frequency of global occurrences
for all descriptors.
(1)
Looking at each normalized profile, the
learning program judges each possible pair of
descriptors in the profile as having a correlation,
and updates that pairs PMI. What results is a
22,00022,000 matrix of PMIs, because there
are 21,000 interest descriptors and 1,000 identity descriptors in the ontology. After filtering
out descriptors which have a completely zeroed
column of PMIs, and applying thresholds for
minimum connection strength, we arrive at a
12,00012,000 matrix (of the 12,000 descriptors,
600 are identity descriptors), and this is the raw
interest fabric. This is too dense to be visualized
as a semantic network, but we have built less dense
semantic networks by applying higher thresholds
for minimum connection strength, and this is the
reason why small clusters seem to appear in the
InterestMap taste fabric visualization.

1531

Unraveling the Taste Fabric of Social Networks

Criticism and limitations. A common critique

heard about our approach is one that questions the
efficacy of using the PMI metric for association.
It has been suggested that we should look at collocations of greater rank than binary. Following
our initial InterestMap publication, we extended
the work by using morphological opening plus
thresholding, as is done in image processing, to
try to discover larger blocks of collocations which
we call neighborhoods. This is to be discussed
imminently. Additionally, another suggestion we
are considering at press is negative collocation,
that is, the collocation of a descriptors absence
with other descriptors. This would address an
apparent flaw of pointwise mutual information,
which is that it overvalues frequent forms
(Deane, 2005), and would shed a new interpretation on the Semiotician Ferdinand de Saussures
structuralist enouncement that meaning must be
negatively defined (1915/1959).

Looking at Topological Features

The raw fabric has two extant topological features
worthy of characterizationidentity hubs and
taste cliques. In addition, we describe what we
believe to be a novel application of mathematical
morphology (Serra, 1982; Haralick, Sternberg,
& Zhuang, 1987) in conjunction with spreading
activation (Collins & Loftus, 1975) to discover the
taste neighborhoods we hinted at in Section 1.
Identity hubs behave like seams in the fabric.
Far from being uniform, the raw fabric is lumpy.
One reason is that identity hubs pinch the network. Identity hubs are identity descriptor nodes
which behave as hubs in the network, being more
strongly related to more nodes than the typical
interest descriptor node. They exist because the
ontology of identity descriptors is smaller and
less sparse than the ontology of interest descriptors; each identity descriptor occurs in the corpus
on the average of 18 times more frequently than
the typical interest descriptor. Because of this
ratio, identity hubs serve an indexical function.

1532

They give organization to the forest of interests,

allow interests to cluster around identities. The
existence of identity hubs allows us to generalize
the granular location of what we are in the fabric,
to where in general we are and what identity hubs
we are closest to. For example, it can be asked,
what kinds of interests do Dog Lovers have?
This type of information is represented explicitly
by identity hubs.
Taste cliques as agents of cohesion. More
than lumpy, the raw fabric is denser in some
places than in others. This is due to the presence
of taste cliques. Visible in Figure 5, for example,
we can see that Sonny Rollins, is straddling
two cliques with strong internal cohesion. While
the identity descriptors are easy to articulate
and can be expected to be given in the special
interests category of the profile, tastes are often
a fuzzy matter of aesthetics and may be harder
to articulate using words. For example, a person
in a Western European taste-echelon may fancy
the band Stereolab and the philosopher Jacques
Derrida, yet there may be no convenient keyword
articulation to express this. However, when the
taste fabric is woven, cliques of interests seemingly governed by nothing other than taste clearly
emerge on the network. One clique for example,
seems to demonstrate a Latin aesthetic: Manu
Chao, Jorge Luis Borges, Tapas, Soccer,
Bebel Gilberto, Samba Music. Because the
cohesion of a clique is strong, taste cliques tend
to behave much like a singular identity hub, in
its impact on network flow. In the following Section, we discuss how InterestMap may be used
for recommendations, and examine the impact
that identity hubs and taste cliques have on the
recommendation process.

Carving Out Taste Neighborhoods

with Mathematical Morphology
From the raw fabric, another step of processing is
needed to reveal taste neighborhoodspatches
of taste cohesion that are larger than taste cliques

Unraveling the Taste Fabric of Social Networks

and more stable than ephemeral taste ethoi.

Taste neighborhoods of course, overlap with
one another, and the discovery and definition
of taste neighborhoods seems even prone to
the Ptolemaic dilemmasome nodes must be
designated as center of the universe, and the
choice of these centric nodes can greatly affect
the resultant neighborhood definition. Two taste
neighborhoods with different Ptolemaic centers
are shown in Figure 4.
Taste ethos from spreading activation. While
the technical details for the discovery process
are potentially lengthy, we sketch a conceptual
overview of the implementation here. The raw n
by n correlation matrix is reviewed as a classic
spreading activation network (Collins & Loftus,
1975). That is to say, activation spreads outward
from an origin node to all the connected nodes,
then from all connected nodes to each of their
connected nodes. The obvious observation here
is that in our correlation situation, all nodes are
connected to a large percentage of the graph, so our
graph is super-connected. However, what makes
the spreading activation meaningful is that the
strength of the spread activation is proportional
to the strength of the PMI along any edge in the
graph. The energy of the spreading is also inhib-

ited as the number of hops away from the origin

grows, according to a per hop discount rate (e.g.,
50%) So, spreading with a low tolerance (or, a
high threshold for activation), and outward from
Jazz, Yoga (two-hops away) is reachable, but
the energy attenuates before the Football (also,
two-hops away) node can be activated.
Spreading activation outward from an origin
node, the result can be likened to that nodes
defeasible (default, in the absence of other inputs or biases) taste ethos. This taste ethos is too
small when spreading activation is configured
with a modest tolerance. On the other hand, if
the tolerance is increased too dramatically, the
taste ethos will grow in size but its stability will
be undermined due to this well-known problem
in graph-based inference: beginning at two hops
away, reached nodes lose their semantic relevance
to the origin node very rapidly. Think of this as
the telephone game effectplaying the childhood
game of telephone, the first couple of hops are
still recognizable, but recognition often rapidly
tapers off after the first couple of hops. The effect
is also observed by Google in their PageRank
algorithm (Brin & Page, 1998) for scoring the
salience of web pages by voting. They noted that
high-rank pages tended to link to high-quality

Figure 4. Two Ptolemaically-centered taste neighborhoods, computer generated with the follow parametersa maximum of 50 nodes in each neighborhood, up to the first three instances of any category type
are shown. Spatial layout is not implied by the neighborhood; nodes are manually arranged here.

1533

Unraveling the Taste Fabric of Social Networks

pages, and those to other high-quality pages, but

after distance=2, reliability tapered off rapidly.
Mathematical morphology. To discover neighborhoods of taste which are larger than particular
node-centric ethoi, but which are still nonetheless stable, we borrow two techniques from the
field of mathematical morphology (Serra, 1982;
Haralick, Sternberg & Zhuang, 1987) and that are
widely used in the image processing literature
which appropriates them for object segmentationmorphological opening and thresholding.
Morphological opening is the mathematical composition of two operators erosion and dilation, in
that order. The intuition is that erosion eats away
at the boundaries of an object, whereas dilation
grows the boundaries of the object. However,
erosion and dilation are not inverses because both
erosion and dilation are injective, that is, they are
many-to-one and lossful transformations. The
effect of morphological opening is also quite intuitiveit removes small objects disturbances
and opens up gaps when they are located near a
boundary. There is morphological opening and
there is also morphological closing which is dilation composed with erosion; closing fills in holes
and around boundaries more than opening. We
employ opening because it is a bit crisper. Opening eliminates blemishes while closing magnifies
blemishes. The other technique, thresholding, is
frequently used to post-process an opened image.
Applying a fixed threshold to an opened image
simply turns every pixel above the threshold to
1, and below the threshold to 0.
Erosion and dilation over spread activations.
We choose identity nodes as the centric origins for
spreading because they are in general more stable
places to start from. This follows the rationale
that identities are stronger cultural fixtures than a
book or a music album, generally speaking. From
the identity nodes, we apply a relatively lenient
discount, e.g., 0.8, and spread to define a fairly
relevant neighborhood. This is repeated over all
identity nodes, begetting an ethos for each identity
node. Where ethoi overlap, the max of the nodes

1534

energy is taken, rather than the sum of the nodes

energies. Now, erosion is applied, trimming back
the weakest boundary nodes, followed by a dilation,
growing the boundary by adding some energy to
all nodes connected to the boundary, pushing some
of them over the activation threshold and thus
growing the mass. In the current implementation,
two iterations of opening are performed, though
the meaning of this varies widely with the choice
of thresholds and other considerations.
In this manner, larger stable masses of nodes,
termed taste neighborhoods, are discovered.
Thresholding can help us trim a neighborhood to
an arbitrary node-size. For visualizations such as
InterestMap, neighborhoods comprised of up to
thirty nodes seem visually appropriate. We believe
that the application of morphological opening and
thresholding to a spreading activation network in
order to discover larger stable neighborhoods is a
novel use, though we do not evaluate this claim
within this papers scope.
Summary. This section discussed an implementation of weaving the interest fabric out of
social networks. Profiles mined from two social
network websites were heuristically segmented
and normalized according to a heterogeneous
ontology assembled together from a variety of
data sources. After normalization, correlation
analysis learned the affinities between descriptors, and mathematical morphology over the
raw fabric enabled taste neighborhoods to be
discovered and overlaid onto the fabric. Next, we
demonstrate the semantic uses of the taste fabric
within application contexts.

What is a Taste Fabric Good

for?
As a rich tapestry of interconnected interests
and identities, the taste fabric brings us closer to
improvisational manipulations of meaning, and
affords us at least three semantic functionsthe
creation of semantically flexible user representa-

Unraveling the Taste Fabric of Social Networks

tions, cross-domain taste-based recommendation,

and the computation of taste-similarity between
people. This section explores these three basic
semantic functions in the context of a survey of
three applications we have developed. InterestMap
is a taste-based recommendation system that leverages interactive visualization of neighborhoods to
make the recommendation mechanism transparent, thereby enhancing users trust perceptions
of the recommender. Ambient semantics uses

the taste fabric to facilitate social introductions

between two strangers, based on their shared
taste. IdentityMirror is a digital mirror for identity
self-management. Whereas a real mirror shows
you what you look like, IdentityMirror shows you
who you are. It explores semantically flexible user
representations by allowing time, orderliness,
and current events in the world to nuance the
representation of the viewer.

Figure 5. Two screenshots of the InterestMap interactive visualization. 5a (top) depicts a user browsing
neighborhoods of taste visually. 5b (bottom) depicts a user visualizing his own taste ethos by dragging
and connecting interesting nodes to the who am i? node.

1535

Unraveling the Taste Fabric of Social Networks

InterestMap
InterestMap (Liu & Maes, 2005a) visualizes the
topology of the taste fabric, and in particular
it depicts taste cliques, identity hubs, and taste
neighborhoods as a navigatable map. As shown in
Figure 5a, users can browse InterestMaps tapestry
of neighborhoods, cliques and identity hubs, or,
as depicted in Figure 5b, they can interactively
build up their own taste ethoi, by searching for
and attaching descriptors to a stationary who
am i? node. The act of connecting a descriptor
to the self is deeper than making a mere superficial keyword association since each descriptor is
actually something more like a semantic cloud.
Once a user has connected several descriptors to
his self, those semantic clouds begin to intersect,
overlap, and mingle. They begin to imply that
other descriptors, which the user has not selected
himself, should be within the users taste. Hence,
the notion of a visual recommendation.
Taste-based recommendation. InterestMap
can, given a profile of the users interests, recommend in a cross-domain way, books, songs,
cuisines, films, television shows, and sports to that
user based on taste. The users interests are normalized according to aforementioned processes
and mapped into the taste fabric. These nodes
in the fabric constitute a particular activation
configuration that is unique to the user, and the
total situation described by this configuration
is the fuzzy taste model of the user. To make
recommendations, activation is spread outward
from this configuration, into the surrounding
nodes. Some nodes in the surrounding context
will be activated with greater energy because
they are more proximal to the taste signature of
the starting configuration. The nodes activated
with the highest energy constitute the users
recommendation. Figure 5b shows a visualization of the recommendation process. The users
self-described interests are the descriptors directly
connected to the who am i? node. Each of these
interests automatically entails other strongly

1536

connected descriptors. This is visually expressed

well in the InterestMap visualization because a
node getting pulled toward who am i? will tug
a whole web of nodes behind it. Since the visualization starts with just the who am i? node
visible on the screen, specifying just a couple
of interests can literally fill up the screen with
its taste entailments. To visualize the spreading
activation mechanism, the size and yellowness of
nodes diminishes as activation spreads outward
from the who am i? node.
Visual recommendation enhances transparency and trust. That a user trusts the recommendations served to him by a recommender system is
important if the recommender is to be useful and
adopted. Among the different facilitators of trust,
Wheeless and Grotz (1977) identify transparency
as a prominent desirable property. When a human
or system agent discloses its assumptions and
reasoning process, the recipient of the recommendation is likely to feel less apprehensive toward
the agent and recommendation. Also in the spirit
of transparency, Herlocker, Konstan, and Riedl
(2000) report experimental evidence to suggest
that recommenders which provide explanations of
its workings experience a great user acceptance
rate than otherwise.
Unlike opaque statistical mechanisms like collaborative filtering (Shardanand & Maes, 1995),
InterestMaps mechanism for recommendation
can be communicated visually. The idiosyncratic
topology of this taste fabric symbolizes the common taste tendencies of a large group of people.
For example, in Figure 5a, it is plain to see that
Sonny Rollins and Brian Eno are each straddling two different cliques of different musical
genres. The rationale for each recommendation,
visually represented as the spreading of flow
across the network, is easily intelligible. Thus it
may be easier for a user to visually contextualize
the reasons for an erroneous recommendation,
e.g., I guess my off-handed taste for Metallica
situated me in a group of metal heads who like all
this other stuff I hate.

Unraveling the Taste Fabric of Social Networks

Although we have not yet implemented such a

capability, the ability to interact with the InterestMap network space would also afford the system
an opportunity to learn more intelligently from
user feedback about erroneous recommendations.
Rather than a user simply stating that she did not
like a particular recommendation, she could black
out or deprecate particular clusters of the network
which she has diagnosed as the cause of the bad
recommendation, e.g., Ill black out all these taste
cliques of heavy metal and this identity hub of
Metal Heads so the system will not make that
mistake again.

Ambient Semantics
Ambient Semantics (Maes et al., 2005) is a wearable contextual information system that supports
users in discovering objects and meeting people
through pithy just-in-time feedback given in the
crucial first moments of an encounter. Here is an
example of a use case involving the discovery of a
new book: Wearing the Ambient Semantics RFID
reader wristband, you pick up a copy of Marvin
Minskys Society of Mind book. Through
your cell phone display, the system tells you that
you would be particularly interested in section 3
because it is relevant to your current research topics. It would tell you that your friends Henry and
Barbara listed this book among their favorites, and
that the authors expressed opinions seem sympathetic to your own, based on semantic analyses of
both your writings. The system can indicate that
you would find the book tasteful because it can
use taste fabric to detect that it is indeed within
close proximity to your taste ethos, translating to
a strong taste-based recommendation.
Exposing shared taste-context between two
strangers. The second use case concerns the system facilitating social introductions by breaking
the ice. This scenario demonstrates using the taste
fabric for the quantification and qualification of
the taste-similarity between two strangers. First,
a scenario. You are at a business networking event

where Ambient Semantics wristwatches have been

given to the attendees. You are tired of the same old
conversation starterswhats your namewho
do you work forhow do you like it here?so
you head to the Ambient Semantics kiosk where
people are meeting each other in a new way. You
introduce yourself to a lady standing next to you.
By virtue of your handshake, the physical surroundings are transformed. The music and lighting in the area change to suit the shared aspects
of yours and the ladys tastes. Some graphics of
kayaking are thrown up on the kiosk display, as
well as the faces of some people. The lady says
to you, so you know Bob and Terry too? Are you
in the Boston Outdoor Society too?
Calculating taste-similarity: quantitatively
vs. qualitatively. There is more than one good way
to use taste fabric to calculate the taste-similarity
of two people. The more direct way is to measure
the intersection of two spread activations. Taking each persons seed profile of interests and
mapping it into the taste fabric, we arrive at an
initial configuration. Spreading activation outward from this configuration defines a semantic
neighborhood, which earlier in the paper we
referred to as a persons taste ethos. Taking the
semantic intersection of two or more persons
ethoi, we arrive at the quantitative calculation of
taste-similarity.
However, another intriguing possibility is
to make a qualitative calculation about tastesimilarity. Although the intersection of two taste
ethoi is mathematically satisfying, it is not easily explainable and articulated. In other words,
having the system explain that the two of you
share taste because you both have interests x,
y, and z in your spreading activation clouds
is inappropriate. More articulate would be to
cite a shared habitation of taste neighborhoods,
for example, this explanationthe two of you
share taste because both of you are adventurers
and lovers of wine. Here, the mechanism of
the recommendation feels more transparent. To
calculate qualitative similarity, each persons

1537

Unraveling the Taste Fabric of Social Networks

taste ethos would be used to score the degree

of a persons habitation across the various taste
neighborhoods, which as you recall, are centered
around identity nodes. Like the classic k-nearest
neighbors classification scheme, here we classify
persons by their k-nearest taste neighborhoods.
Having completed this mapping, the subset of
neighborhoods shared among the two or more
persons become those persons shared situation.
To communicate shared neighborhoods to the
persons, the neighborhoods could be effectively
visualized on a screen, or, neighborhoods are
safely summarized by stating the identity nodes
which live within that neighborhood.

IdentityMirror
What if you could look in the mirror and see
not just what you look like, but also who you
are? Identity mirror (Figure 6) is an augmented
evocative object that reifies its metaphors in the
workings of an ordinary mirror. When the viewer
is distant from the object, a question mark is
the only keyword painted over his face. As he
approaches to a medium distance, larger font
sized identity keywords such as fitness buffs,
fashionistas, and book lovers identify him.
Approaching further, his favorite book, film, and
music genres are seen. Closer yet, his favorite
authors, musicians, and filmmakers are known,

and finally, standing up close, the songs, movies,

and book titles become visible.
The Identity Mirror learns and visualizes a
dynamic model of a users identity and tastes.
Looking into it, the viewers face is painted
over with identity and keywords, sourced from
this dynamic user model. Taste fabric is used to
interpret an initial seed profile into a semantic
situation within the fabric. For instance, the viewer
specifies that he listens to Kings of Convenience
and enjoys the fiction of Vladmir Nabakov, and
using this, taste fabric situates the viewer within
its multiple neighborhoods of taste. The keywords
which paint over the viewers face represent his
context within taste-space.
Dynamic model of taste. The richness of
connections in the taste fabric allow for a truly
dynamic model of a users tasteone that can
evolve over time, and can absorb the influence
of each mornings world events. First, we explain
how world events effect the dynamic user model.
In Liu (2003), one of the authors gave a model of
context-sensitive semantic interpretation for noun
phrases. The phrase fast car, under a default
context, would mean a car that is moving fast,
but uttered at a carwash, it could also mean a
car that is washed fast. Similarly, a persons
identity can be interpreted differently based on
each mornings news. For example, supposing
that the morning news reveals an international

Figure 6. Three screenshots of one of the authors gazing into the IdentityMirror. (left) Far away, only general identities can be seen; (center) at mid-distance, favorite music, book, and film genres emerge; (right)
finally, up-close, all of the details and specific interests in the viewers taste ethos become visible.

1538

Unraveling the Taste Fabric of Social Networks

conspiracy, that event could tease out from the

users taste ethos her one-time penchant for Tom
Clancy mystery and suspense novels, even though
on any other day, that part of her taste would not
have been visible. IdentityMirror implements a
feature to allow world events to bias who you
look like. It operationalizes this by parsing each
mornings news feeds for major topics, and activating those topics in the taste fabric as contextual
biases. When the users taste ethos is generated
by spreading activation away from a starting
configuration, the activation would now flow in
an unusual way because of the new contextual
biases. Thus, the image of the users identity has
been biased by world events.
Second, we explain how the users model
can evolve over time by recording the history of
changes that a viewer makes to himself in the
mirror. By gazing into IdentityMirror, a viewer
can glean his identity-situation. Is his hair out of
place? Are one of his interests out of place? How
do his facial features combine to compose a gestalt? How do his various interests come together
to compose an identity or aesthetic gestalt? We
implement a feature called Identity Fixing which
allows a viewer to fix himself as he would fix his
hair. Keywords are distributed between a hearth
(keywords that are taste-cohesive) and a periphery
(outlier keywords seemingly out-of-place about a
person); the hearth covers the face, the periphery
covers the hair. A person with a strong degree of
taste-coherence has ruly hair, whereas a postmodernist with scattered interests has unruly hair. The
viewer can use his hands to adjust his hairhe
can dishevel those unwanted peripheral keywords,
or accept them by packing them into his hair. In
the user model, rejecting a keyword de-energizes
that keyword in the users taste ethos, whereas
affirming the keyword energizes that keyword
in the users taste ethos. As world events tease
out new and hidden aspects to the viewers taste
over time, and as the viewer continues to fix his
identity, over time, the viewers identity will be
well groomed and even well vetted.

Advanced Discussion
In this section, we present an evaluation of the
taste fabric, present related work, and discuss
other ways in which this work is of consequence
to the semantic mining and Semantic Web communities.

Evaluation
We evaluate the quality of the taste fabric apropos a telos of recommendation, scrutinizing
the performance of recommending interests via
spreading activation over the taste fabric, as
compared with a classic collaborative filtering
recommender. Much of this discussion is adapted
from (Liu & Maes, 2005a).
In this evaluation, we introduced three controls
to assess two particular features: (1) the impact that
identity hubs and taste cliques have on the quality
of recommendations; and (2) the effect of using
spreading activation rather than a simple tally of
PMI scores. Notably absent is any evaluation for
the quality of the produced taste neighborhoods,
because here we consider only quantitative and not
qualitative recommendation. Qualitative recommendation is not claimed to outperform quantitative recommendation in terms of accuracyour
suggestion was that linguistically identifying and
visually illustrating two persons cohabitations
of taste neighborhoods should facilitate trust and
transparency in the recommenders process.
In the first control, identity descriptor nodes
are simply removed from the network, and spreading activation proceeds as usual. In the second
control, identity descriptor nodes are removed,
and n-cliques9 where n>3 are weakened10. The
third control does not do any spreading activation,
but rather, computes a simple tally of the PMI
scores generated by each seed profile descriptor
for each of the 11,000 or so interest descriptors.
We believe that this successfully emulates the
mechanism of a typical non-spreading activation

1539

Unraveling the Taste Fabric of Social Networks

item-item recommender because it works as a

pure information-theoretic measure.
We performed five-fold cross validation to
determine the accuracy of the taste fabric in
recommending interests, versus each of the
three control systems. The corpus of 100,000
normalized and metadata-expanded profiles was
randomly divided into five segments. One-by-one,
each segment was held out as a test corpus and the
other four used to train a taste fabric using PMI
correlation analysis. The final morphological step
of neighborhood discovery is omitted here.
Within each normalized profile in the test
corpus, a random half of the descriptors were
used as the situation set and the remaining half
as the target set. Each of the four test systems
uses the situation set to compute a complete
recommendationa rank-ordered list of all
interest descriptors; to test the success of this
recommendation, we calculate, for each interest
descriptor in the target set, its percentile ranking
within the complete recommendation list. As
shown in (2), the overall accuracy of a complete
recommendation, a(CR), is the arithmetic mean
of the percentile ranks generated for each of the k
interest descriptors of the target set, ti.
(2)
We opted to score the accuracy of a recommendation on a sliding scale, rather than requiring that descriptors of the target set be guessed
exactly within n tries because the size of the target
set is so small with respect to the space of possible guesses that accuracies will be too low and
standard errors too high for a good performance
assessment. For the TASTEFABRIC test system
and control test systems #1 (Identity OFF) and
#2 (Identity OFF and Taste WEAKENED), the
spreading activation discount was set to 0.75). The
results of five-fold cross validation are reported
in Figure 7.
The results demonstrate that on average, the
full taste fabric recommended with an accuracy of

1540

Figure 7. Results of five-fold cross-validation of

taste-fabric recommender and three control systems on a graded interest recommendation task.

0.86. In control #1, removing identity descriptors

from the network not only reduced the accuracy
to 0.81, but also increased the standard error by
38%. In control #2, removing identity descriptors and weakening cliques further deteriorated
accuracy slightly, though insignificantly, to 0.79.
When spreading activation was turned off, neither
identity hubs nor taste cliques could have had any
effect, and we believe that is reflected in the lower
accuracy of 73%. However, we point out that since
control #3s standard error has not worsened, its
lower accuracy should be due to overall weaker
performance across all cases rather than being
brought down by exceptionally weak performance
in a small number of cases.
We suggest that the results demonstrate the
advantage of spreading activation over simple
one-step PMI tallies, and the improvements to
recommendation yielded by identity and taste
influences. Because activation flows more easily
and frequently through identity hubs and taste
cliques than through the typical interest descriptor node, the organizational properties of identity
and taste yield proportionally greater influence
on the recommendation process; this of course,

Unraveling the Taste Fabric of Social Networks

is only possible when spreading activation is

employed.

Related Works
A cultural metadata approach to musical taste.
Whitman and Lawrence (2002) developed a
metadata model for characterizing the taste
coherence of musical genres. Mining adjectival
and noun phrases collocated with musical artist
discussions in newsgroups and chatrooms, they applied machine learning to automatically annotate
music artists with what they termed community
metadata. Then Whitman and Smaragdis (2002)
applied community metadata to build cultural
signatures for music genres that could be used, in
conjunction with the auditory signal, to classify
unknown artists based on style similarity. Their
notion of a metadata signature for musical styles is
sympathetic to our notion of taste ethos and taste
neighborhood, and both systems take a bottom-up
metadata-driven view of meaning definition. A
chief difference between our two works is that
taste knowledge is located in descriptive wordchoice in their system (e.g., wicked, loud), and
located in interest-choice in our system, that is,
the choices of what people consume (e.g., Britney
Spears, Oreo cookies).
Social information filtering. In prior work,
one of the authors co-developed a well-known
technique for item recommendation based upon
nearest taste-neighbor, the approach known variously as social filtering, or collaborative filtering.
Shardanand and Maes (1995) represent users
as vectors of (item, rating) pairs, and compute
taste-similarity as statistical correlation between
user vectors, or alternatively as cosine similarity
of vectors in n-dimensional item space. In their
Ringo social music browsing system, users were
recommended a list of potential tastemates on
the basis of taste-similarity. One difference between our two approaches is that social filtering
maintains distinct user profiles, whereas taste
fabrics dissolves user boundaries, and is, in their

terminology, a content-based filtering approach.

In distilling a reusable knowledge resource out
of social network profiles that can be reappropriated for a variety of other purposes not concerned
with the original social network community, it is
necessary to protect the privacy of the original
users, and we suggest that taste fabrics serves
as a model for doing so. Also relevant is Sarwar
et al.s (2001) item-based collaborative filtering
approach to recommendation, which, like taste
fabrics, relies upon item-item correlation rather
than user-user correlation. Taste fabric exceeds
item-based filtering by use of extensive metadata
to relax the meaning from the item itself, by
defining identity descriptors as supernodes, and
by representing users as k-nearest neighborhoods.
In general, collaborative filtering is more representation-opaque whereas spreading activation
over neighborhoods can be visualized and more
easily debugged.
Social network analysis and relational mining.
Much research has examined the explicit structure
of social networks, and studied their topologies
via graph theory. Newman (2001) mined scientific
coauthorship networks and found that collaborations funneled through gatekeeper scientists.
In taste fabrics, identity hubs, and hubs created
around particularly salient interest descriptors
constitute a similar topological feature. Jensen
and Neville (2002) mined structured metadata
relations from the Internet Movie Database (imdb.
com) called schema and learned a Bayesian
network model to represent and predict item
distances probabilistically. They also model the
relational semantics of social network relations
implied between movie actors from the Internet Movie Database and the Hollywood Stock
Exchange (www.hsx.com). Finin et al. (2005)
examine how the FOAF (friend-of-a-friend)
ontology applies Semantic Web concepts to enable efficient exchange of and search over social
information, illustrating how social networks
could develop with its semantics already explicit.
Finally one work which considers the semantic

1541

Unraveling the Taste Fabric of Social Networks

content entailments of social network users is McCallum, Corrada-Emmanuel, and Wangs (2005)
modeling of Author-Recipient-Topic correlations
in a social network messaging system. Given the
topic distributions of email conversations, the
ART model could predict the role-relationships
of author and recipient. The work considers group
clusters and dyadic relationship dynamics but does
not consider cultural aggregates as is the concern
of our present work.
Large-scale commonsense knowledge networks. Taste fabrics are a rich tapestry which
define the meaning space of taste and interests.
They are represented as semantic networks and
reasoning is performed via spreading activation
over this network. This approach to knowledge
representation and reasoning builds upon previous
work in large-scale semantic knowledge bases such
as WordNet (Fellbaum, 1998) and ConceptNet (Liu
& Singh, 2004). WordNet is a semantic network
whose nodes are words, and edges are various
nymic lexical relations between the words, e.g. a
dog has the hypernym of canine. ConceptNet
is a semantic network of commonsense knowledge whose 200,000 nodes are verb phrases (eat
burger, take shower), and 1.6 million edges
are one of 20 kinds of world semantic relations
(e.g., EffectOf, PropertyOf, DesireOf), e.g.,
(EffectOf be hungry cook food). ConceptNet
and taste fabrics reason similarly by activating
a seed configuration of nodes, and spreading
activation outward to define a semantic context.
Both resources are densely connected, semantically extensive within their respective domains,
and allow for improvisational manipulations of
meaning to take place atop them.

Reusable Methodologies
Sanitary semantic mining. The sanitariness of a
mined knowledge resource is the degree to which
it is purged of idiosyncrasy, especially idiosyncratic traces of user-specific information, and
also idiosyncrasies which implicate the original

1542

application domain from which the resource was

mined. When a knowledge resource is sanitary,
assurances can be made that private user data is
not recoverable, and that the resource is sufficiently
context-free so that it could potentially be used
to solve problems across a variety of domains.
Taste fabrics are an illustration of how a sanitary
knowledge resource can be mined out of a highly
idiosyncratic and application-specific data source
such as self-descriptive social network profiles.
Because it is sanitized, taste fabrics can be publicly
distributed and used to power applications living
in other domains.
When mining social network data, concern
for privacy and copyrights of user data make
derivative works especially problematic; yet there
is a great need and opportunity to infer valuable
semantic knowledge from these sources. Ensuring data anonymity in the produced knowledge
resource is a particularly sensitive issue. An early
phase of the taste fabric construction process is
to normalize the casually-stated keywords and
phrases into formal ontologies of non-idiosyncratic form (e.g., Nietzsche into Friedrich
Nietzsche, dogs appearing under the passions category into Dog Lover). Already, the
unrestricted idiosyncratic language which bears
traces of an authorship are beginning to be wiped
away. In contrast, collaborative filtering systems
maintain ratings for each user, and while users do
not have to be named, even unnamed users are
not anonymous, they are only pseudonymous. A
users name is simply wiped away and replaced
with a unique id (renamed from John Smith to
User #123), but the profiles integrity is intact.
Because the number of instances is quite large in
the space of tastes, it may be possible to recover
the identities of pseudonymized users because
the constitution of profiles are quite unique.
At the very least, maintaining any information
structured around the notion of a user lends itself
to the perception that privacy of the source data
may be violated.

Unraveling the Taste Fabric of Social Networks

Rather than preserving individual profiles, the

taste fabric simply uses these profiles to learn the
strengths of connections on a network whose nodes
already exist (they are simply an exhaustively
enumeration of all features in the ontology). The
method of the learning is non-linear so explicit
frequency counts cannot easily be recovered.
Thresholding and neighborhood definition are
further lossful transformations which make
details of the original application data virtually
unrecoverable. The final structure is sanitaryit
assures the anonymity of the data source, and is
much easier to distribute.
Instance-based semantic webs and ethotic
representation. In the Semantic Web community,
ontology and metadata systems are often seen as
top-down and bottom-up approaches to knowledge
representation, respectively. To draw parallels
with the artificial intelligence literature, ontology
is a category-based representation, and metadata
is a feature-based representation. However, taste
fabrics introduces the notion of an instance-based
representation, which we feel to be a promising
methodology for the Semantic Web community
that warrants further study, especially into the
issue of scalability. An instance-based representation lacks categories or features, having only items
and dense numerical connections between them.
Knowledge is thus unpacked from the linguistic
symbolism of a category or features name, and
instead, is found in connectionismthe flow of
semantics through a graph of items. The shift
from symbolic interpretation toward continuous
interpretation parallels Zadehs efforts in attempting to soften the bivalence of logic representation
by giving a fuzzier, more continuous account of
meaning (Zadeh, 2004).
Instance-based representations are more appropriate for semantic recognition and semantic
mediation because they offer continuous numerical interpretation of entity similarity. In taste
fabrics, users, groups of users, and cultures can
all be represented uniformly as clouds of node
activations in the fabric. A taste fabric allows the

meaning of a users keyword profile to be relaxed

into a semantic cloud which we term an ethos.
Using ethotic representation, semantic mediation
between two users or entities in the fabric can be
computed quite easily as shared activation, and
even effectively visualized. By interpreting an
ethos as a membership into k-neighborhoods, the
resource can be used to classify users or entities
into an ontology of neighborhoods (the organizing
force of ontology, in fact, is still present in the
resource via neighborhoods and identity descriptors). Instance-based representations and ethotic
representations would be well-suited for semantic
resources meant for mediation and classification
in the Semantic Web.

Conclusion
This paper presented a theory and implementation
of taste fabricsa semantic mining approach
to the modeling and computation of personal
tastes for lifestyle, books, music, film, sports,
foods, and television. Premised on philosophical
and sociological theories of taste and identity,
100,000 social network profiles were mined,
ontologically-sanitized, and a semantic fabric
of taste was weaved. The taste fabric affords a
highly flexible representation of a user in tastespace, enabling a keyword-based profile to be
relaxed into a spreading activation pattern on
the taste fabric, which we termed a taste ethos.
Ethotic representation makes possible many
improvisational manipulations of meaning, for
example, the taste-similarity of two people can
be computed as the shared activation between two
ethoi. Taste-based recommendation is already
implied by a taste ethos, as all items within an
ethos are intrinsically relevant to the taste of the
individual. Indeed, an evaluation of recommendation using the taste fabric implementation shows
that it compares favorably to classic collaborative
filtering recommendation methods, and whereas
collaborative filtering is an opaque mechanism,

1543

Unraveling the Taste Fabric of Social Networks

recommendation using taste fabrics can be effectively visualized, thus enhancing transparency
and cultivating user trust.
Two models of taste-based recommendationone quantitative based on shared activation,
and one qualitative based on k-nearest neighborhoodswere presented. Recommendation, time
and world-sensitive user representation, and interpersonal taste-similarity, were illustrated within
a survey of three applications of taste fabrics.
This paper makes three contributions to the
literature. First, it presents a novel mechanism for
mining and modeling the taste-space of personal
identities and interests. Second, the mining and
weaving of taste fabrics from idiosyncratic social
network profiles raises the issue of sanitation of
knowledge resources, and this paper illustrated
how ontology and non-linear correlation learning
can be used to purge idiosyncrasy and prepare
a general-purpose grade knowledge resource.
Finally and third, in addition to ontology-based
and metadata-based knowledge resources, taste
fabrics introduces a novel third approach to the
literatureinstance-based fabrics, where the
notion of knowledge is a purely relational one.
Fabrics, we suggest, excel at semantic mediation,
contextualization, and classification, and may
play a valuable role as a context mediator in a
recently complicated Semantic Web of formal,
semi-formal, and now, informal, entities.

Acknowledgments
This research was supported by a British Telecom Fellowship, an AOL Fellowship, and by the
research consortia sponsors of the MIT Media
Lab.

References
Aberer K., et al. (2004). Emergent semantics. Proc.
of 9th International Conference on Database

1544

Systems for Advanced Applications (DASFAA

2004), LNCS 2973 (pp. 25-38). Heidelberg.
Aristotle. (350 BCE). Nichomachean Ethics.
Boyd, D. (2004). Friendster and publicly articulated social networks. Conference on Human
Factors and Computing Systems (CHI 2004).
ACM Press.
Brin, S., & Page, L. (1998). The anatomy of a
large-scale hypertextual Web search engine.
Computer Networks and ISDN Systems, 30(1-7),
107-117.
Church, K.W., & Hanks, P. (1990). Word association norms, mutual information, and lexicography.
Computational Linguistics, 16(1), 22-29.
Collins, A.M., & Loftus, E.F. (1975). A spreading-activation theory of semantic processing.
Psychological Review, 82, 407-428.
Csikszentmihalyi, M., & Rochberg-Halton, E.
(1981). The meaning of things: Domestic symbols and the self. Cambridge, UK: Cambridge
University Press.
Deane, P. (2005). A nonparametric method for
extraction of candidate phrasal terms. Proceedings of ACL2005.
Fellbaum, C. (Ed.). (1998). WordNet: An electronic
lexical database. MIT Press.
Finin, T., Ding, L., Zhou, L., & Anupam J. (2005)
Social networking on the Semantic Web. The
Learning Organization: An International Journal,
12(5), 418-435.
Goffman, E. (1959). The presentation of self in
everyday life. Garden City, NY: Doubleday.
Haralick, R.M., Sternberg, S.R., & Zhuang, X.
(1987). Image analysis using mathematical morphology. IEEE Transactions on Pattern Analysis
and Machine Intelligence, 9(4), 532-550.
Herlocker, J., Konstan J., & Riedl, J. (2000). Ex-

Unraveling the Taste Fabric of Social Networks

plaining collaborative filtering recommendations.

Conference on Computer Supported Cooperative
Work (pp. 241-250).
Jensen, D., & Neville, J. (2002). Data mining
in social networks. National Academy of Sciences Symposium on Dynamic Social Network
Analysis.
John, O.P. (1990). The Big Five factor taxonomy:
Dimensions of personality in the natural language
and in questionnaires. In L. A. Pervin (Ed.),
Handbook of personality: Theory and research
(pp. 66-100). New York: Guilford.
Landauer, T.K., Foltz, P.W., & Laham, D. (1998).
An introduction to latent semantic analysis. Discourse Processes, 25, 259-284.
Levenshtein, V. (1965/1966). Binary codes capable
of correcting deletions, insertions, and reversals,
Doklady Akademii Nauk SSSR, 163(4), 845-848,
1965 (Russian). English translation in Soviet
Physics Doklady, 10(8), 707-710.
Liu, H. (2003). Unpacking meaning from words:
A context-centered approach to computational
lexicon design. In Blackburn et al. (Eds.), Modeling and Using Context, The 4th International and
Interdisciplinary Conference, CONTEXT 2003,
LNCS 2680 (pp. 218-232). Springer.
Liu, H., & Maes, P. (2005a, Jan 9). InterestMap:
Harvesting social network profiles for recommendations. Proceedings of IUI Beyond Personalization 2005: A Workshop on the Next Stage
of Recommender Systems Research, San Diego,
CA (pp. 54-59).
Liu, H., & Singh, P. (2004). ConceptNet: A practical commonsense reasoning toolkit. BT Technology Journal, 22(4), 211-226.
Maes, P., et al. (2005). Ambient semantics and
reach media. IEEE Pervasive Computing Magazine. Submitted.

McCallum, A., Corrada-Emmanuel, A., & Wang,

X. (2005). Topic and role discovery in social
networks. Proceedings of 19th International
Joint Conference on Artificial Intelligence (pp.
786-791).
McCracken, G. (1991). Culture and consumption:
New approaches to the symbolic character of
consumer goods and activities. Indiana University Press.
McCracken, G. (1997). Plenitude. Toronto: Periph:
Fluide.
Newman, M. (2001). Who is the best connected
scientist? A study of scientific coauthorship networks. Phys. Rev., E 64.
Sarwar, B.M., et al. (2001). Item-based collaborative filtering recommendation algorithms.
The 10th Intl World Wide Web Conference (pp.
285-295). ACM Press.
Saussure, Ferdinand de (1915/1959). Course in
general linguistics (W. Baskin, Trans.). New York:
McGraw-Hill.
Schleiermacher, F. (1809/1998). General hermeneutics. In A. Bowie (Ed.), Schleiermacher:
Hermeneutics and criticism (pp. 227-268). Cambridge University Press.
Serra, J. (1982). Image analysis and mathematical
morphology. London: Academic Press.
Shardanand, U., & Maes, P. (1995). Social information filtering: Algorithms for automating
word of mouth. Proceedings of the ACM SIGCHI
Conference on Human Factors in Computing
Systems (pp. 210-217).
Sheth, A., Ramakrishnan, C., & Thomas, C.
(2005). Semantics for the Semantic Web: The
implicit, the formal and the powerful. International Journal on Semantic Web and Information
Systems, 1(1), 1-18.
Simmel, G. (1908/1971). How is society possible?
In D. N. Levine (Ed.), On individuality and social

1545

Unraveling the Taste Fabric of Social Networks

forms: Selected writings. University of Chicago

Press.
Staab, S., Santini, S., Nack, F., Steels, L., &
Maedche, A. (2002). Emergent semantics. IEEE
Intelligent Systems, 17(1), 78-86.
Wasserman, S. (1994). Social network analysis:
Methods and applications. Cambridge University
Press.
Wheeless, L., & Grotz, J. (1977). The measurement of trust and its relationship to self-disclosure.
Communication Research, 3(3), 250-257.
Whitman, B., & Lawrence, S. (2002). Inferring
descriptions and similarity for music from community metadata. In Voices of Nature, Proceedings of the 2002 International Computer Music
Conference (pp. 591-598).

Whitman, B., & Smaragdis, P. (2002). Combining

musical and cultural features for intelligent style
detection. Proceedings of the 3rd International
Conference on Music Information Retrieval.
Zadeh, L.A. (2004, Fall). Precisiated natural
language. AI Magazine.

Endnotes


3

4

5

6

7

8

9

1
2

10

http://www.dmoz.org
http://www.imdb.com
http://tvtome.com
http://tvguide.com
http://www.wikipedia.org
http://www.allmusic.com
http://allrecipes.com
http://www.foodsubs.com
A qualifying clique edge is defined here
as an edge whose strength is in the 80th
percentile, or greater, of all edges.
By discounting a random 50% subset of the
cliques edges by a Gaussian factor (0.5 mu,
0.2 sigma).

This work was previously published in International Journal on Semantic Web & Information Systems, Vol. 2, Issue 1, edited
by A. Sheth & M. Lytras, pp. 42-71 , copyright 2006 by IGI Publishing, formerly known as Idea Group Publishing (an imprint
of IGI Global).

1546

1547

Chapter 4.8

Computer-Mediated
Communication:
Enhancing Online
Group Interactions
J. Michael Blocher
Northern Arizona University, USA

Abstract

Background

Advancements in information technology have

transformed much in all aspects of todays society. In particular, synchronous and asynchronous
electronic communication systems support countless interactions every day. However, these communication systems have specific limitations that
can cause miscommunications as well. Because
of that, computer-mediated communications
(CMC) has been a robust research agenda in many
disciplines. This is especially true of education,
where online learning has become common place
in all but a few universities, thus requiring learners to interact via CMC within virtual learning
environments. This chapter will use educational
CMC research as a lens to promote and support
an understanding of how to better utilize and
facilitate electronic communication, regardless
of the field or endeavor.

Over a decade ago, Rheingold (1993) described

how the Internet, specifically, synchronous and
asynchronous electronic communication could
create a non-linear and level environment that
provides the conduit for human interaction that
is culturally neutralwhere members meet in
virtual communities judged by ideas, thoughts,
and contributions, rather than by race, gender,
age, physical appearance, or national origin. In
particular, CMC was described as a venue where
participants could engage in discussions on bulletin boards and listservs with equal impunity
and have the opportunity to create a cyber community. Due to the lack of visible and contextual
cues, participants engaging in this, somewhat,
anonymous form of communication were free
from prejudice and able to speak their mind more
freely and openly than in traditional face-to-face

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Computer-Mediated Communication

group interactions. However, this turned out be

only somewhat correct, and the literature provides
ample examples of how CMC group interactions
follow similar patterns and styles that have been
identified within traditional face-to-face group
interactions.

Communication Style Differences

Communication style differences, particularly,
gender-based differences have been identified
as one area that has been investigated in more
historic group interaction literature. Specifically, women tend to engage in group interactions with a style distinctly different than men.
Eakins and Eakins (1978) in researching gender
styles within group interactions discovered that
women tend to be more engaged in developing
intimacy and maintaining relationships than their
male counterparts. In other words, women tend
to be more interested in building relationships
and developing closeness within their group than
men do. For example, women are more apt to ask
for anothers opinion, desiring to engage others
within the conversation. Men, on the other hand,
tend to be more directive and less apt to draw out
the interaction (Fishman, 1983). Perhaps this is
because, in general, women are more interested
in the social aspects of group interaction. For
example, Carli (1984) reported that women are
more social-oriented, while men are more taskoriented in their interactions. Furthermore, Briton
and Hall (1995) reported that women tended to
be better at using nonverbal cues, being more
expressive, utilizing body language, eye contact,
and gestures to send and receive subtle messages
to promote and enhance communication in group
interactions. Men, on the other hand, leaned
toward competitiveness and dominance seemed
more apt to be more competitive and less socially
motivated in their group interactionsusing
fewer subtle nonverbal cues.
In terms of group interactions where productivity is at issue, Maguire (1999) reported that

1548

within studies of marketing executives women

work more to build consensus while men tended
to make faster decisions. One could argue that
one style is better than the other at supporting
communication in group interactions, as both
have benefits and limitations. For example, being
more direct could be seen as a way to improve
communication by not clouding the issue with
possibly vague nonverbal cues. On the other hand,
by providing more information via other channels
(e.g., body language, facial expressions, etc.), the
verbal message might be further enhanced, with
less chance of misunderstandings. Furthermore,
by utilizing a communication style that seems
competitive and dominating, issues of power may
cloud or hinder the real message. Again, on the
contrary, using less direct verbal communication
could be seen to hinder ones ability to provide
leadership. Which style makes a better communicatorone who builds community or consensus
through social group interactions enhanced with
non-verbal cues, or one who improves productivity
with more directive style communication?
It could be argued that it might be better to
try to understand these communication style
differences, rather than trying to debate which
might be more effective, as each has benefits
and limitations. This is especially important
within the CMC environment. As we will see in
the next section, CMC group interactions have
other issues based upon the various CMC media,
which can greatly impact these more traditional
communication style differences.

CMC Communication Style

Differences
Of course, these gender differences are generalities. Ones individual group communication style
is just that, individual. However, these genderrelated communication style differences seem
to parallel a CMC environment as well. As in
all social environments, membership within a
CMC group environment can be dominated by

Computer-Mediated Communication

more spirited individuals, and certain hierarchies

can develop or emerge based upon the members
interactions within that virtual community. Of
particular note, Herring (1993) details that virtual
communication spaces may not be the democratic
spaces described earlierprimarily because of
the communication style differences brought
naturally to the environment by the various group
members and the community ethos that the group
develops. For example, her study reported that
men generally made strong assertions, engaged
in self-promotion, asked rhetorical questions,
came from an authoritative orientation, provided
challenges and humor. Women, on the other
hand, tended to ask questions, provided attenuated assertions, presented apologies, had a
personal orientation and provided support in
their CMC. As discussed above, these two genderbased communication styles mirror those of the
traditional research. Generally, women tend to
foster consensus and community building, while
men tend to demonstrate a communication style
that is more directive in nature.
While this supports the argument that men are
more apt to utilize a more directive CMC style,
perhaps more interesting are specific investigations into chat (synchronous communications)
that have detailed how gender communication
style differences were power-based in particular.
Soukup (1999) stated that in his study, Masculine participants were aggressive, argumentative,
and power oriented. While feminine participants
sought relationships and intimacy, they were often
dominated and overpowered by the aggressive
discourse of the masculine members (p. 169).
Perhaps, even more disturbing, findings were
reported by Herring (1999). She described maledominated harassment of female participants in
two case studies, stating: Due perhaps to the
youth and sexual preoccupations of the majority of its users, Internet relay chat (IRC) tends
to sexualize female participants, and to involve
them in flirtatious interactions. Accordingly, the
ideal for female interactions in IRC appears to be

cooperative flirtation, as it is also in many off-line

recreational social realms (p. 163).
One would think that this type of interaction
might inhibit female members when engaging
in online communication, perhaps making them
more apt to adjust or change their online communication engagement; perhaps even hide or mask
their gender identity in CMC environments to
guard against this type of interaction. To answer
that question, Jaffe,Lee,Huang, andOshagan
(1999) investigated how participants, when given
opportunities, chose to select anonymous pseudonyms for their CMC. Some participants chose
pseudonyms with same gender, others chose
pseudonyms that were cross gender, and finally
some chose pseudonyms that were gender neutral.
Gender-neutral and cross-gender pseudonyms
then collapsed into gender-masking, thus providing two categories as same-gender and gendermasking. They reported that, Seventeen (81%)
of the 21 male participants in the pseudonymous
conference chose pseudonyms that retained their
gender identification. In contrast, 13 (81%) of the
16 females in the pseudonymous conference chose
pseudonyms that masked their gender identification (1999, p. 227). This significant difference (p
< .001) between men and women demonstrates a
greater tendency, by women, to utilize a pseudonym that masked their gender. The authors of
this study saw this as evidence of a direct strategy
to level the playing field stating:
This finding is interpreted to reflect an effort to
rectify an imbalance, felt by women, of social
parity when interacting in mixed-gender situations. An especially pertinent example of previous
gender-based naming practices provides historical justification for such sentiments. In anecdotal
recollections within the scientific research profession, female authors were often referred to by first
and last names, while only initials and last name
identify males. (Jaffe,Lee,Huang, &Oshagan,
1999, p. 230)

1549

Computer-Mediated Communication

The CMC literature provides examples where

group interactions within CMC environments
seem to exacerbate the communication style
differences, perhaps making men aggressive and
making women more engaged in building social
relationships in their communication styles.
However, it seems that if given the opportunity,
women tend to mask their gender for strategic
purposes.

Summary
Although the CMC technology does have the
potential to provide a non-linear and level environment that provides the conduit for human
interaction that is culturally neutral, it seems that
it may not. Indeed, the studies described above
strongly indicate that CMC may not be the level
playing field that it was once thought to be. In
particular, there are communication style differences within the CMC environment that parallel
traditional communication styles, thus creating
a group communication environment where the
communicators interact in similar ways that they
would in traditional face-to-face environments
where gender, power, and communication style
may have a strong impact on ones communication
within group interactions.
It is important to remember that these are
general tendencies and that not all CMC group
interactions are destined to exhibit the type of
interactions described above. Rather these are
tendencies that could impact ones experience
engaging in CMC, which might be exacerbated
by some of the unique communication issues
that are specific to current text-based CMC. For
example, often a message is sent that is totally
benign. However, the message even when taken
simply for the written text may be taken for a
completely different meaning than originally
intended, because the sender and recipient engaged in the communication at different times,
locations, and states of mind. Being more knowledgeable about the tendencies and limitations of

1550

CMC could help limit more adversarial interactions that were based upon miscommunication
due to CMC issues or communication styles. In
particular, becoming aware of and monitoring
ones personal communication style could help
make the individual more thoughtful and strategic about their communiqus, thus reducing the
tendencies for miscommunication to occur. The
next section will discuss specific issues at work
within the current CMC environments that might
hamper communication and strategies that could
be implemented to enhance more equitable and
clear CMC engagement.

Issues
As described, gender, power, and communication
style can impact ones personal communication
interactions within a CMC environment. Although
these elements parallel traditional face-to-face
communication interactions, there are issues
with CMC that are unique to that environment.
In particular, the discussion will focus on current
CMC issues in terms of how asynchronous and
synchronous electronic communication tools limit
the capacity of communication cues and metainformation as well as communication styles in
terms of gender, and whether these issues can be
influenced to help ameliorate those limitations.

Media Richness
Much of the CMC literature suggests that communicating within electronic messaging systems,
which are primarily text based, limits communication because of the lack of other communication cues or components. Media richness theory
(Daft & Lengel, 1986) defines communication
media in terms of face-to-face communication
elements. These elements include immediacy of
feedback, non-verbal, and other cues that enrich
communication. In other words, a medium is
considered richer or thicker when it can sup-

Computer-Mediated Communication

port a greater number of cues to communicate a

persons ideas (e.g., facial expressions, gestures,
voice inflection, etc.). In light of media richness
theory, current CMC is defined as being lean
because much is asynchronous communication,
lacks immediacy of feedback and generally lacks
multiplicity of cues.
In terms of immediacy of feedback, imagine
a face-to-face interaction where participants give
and receive concurrent and instant feedback while
they monitor the message reception and consider
its impact. Compare that to asynchronous CMC
where interactions are often delayed by hours, if
not days. Indeed, can a sender remember the exact
meaning of a message or state of mind they were
in upon sending a message when they receive the
reply a day or two later? Even synchronous CMC
(chat) lacks in immediacy, where there is a lag time
of waiting for the individual to complete their typing. Furthermore, synchronous chat is dependent
upon the individuals keyboarding skills.
In addition to the limitations of asynchronous
communication, text-based CMC also does not
include non-verbal cues, inflection, and other
elements of face-to-face interactions. As most
current CMC messages are text only, they are
devoid of very rich communication cues that can
cause miscommunication. These missing cues can
impact CMC with possible miscommunications in
several ways. For example, often in verbal communication of most languages inflection describes
how the message is to be taken. English speakers
often use inflection to relay whether something
said is a statement or a question. Punctuation
provides cues in the written language to account
for this issue. However, many times in CMC group
interactions, communicators use less than formal
writing styles, which often can cause confusion.
An additional example is sarcasm. While one
might fully understand ones intent of a sarcastic
statement in a verbal interaction, the message can
very easily be mistaken in a written interaction.

Media Richness Impact on

Communication Style
Can media richness in CMC impact ones ability
to communicate, depending upon an individuals
communication style? In other words, if ones
communication style utilizes communication
cues that are missing in the leaner CMC media,
will it create a less equal communication environment? To help answer that question Dennis,
Kinney, and Hung (1999) investigated decisionmaking task performance in terms of media
richness and gender differences. In particular,
they attempted to answer questions of the impact
of media richness on performance in terms of
decision time, decision quality, consensus
change, and communication satisfaction. The
study investigated gender-based teams performing equivocal tasks both in synchronous chat and
face-to-face. As women tend to better utilize and
read non-verbal cues, which are missing in CMC,
they were especially interested in the results of
the impact of gender on the performance tasks.
Teams of all male, all female, and mixed gender
were formed and compared performing equivocal
decision-making tasks in both face-to-face and
chat environments. The results indicated that
only decision time was impacted by the use
of CMC. Indeed, decision quality, consensus
change, and communication satisfaction were
not significantly impacted by the use of CMC.
Interestingly, the all-female team took five times
as long to complete an equivocal task of reaching
a decision utilizing CMC than it did face-to-face.
This supports the idea that CMC has a more
profound impact on women than men due to the
lack of cues that women tend to better utilize as
they needed more time to complete the tasks.
More interesting, however, is that the decision
quality of the all-female team utilizing CMC was
higher than the all-male and mixed-gender team;
second only to the all-female face-to-face decision-making team. In other words, while missing
the nonverbal cues, it took this team much longer

1551

Computer-Mediated Communication

to perform the task, but the female members still

took the time to perform at a high level regardless of mediasuggesting that they could, and
did, modify their engagement to successfully
complete the task.
This study provides evidence that media richness, or leanness of, in this case, does impact
ones ability to engage in CMC group interactions.
More specifically, since women tend to utilize
nonverbal communication cues more often, a
CMC medium that strips out those communication
cues, can provide a distinct gender disadvantage.
However, it is very important to remember here
that these are tendencies. Indeed, anyone who
utilizes a communication style that is supported
by nonverbal communication cues and engages
within a CMC group interaction where those cues
are missing due to the mediums leanness is at a
disadvantageregardless of gender.

CMC Strategies
From the study detailed above it seems that the
participants modified or enhanced their communication within the limited media in some way to
better perform their tasks. While it is clear that
various CMC media have limitations, users can
and do use strategies that can help ameliorate the
limitations of a particular media. There are specific examples of communicators modifying their
communication to better fit a medium. As a case
in point, text shortcuts, such as: btw = by the way,
irl = in real life, cu ltr = see you later, and imho =
in my humble opinion, have been developed and
utilized within in CMC environments providing
evidence of communicators utilizing new methods
to speed up and facilitate their interaction. The
practice of abbreviated-text messaging and instant
messaging has become common place. Indeed,
mobile phone text messaging has become a social
phenomenon practiced by young people in many
parts of the world. For example, Grinter and Eldridge, (2003) describe how British teenagers are
similar to teenagers in other parts of the world in

1552

their use of a meta-language of shortened words

to send text messages for a variety of purposes,
including coordinating future communications
within other media, generally phone calls. This
trend suggests that these young communicators
have developed an understanding and the strategic utilization of various media for a variety but
connected purposes.
The above example shows modifying a simple
communication method to be more strategically
purposeful of the specific communication media.
However, as discussed above, gender-based communication style is much more complex. Will one
utilize CMC strategies to reduce the limitations
of CMC media? Fahy (2003) investigated supportive CMC strategies within an online learning environment to see if participants utilized
strategies that would increase interaction and
reduce interpersonal distance within the leaner
CMC media. Specific strategies were defined
as supportive and were seen to replace or approximate sociolinguistic devices and strategies,
including nonverbal elements normally available
in face-to-face interaction, (Ridley & Avery,
1979) (Fahy, 2003. pg. 5). The results suggest
that these participants were able to and took the
time to utilize important CMC strategies. As
Fahy states: One-quarter (25.9 per cent) of all
interaction in these conferences was classified as
supportive, a proportion which, though it must
be tested against wider and larger conferencing
samples, indicates the importance attached to
the interpersonal in these online interactions
(Fahy, 2003, p. 11). Students who engaged in these
online conferences did, in fact, utilize the CMC
strategies that approximated the missing cues,
thus enriching the lean medium that promoted a
more supportive group interaction.

Gender-Based Communication Style

Identification
While the CMC literature suggests that men and
women tend to utilize different communication

Computer-Mediated Communication

styles in CMC, one might ask: How strong are

these tendencies and can the communication
style give away the authors gender identity in
an anonymous interactive environment? Can
one judge the gender of a CMC author based
on the CMC style without knowing the author
or his/her gender? Does ones awareness of
CMC styles provide a tool to distinctly judge an
authors gender? In other words, is a persons
gender given away by his or her communication
style even in an anonymous CMC environment?
Savicki, Kelley, and Oesterreich (1999) detail a
series of studies where participants were grouped
in male-only, female-only and mixed-gender
groups for CMC interactions. Their anonymous
messages were then provided to participants to
be judged as to the authors gender based upon
the communication style analysis. They analyzed
messages that included a greater number of I
statements, statements of personal opinion, selfdisclosure, and coalition language. Since this
pattern of communication was related to high
levels of satisfaction with the group process and
high levels of group development, it was labeled
the high group development communication style
(HCS) and contrasted with the opposite style, low
group development communication style (LCS)
(p. 187).
As might be predicted, participants were quite
accurate in judging a LCS message authored by
men (.708), while significantly less accurate (p <
0.001) in judging an HCS message authored by
men (.395). In opposition to prediction, however,
participants were significantly less accurate (p
< 0.05) in judging HCS messages authored by
women (.549) than LCS messages authored by
women (.662). These results support the finding
that participants were significantly more accurate when judging LCS messages than HCS
messages, regardless of the gender of the author.
In this study, if a message exhibited HCS it was
more difficult to accurately judge the authors
gender than messages that exhibited LCS. They
also reported that the judges gender or CMC

experience did not seem to factor into the level

of accuracy as no significant differences for these
variables were found.
These mixed results suggest that, while gender
identification may be possible, it is quite limited.
In other words, if authors utilize HCS within
their anonymous CMC messages, there is less
likelihood of being identified as male or female.
If ones communication style includes HCS in
the CMC messages, they may well be taken for
their information and not for their gender. This
may have an impact on helping promote a more
democratic communication environment based
on gender inequities previously discussed.

Gender-Based Communication Style

Training
Certainly, promoting more democratic CMC
group interaction seems a prudent thing to try to
achieve. However, can one be trained, and more
importantly to become motivated, to utilize HCS
to support participation, satisfaction, and group
development within a CMC environment? Savicki,
Kelley, and Ammon (2002) investigated that question by grouping participants in male only (MO),
female only (FO), and mixed gender (MIX). They
examined how HCS training would impact participation, satisfaction, and group development.
The results were mixed, For example, while they
reported that the FO groups scored higher on
group development, it may have been due to the
potential of gender-based communication style
tendencies for that group of females rather than
HCS training. However, HCS training for the
MIX groups had a greater impact than the MO.
The authors state that:
. . . the MIX groups responded somewhat
more to the training than the MO groups. The
combination of men and women in the same
group seemed to change the norms toward more
acceptance of HCS communication. Gender is an
important variable influence communication in

1553

Computer-Mediated Communication

online groups (Savicki & Kelly, 2000). Gender

communication norms seem to be quite strong
and should be taken into account when selecting
participants for virtual teams. (p. 267)
It seems logical to presume, and from the
findings of this study, that it is possible to provide HCS training to impact user engagement.
The MIX groups did provide some evidence of
the success of the training. However, the HCS
training MO groups had less success, according
to the authors, because of previously established
gender norms. Training individuals to change
their communication style for group interactions,
in terms of developing higher group development
communication skills, may be more complex
than first thought. For example, it may not be as
simple as providing HCS training, as more social
communication interaction norms are at play.
However, this study also instigates a question.
What interactions in the MIX groups promoted
consensus or community building to change the
communication norms of accepting increasing
use of HCS communication more than the MO
groups?

Summary
The studies described above detail some of the
major issues of CMC. Specifically, current CMC
systems tend to be lean media in terms of media
richness theory, which can impact certain populations based upon their communication strengths.
However, it also would seem that communicators
believe it is important and take steps to ameliorate
the limitations that the lean media present by
utilizing communication methods and strategies
that are supportive in nature. Interestingly, utilizing higher group development communication
style (HCS) might limit the identification of the
communicator, thus possibly providing a more
level communication venue in terms of gender.
More importantly, however, it seems that the
studies above demonstrate that communication

1554

styles, although tend to be gender-based, may

be a matter of choice. Furthermore, although it
seems possible to provide training to support more
HCS communication, the process may be quite
complex. Indeed, higher group communication
development may be greatly impacted by the
membership of that group. Although this might
seem logical and somewhat simplistic, it could
have very complex implications. For example,
in becoming knowledgeable about, and strategic
in the use of their communication style, CMC
users could become better more effective and
thoughtful communicators by utilizing a style
that is more engaging, supportive, and more likely
to foster group development. However, they may
not be motivated, or perhaps do not believe they
have the permission to do so based upon the
membership and communication norms of their
virtual group.

Recommendations
From the review of the literature there are several
issues that dominate the various common CMC
systems and the participants who engage in group
interactions within them. In particular, there are
issues of media richness and communication
style differences. In addition, there is the issue
of a groups communication norms. These factors
can impact the groups communication norms,
which might be defined by: (A) the makeup of the
group membership, (B) the purpose of the group,
and (C) the media they employ to communicate.
These issues can provide some challenges for
CMC users. Drawing from distance or distributed
learning theory will provide some recommendations to help support electronic communication
by building online communities where group
interaction is enhanced by developing membership within a community of practice. In addition,
the literature suggests CMC strategies that might
help communicators better their electronic communiqus, thus reducing miscommunications

Computer-Mediated Communication

and more equitable group interactions. Finally,

current CMC technologies generally provide for
text-only communication. In particular, there is
the issue of media richness with communication
environments that utilize a suite of tools that include only the traditional text-based synchronous
and asynchronous e-communication systems.
Utilizing multimedia CMC systems that include
audio and/or video in addition to the lean textbased synchronous and asynchronous systems
can help thicken the communication media, thus
alleviating some of the limitations and issues
discussed above. All of these will be discussed
in the next section.

Community Building
As discussed earlier in this chapter, Rheingold
(1993) described virtual communities where
members interact using CMC within a level social
environment. Although this ideal may have been
suggested prior to having a greater understanding
of some of the media limitations and how group
interactions are impacted by them, it does not mean
that the idea should necessarily be abandoned. This
is especially important as group communication
norms can dictate the membership engagement of
communication style. One only needs to look at the
incredible growth in online tools like, MySpace
(n.d.), Orkut (n.d.), and the plethora of blogs to
see that people want to communicate and engage
in CMC group interactions. For example, more
advanced robust open-community CMC systems,
such as BuddySpace, are being developed to
provide the users with greater flexibility, in terms
of connectivity with other platforms. In addition,
BuddySpace provides the users location, which
can help group individuals who might have similar
interests because of their locale. However, as with
any community, online community members
may not be as supportive to one another as they
ultimately could, or perhaps should. Therefore,
it would be wise to understand the elements of

community building as it may impact CMC group

interactions in online communities.
In a very well-established text, Kim (2000)
provides details and recommendations of community-building strategies that are supported by
three underlying design principles: (a) design
for growth and change, (b) create and maintain
feedback loops, and (c) empower members over
time. These principles suggest that the placement
of responsibility of community development begins with the designers or developers rather than
members. However, it supports the cultivation of
member ownership by its three design principles.
For example, designing for growth and change,
allows for the community to be molded by the
feedback received from the members as they gain
empowerment. With the empowerment of the
members over time, the organizational ownership and responsibility of the community will be
cultivated and be accepted by its members. That
it not to say, however, that in an online community, the members cannot be the designers and or
developers. Regardless, in Kims recommendations the members will assume responsibility and
ownership of the community.
Another example of community building is
detailed in the work of Wenger (1998). In defining
a community of practice, he provides a detailed
outline that identifies a more complex view of
communities, stating:
A community of practice defines itself along
three dimensions:

What it is about: Its joint enterprise as

understood and continually renegotiated by
its members.
How it functions: Mutual engagement
that binds members together into a social
entity.
What capability it has produced: The
shared repertoire of communal resources
(routines, sensibilities, artifacts, vocabulary,
styles, etc.) that members have developed
over time. (p. 54)

1555

Computer-Mediated Communication

Wengers description of a community of

practice suggests that members bind together in
a social entity by sharing routines, sensibilities,
artifacts, vocabulary, and style. Wenger goes on to
say, Even when a communitys actions conform
to an external mandate, it is the communitynot
the mandatethat produces the practice. In this
sense, communities of practice are fundamentally
self-organizing systems (p. 2). The concepts
outlined in Wengers community of practice have
very interesting implications in light of the HCS
training findings of Savicki et al. (2002). In that
study, the mixed-gender groups moved toward
building a community of practice by changing
their norms to accept and use more HCS in their
CMC within their group interactions. However,
one also could argue that the male-only group
members also moved toward building a community of practice by not changing their groups
communication norms to accept and use the HCS
communication strategies. The HCS training
had little or no effect on the male-only group.
On the contrary, the group maintained their
communication style regardless of the training,
perhaps because of the strong influence of the
male group communication norms. The external
mandate, as Wenger describes, could be seen as
the training to change to HCS communication,
which was not accepted by the male-only groups,
but was more so by the mixed-gender groups.
In this case, the CMC norms were developed
and determined by the group. This supports the
idea, in terms of online communities, that the
ownership and responsibility for the practice of
the community lays within the members of the
community, not the administrators or developers
of an online space.
To provide an example in practice, let us look
at designing and developing an online course
of study, which utilizes CMC systems to build
communities of practice that support enhanced
group interactions. It is important to remember
that online course delivery has its roots in the
distance learning model originally developed

1556

that utilized the traditional mail system to send

materials to learners in remote locations in the
early part of the 20th century. As technology
developed, course material delivery moved to
more advanced systems: radio, television, CDROM, and, eventually, to the Internet. As the
delivery methods advanced, opportunities for
learner interaction became greater because of
the possibilities for group interactions using
CMC. Distance learners were no longer limited
to primarily one-way information dissemination
for their learning experience.
Distributed learning is a term that has been
used to replace distance learning by some. Dede
(2004) defines distributed learning communities
where learners engage collaboratively to develop
knowledge as they interact within the various electronic communication and information delivery
systems. In this concept of community building,
group interactions are distributed via CMC and
other more advanced electronic tools, to help
members actively engage with one another for
the specific purpose of learning. Here, learners
within the distributed learning community utilize a variety of resources in a highly structured
learning environment. Learning management
systems, such as WebCT (n.d.) and BlackBoard
(n.d.) include the tools to support this definition
of distributed learning.
Although the use of the CMC systems will
provide the opportunity for group interactions
within a distributed learning environment, the
members within the community are still responsible for their interaction and, thus, for their
learning. Therefore, designing and developing an
online course of study that utilizes distributed
learning elements is more complex than using a
system to make course materials and opportunities for group interaction available. The key to a
distributed learning environment, accordingly,
is building communities of practice through the
utilization of multiple electronic communication
systems that advances the opportunities for the

Computer-Mediated Communication

members to engage, thus making CMC the vital

means of group interaction.
In this example, the designer, developer, and/or
the instructor begin by selecting the interactive
communication systems to utilize. They also will
need to design learning activities and assignments that will have the learners engage with
the course materials and activities. Finally, they
also provide guidelines for group interactions the
learners will have with the instructor and their
peers. Palloff and Pratt (1999) described effective
strategies for building and the cultivation of an
online learning community that fosters greater
group interaction that augments online learning.
They outline a framework for online instruction
that provides for: (a) mutually negotiated guidelines, (b) focused outcomes, (c) active creation
of knowledge and meaning, and (d) collaborative
learning. Interestingly, these elements echo the
communities of practice work of Wenger (1998) in
terms of sharing the responsibility of community
building and, in this case, learning throughout the
membership of the community. All of the members,
including learners and instructors, become responsible for developing the community of practice.
For example, by promoting mutually negotiated
guidelines, learners are given the opportunity to
share ownership of the learning environment that
will bind the members together in a joint enterprise. Learners utilize CMC to communicate with
their peers and instructors to mutually produce
the shared repertoire of communal resources and
learning guidelines that will structure their social
entitythe learning community. Through their
collaborations they actively create knowledge and
meaning based on the focused outcomes, as defined
by the community. Designing an online learning
environment with these four strategies will then
support the members in building a community of
practice for learning that utilizes the pedagogy and
technology of distributed learning. Here members
become responsible for and have the tools to bind
together in a social entity where they share routines, sensibilities, artifacts, vocabulary, and style.

As learners collaboratively build their learning

community of practice through the construction
of artifacts, vocabulary, routines, sensibilities, and
style, they are building shared knowledge.
The idea of building shared knowledge is at the
heart of the distributed cognition work of Hutchins
(2000). Distributed cognition theory suggests that
knowledge and knowledge acquisition go beyond
the individual. Rather, knowledge is constructed
through the social interactions, where individuals
share and co-construct knowledge. In other words,
knowledge is not constructed by the individual
alone, but instead by the interacting members engaged in the learning process. This is an important
characteristic of distributed learning that supports
the concept of communities of practice producing
a shared repertoire of communal resources (routines, sensibilities, artifacts, vocabulary, styles,
etc.) that members have developed over time (p.
54). If the joint enterprise of that online group
is that of learning, then membership within that
community supports and fosters shared knowledge
construction. This suggests that, in terms of community building, through their group interactions
the social entity and its members build knowledge
as they build community.
This section has focused on the elements
of community building within online groups,
emphasizing that through the development of
an online community of practice where group
interaction is enhanced by members negotiating
the communication norms and behaviors and
working collaboratively to build and distribute
knowledge. In doing so, through enhanced group
interactions the social entity builds community,
shares elements and resources that support the
individuals within that community. Therefore,
to support the initial building of community, it
is suggested that administrators and developers
of those groups provide ample tools and, perhaps
structure, but that it is ultimately the membership
that makes the rules. This does not fully address
issues of power, as individuals who tend to be
more aggressive may still dominate discussions

1557

Computer-Mediated Communication

and communications. Although the CMC administrators of the online community could censure
inappropriate individuals, and some communities may decide that some form of policing is
appropriate, still, here, the membership should
be responsible for the decision making of that
policy. It is recommended that by building an
online community of practice, inequities, and
disagreements would be settled by the membership of that community, based upon that groups
agreed-upon norms. The next section will outline
various communication strategies that will help
reduce miscommunications that can often be the
cause for online disagreements.

T2 (b) answers or comments

T3 Reflections
T4 Scaffolding
T5 References, Authorities
T5 (a) references, quotations, paraphrases
T5 (b) citations or attributions
In a later study, Fahy (2003) utilized a specific
subset of the categories from the TAT, listed
above, to investigate participant use of supportive
communication strategies. The supportive communication strategies investigated included:

Utilizing CMC Strategies

Building a community that shares and co-constructs knowledge could be enhanced by what has
been learned about CMC strategies. Although,
as has been argued above, the membership of an
online community is responsible for setting the
communication norms of their group, it still seems
prudent to be aware of helpful communication
strategies. For example, if the membership of the
online community creates routines, sensibilities,
artifacts, vocabulary, styles, as Wenger (1998) suggests, it would seem that communicators would
benefit from understanding, being aware of, and
utilizing CMC strategies within their messages
that support more personal interactions. By doing
so, there may be less miscommunication, which
can often erupt into disagreements.
To better understand communication interactions, Fahy, Crawford, and Alley (2001) developed
the transcription analysis tool (TAT), which
provides five CMC interaction categories, three
of which have subcategories.
T1 Questioning
T1 (a) vertical
T1 (b) horizontal
T2 Statements
T2 (a) direct

1558

T1 (b) horizontal questions, which promote

more discussion as this type of question does
not presume one correct answer, but rather
invites any plausible answer.
T2 (b) referential statements, message
statements that directly or indirectly refer
to anothers message, thus engaging more
than just the author of the message.
T4 scaffolding and engaging comments
including:
Acknowledgement: Recognizing or
acknowledging comments or statements from prior messages.
Agreement: Expressing agreement.
Apology, or self-criticism: Making
an apologetic statement.
Closing: using some sort of summary
or closing (e.g., later).
Emoticon: using emoticons (e.g., ;-)).
Humor: Using humor to lighten the
mood.
Invitations: Statements that invite a
response.
Rhetorical questions: Posing a rhetorical question.
Salutation: Using a greeting, especially
with the individuals name.
Signature: Using a signature or nickname.
Thanks: Expressing thanks.

Computer-Mediated Communication

He found that participants believed that supportive CMC strategies were important and utilized them within an online learning environment
in 25.9% of the interactions. In particular, he found
that the most commonly used supportive communication strategies utilized by his participants
included: (a) referential statements, (b) signatures,
(c) greetings, and (d) horizontal questions.
Within a community of practice or learning,
it is recommended that members become aware
of and begin to utilize these strategies. While it
might seem logical to use some of these strategies
in group interactions, messages are often received
that do not include any closing and, at times, not
even something as important as the senders name.
In utilizing these strategies the CMC message will
include more communication cues to help enrich
the medium with more information. As described
earlier, a lack of nonverbal cues can be a possible
issue for female members of a community, and this
simple strategy might better support a members
understanding of a particular message. Simply
utilizing emoticons to replace the nonverbal cues
(e.g., facial expressions) missing from CMC can
provide more information to the recipient of the
message. For example, utilizing the winking
emoticon can provide information to the recipient that the statement was not meant to be taken
seriously. Other more complex strategies, such as
stating acknowledgement and agreement, provide
the recipient with information that might replace
a head nod or other gestures, not communicated
in text-only CMC. Furthermore, apologies for
miscommunications or other miscues provide I
statements and self-disclosures that are associated
with higher group development communication
style (HCS). The use of horizontal questions,
inviting responses, and providing referential
statements of anothers message support the
notion of enhancing social interactions, which
encourages additional participant engagement
in group interaction. With the use of these more
supportive CMC strategies, members will engage
in richer CMC to better understand one another,

and, thus, augment community building. Again,

it is the membership of the community that
decides upon the communication norms of that
community. However, with more members of the
online community being aware of and utilizing
these communication strategies the membership
is more apt to adopt them, making them a routine
of their social entitys repertoire of vocabulary
and style.
Although the various CMC strategies adopted
and utilized by a community might improve
text-based communication, advances in Internet
technologies have brought about the development
of multimedia CMC tools also could improve
communication. The next section will deal with
some of types of current multimedia tools that
can better support the needs and predilection of
the community.

Multimedia CMC Tools

Although computer desktop video conferencing
and telephony have increased the media options
for CMC, they still do not provide all of the rich
aspects of face-to-face interaction. However, these
multimedia technologies do add additional communication cues beyond the text-based communication strategies mentioned above. Therefore, it
does seem prudent and is recommended to utilize
advanced online multimedia communication
technologies to the traditional tool set to thicken
the communication conduit when possible and,
perhaps more importantly, when appropriate.
By utilizing a video conferencing, for example,
communication can be enhanced with facial
expressions, hand gestures, voice inflection and
other cues missing from text-only CMC. In addition, many of these tools provide interactive
whiteboards for shared graphic demonstrations.
These tools can be used to provide more graphic
elements to enhance ones ideas or concepts.
Furthermore, some of these tools also permit
document or application sharing. This can be a

1559

Computer-Mediated Communication

valuable tool that can provide technical support

and foster online collaboration.
Although these tools can provide the added
communication support, they also can complicate
the communication process. In terms of video
conferencing, for example, some individuals may
dislike being on display and this use may cause
some discomfort. Also, utilizing the audio and
video components, one must have a camera and
microphone for each individual, which requires
the users, to some extent, to be more technically
proficient to set up and use these multimedia
peripherals. In addition, high speed Internet
connectivity for all parties interacting within the
group communication is really a must. Although
it is possible to utilize these tools with low bandwidth, it can become very stilted and, thus, very
frustrating, perhaps negating the advantages.
Furthermore, some of these multimedia CMC
tools are open systems allowing one to communicate with anyone, while others are more closed
to specific group membership. Because the various types of systems, it is important to know the
strengths and weaknesses, and to understand their
relative costs and benefits, as they have specific
requirements and individual featureskeeping
in mind the purpose or objective of the groups
communication. The multimedia CMC tools
listed below were chosen to provide examples of
the different types currently available and are not
inclusive of all tools.

Skype
Skype (n.d.) is an Internet telephony service that
utilizes a computer with Internet connectivity, primarily for audio, provides computer to computer
Internet Telephony. Although it does provide a
video option, it currently is in the beta version.
However, once connected users also have a chat
window available for synchronous messaging as
well. Users can set up conference calls with up
to four participants plus the host.

1560

Skype client software includes all of the features one might expect in such a tool, such as an
address book of contacts for organizational and
ease of use purposes. Skype accounts also provide
several services for various fees, such as SkypeIn,
which provides the user a phone number much like
a tradition phone including voicemail, and SkypeOut, which permits users to make phone calls to
both land and cell phone lines, including SMS, a
form of instant messaging. Requirements include
a headset with microphone, which is used instead
of a microphone and speakers to keep feedback
from occurring. The easy-to-use client software
can be downloaded for free. More importantly,
is that computer-to-computer communication is
free. There is a fee to utilize calls to and from
land or cell phones; users need a pre-paid account. However, the costs are quite inexpensive,
especially for international calls. Because this
form of Internet Telephony primarily focuses on
audio, Skype might be a good tool to consider
when beginning utilizing multimedia CMC, as it
only requires a headset and the easy-to-use free
software to get started.

iVisit
iVisit (n.d.) combines a variety of CMC tools
including, video conferencing, instant messaging, file sharing, and desktop sharing. Although
server licenses are available, most users have to
obtain access to the software application in the
traditional server/client relationship, where iVisit
provides the server access, and users download
client software and sign up for one of two types of
accounts, Light and Plus. The software includes
several windows, including; the video feed with
controls, an address book that lists both people
and places (available conferences), and a chat
window for both the individual and all chatting
within a conference.
An iVisit Light is a free account that permits
users to interact as guests in open or public con-

Computer-Mediated Communication

ference rooms or invited into private conference

rooms. iVisit Plus accounts are yearly subscriptions that permit the ability to host a conference
room that can accommodate eight visitors, including the host. Although this can be increased to
16 simultaneous users for no extra fee, it is not
recommended due to the complexity of managing
that many in an online video conference. In addition to being able to host a conference, the iVisit
Plus account provides greater audio quality and
video resolution, plus other benefits. One rather
interesting feature of iVisit is the ability to make
recordings of participants audio and video feeds.
Although the Light account only permits saving
the file into an iVisit proprietary file format, the
Plus account provides for QuickTime recordings.
While using iVisit for multimedia, CMC provides
greater cues that video provides, it also is more
complex to setup, requiring a camera, microphone,
and/or headset. Depending upon the technical
sophistication and needs of the particular community, this tool may be worth the complexity.

Microsoft NetMeeting
and Messenger
For quite a few versions, Microsoft Windowsbased operating systems have provided multimedia CMC tools that include various features
depending upon the particular version. The
tools range from earlier versions of Microsoft
NetMeeting (n.d.) to, most recently, Microsoft
Live Messenger (n.d.). While Live Messenger is
quite similar to Yahoo Messenger and focuses
on instant messaging (IM), audio-video Internet
telephony (somewhat like Skype), and file sharing, NetMeeting also includes application sharing. NetMeeting, will no longer be part of the
new Windows Vista operating system. Instead,
Microsoft will include a new suite of CMC tools
called, Windows Meeting Space for home and
home office users, Office Live Meeting Windows
Meeting Space for small and medium business
users, and Office Live Meeting Office Com-

munications Server 2007 for large enterprises

(Microsoft NetMeeting, n.d.). Each level of the
application will work similarly to NetMeeting in
that it will still support desktop, application, and
file sharing across networks. One of the biggest
changes from earlier versions of NetMeeting,
however, will be the increase in the number of
users that can connect simultaneously. One of the
biggest advantages of the Microsoft multimedia
CMC tool set is that it is readily available within
ones operating system. However, with the newer
versions, outlined on the Microsoft Web site,
members within online communities will need
to decide which tool best suits their needs and
available resources.

Elluminate
Elluminate (n.d.) is a comprehensive online
multimedia-conferencing tool that includes audio,
chat, interactive whiteboard, application sharing,
file transfer, and direct messaging. It truly goes
beyond the tools previously listed, as its primary
focus is to support group interactions. As such,
Elluminate includes additional features designed
to provide greater user interactivity, such as: (a)
participant profiles to provide greater information
(photo & bio) about the participants, (b) polling,
which permits a moderator to get some feedback
regarding an issue, question, or comment, and
(c) breakout rooms, a great feature should users
want to break up a large conference for smaller
group discussions. One important feature is that
it utilizes full duplex audio, meaning that more
than one individual can speak at the same time.
However, the tool also provides for moderator
control, should that be desired. In addition, the
tool is designed for assistive access with closedcaption transcript and key stroke configurability.
Elluminate also provides for input from a Web
cam, which provides video feed of the speaker,
although this feature is not the primary focus as
with iVisit.

1561

Computer-Mediated Communication

Elluminate provides multimedia CMC on a

much more enterprise scale, and, as such, the
cost is commensurate. There are several versions
available, each appropriately priced, including
Lite, Academic, and Enterprise editions, depending upon the users size and needs. Currently,
Elluminate is being utilized by corporate trainers
and universities. In addition, Elluminate has partnered with WebCT Vista, eCollege, Blackboard,
and others to devise power links that permit Elluminate to work seamlessly within these course
management systems. This multimedia CMC tool
is designed for communities that want to focus
on the synchronous interactive components that
this tool provides. The software also provides
for asynchronous delivery by including a session
recording feature, which makes sessions available
for later viewing. Although Web-delivered video
via a Web cam is available, Elluminate does not
focus on multiple, simultaneous video feeds such
as iVisit. Depending upon need, this tool might
provide users with a very interactive multimedia
CMC tool, but as with any complex tool, deployment and implementation are a consideration.
Again with this tool, users will be required to
have a microphone, speakers, or headset to interact
with the multimedia elements.

Multimedia CMC Uses for Support

and Collaboration
The level of multimedia CMC tools available
varies from those that are very simple to those
that are much more comprehensive and quite
very complex. As mentioned, it should be the
membership of the community that selects the
CMC tools. In addition to increasing the number
of communication cues engaged users receive
due to the addition of audio and video, there are
further ways to utilize the multimedia CMC tools.
In particular, the application-sharing feature of
some multimedia CMC tools better support col-

1562

laborations that enhance community building and

augment the shared construction of knowledge.
For example, one of the primary uses for Microsoft
Application Sharing is to provide remote technical
assistance. When individuals are having difficulties with a computer application they can be supported by sharing that application with a distant
help agent. The user can watch and listen as the
help agent provides instruction and demonstrates
the application elements that caused confusion.
In this way the utilization of a multimedia CMC
tool for remote assistance can be used to help
members by any more knowledgeable member
within the community.
A more collaborative example of utilizing
multimedia CMC tool to support a distributed
learning environment is using the application
sharing for group authorship. In this case, team
members can schedule synchronous online meetings where they utilize application sharing to show
and discuss Web resources. They also can use
application sharing to work collaboratively on the
same file to produce a shared Word document or
presentation. In doing so, they are collaboratively
engaging in the production of an artifact, thus,
co-constructing knowledge.

Summary
This section provided recommendations that could
enhance CMC group interactions. Specifically, it
is recommended that the users of CMC consider
fostering the building online communities of
practice where members co-construct, hopefully,
equitable communication norms in their online
group interactions. It also is recommended that the
members engaged in group interactions become
aware of and utilize CMC strategies that enhance
their interactions by fostering more clear communications. Finally, it is recommended that multimedia CMC tools be considered to enhance and
enrich CMC when practical and appropriate.

Computer-Mediated Communication

Future Trends
This chapter has discussed the issues and recommendations of current CMC group interactions.
Over the past few years, incredible advances
in communication technologies have become
available and have been embraced by our society. So what does the future hold? In reviewing
the literature of those whove studied aspects,
issues, and elements of CMC group interaction,
one observation is evidentelectronic computermediated communication has expanded in use,
sophistication, and reach. More people now use
various types of computer-mediated communication than ever before to communicate with others
in all parts of the world. However, the literature
provides evidence that online group interactions
can suffer the same issues as traditional face-toface group interactions. With that in mind, one
could predict that future trends in CMC group
interaction will include the continued expanding
use, advancement, and sophistication of CMC
tools to an even more global scale than currently
exists. With the expanding global reach, users
skills, in terms of more sophisticated tools and
more strategic use, will need to increase as well.
With more diverse users making use of the various
CMC tools, communicators will need to become
better accomplished CMC users to compensate
for the limitations of the various CMC systems.
In addition, they will need to be more knowledgeable about possible group interaction issues that
pertain to newer systems. If that prediction holds
true, future global CMC users would benefit from
greater investigation of cultural communication
differences and global CMC interactions as they
relate to the various media that transmits their
messages.

Expanding and Changing CMC Skills

Computer-mediated communication has become
an important aspect of our society. However, as
detailed in this chapter, there are issues that are

embedded within CMC group interactions that

may impact ones ability to clearly communicate
via the various asynchronous and synchronous
systems. It would seem that being aware of the
strengths and limitations of the CMC tools and
having skill using those strategies, would provide
an advantage. For example, having knowledge of,
monitoring, and then intentionally utilizing some
of the communication strategies listed above may
make one more apt to better communicate within
a lean or limited CMC system. By utilizing some
of the CMC text-based strategies listed above,
regardless of gender, one could promote his or her
higher group communication style (HCS), thus
enhancing the message with richer, more engaging
levels of communication. In other words, those
individuals who are more CMC savvy, in terms of
being able to more purposefully utilize communication strategies would be better able to foster and
support persuasive arguments, promote agenda,
and clearly share their ideas. Given knowledge
of the specific strengths and weaknesses of various communication styles within specific CMC
media would provide a powerful advantage to
those who might strategically adjust their own
communication style accordingly.
Parallel events are beginning to happen within
the use of mobile phones for text messaging. For
example, users are utilizing text messaging with
their mobile phones or instant messaging on their
computers for quicker, asynchronous communication when they cannot talk aloud or in lieu of e-mail
or other slower CMC systems. When thinking
about future possible trends, it seems logical to
focus future practical efforts toward educating
members of communities engaged in CMC to
better understand the limitations of various CMC
systems and how best to understand, monitor, and
promote clear, concise, and perhaps strategic CMC
style. This may become much more important as
higher-speed Internet bandwidth becomes more
widely available making multimedia CMC tools
more common place.

1563

Computer-Mediated Communication

Expanding Need for Global

Understandings
This chapter has focused upon detailing specific
issues of CMC style differences and media issues,
and has provided some recommendations on building communities of practice within these CMC
systems. For the most part, the discussion has
focused on the research literature that discusses
CMC style issues as they relate to group interaction differences and the various media available that might impact the communication cues
available. As education, business, and industry
extend their reach to more global markets, CMC
will be utilized more extensively for interaction
by members of various cultures and languages.
While certain elements of CMC may better support
these endeavors, there will be other issues that
could compound good CMC group interaction.
For example, asynchronous CMC better supports
members in various time zones allowing for business to operate at their normally desired time.
With more synchronous multimedia utilization
globally, users will be required to interact 24/7.
In addition, more global group interactions will
require users to be sensitive to diversity in both
language and culture. For example, an international student from China was concerned when
his advisor sent him an e-mail where some of
the text was colored red. He thought she was
angry with himwhen in truth, she simply was
trying to emphasize a point by making it stand
out. From her perspective, this was a very benign
strategy. Because the color red has specific connotations within the Chinese culture, there was a
pretty serious cultural miscommunication for this
Chinese student. While no harm was intended by
this communication strategy, its use created an
unintended consequence of causing great alarm
for the student.
Future research in CMC utilization will continue to include the issues of gender and communication style. However, these issues are bound
to be expanded by cultural differences as CMC

1564

tools are used by members across a more global

society trying to build communities of practice
with various cultural differences. A very powerful
example of building community within a multiethnic online discussion group is described in the
work of Kadende-Kaiser (2000). She examined
CMC interactions of members of Burundinet, a
multiethnic discussion forum for Burundians in
the Diaspora. Most of the discussion focused on
the terrible ethnic violence between the Hutus and
Tutsis by members that spoke Kirundi (the native
language), French (the colonial language), and
English. In addition to their language differences,
members came from very different perspectives
on a very divisive topic. In this example, social
rules were changed to provide more equitable
social interactions. This would not be the case
in Burundian social settings, where seniority and
social status determine who can interact with
whom, or who is even excluded in the interaction. . . . In internet communication, however,
these cultural rules regarding turn-taking on the
basis of age and social rank are irrelevant; any
subscriber can play the role of social mediator
(p. 131). Indeed, the membership of Burundinet
fashioned the rules as she states:
Mutual respect and politeness, on the one hand,
and the freedom of expression, on the other, do not
have to be mutually exclusive. Therefore, many
Burundinet members seek to preserve individual
subscriber rights to express their opinion, regardless of what views they have in mind. As one net
member advised: It seems more wise that we try
to preserve a minimum of formal politeness toward
the others, and at the same time we need to allow
every Burundinetter to express his/her opinion on
Burndi, regardless of what this opinion is(Author
I, 1996). These statements and individual opinions
serve as the foundation for message composition
and they contribute to the establishment of the
rules for interaction on Burundinet. (p. 134)

Computer-Mediated Communication

Although cultural differences, such as this,

may cause issues when filtered through the lean
medium of current CMC technologies within a
more global community, the membership of Burundinet demonstrated how it established unique
rules for CMC interaction within their virtual
community to permit discussion regardless of their
differences. This provides a very good example of
how the CMC group interaction communication
norms are established by the membership.

Summary
This section provided some ideas of what the
future might hold for CMC group interactions.
Certainly, we can presume that CMC usage will
increase and become more complex and, perhaps,
robust. CMC users will be required to become
more technically proficient. A more important
concern, however, will be that as the usage of
CMC expands more globally, the communities
that will better prosper will be those that trend toward the development of communities of practice,
where members think about their responsibility
in the development of their shared repertoire of
communal resources in light of possible cultural
differences. It will be vital for members engaged
in CMC group interactions to be aware of, and
provide consideration for, cultural differences.
While the CMC strategies outlined above may
work well for certain cultures, will they work
well within a global community? Or, will they
have unintended consequences?

Future Research Directions

There is a history of research that has investigated
face-to-face group interaction that supersedes
the CMC tools that are now employed. The findings from that earlier research seems to parallel
research findings of those engaged in CMC group
interactions. For example, gender-based CMC
style differences can impact those engaged in

online group interactions. However, there are

strategies that users can and do employ to adjust
their CMC group interactions in an attempt to
replace missing nonverbal cues, thus supporting
and fostering better communication. Therefore,
future research that further investigates how
users manipulate their CMC style strategically
by employing CMC techniques and/or adjusting
their CMC group interactions, based upon the
medium, could prove very enlightening. Because
members engaging in CMC group interactions
can and do take responsibility for developing
their community of practice, researchers also
would be advised to study the process of defining
and developing communication norms within an
online community of practice.
As technologies continue to advance, richer
media will become commonplace for CMC group
interactions. Regardless of new, more advanced
multimedia electronic communication systems,
there still will be a filtering of sorts on communication cues within CMC systems. Future researchers
would be advised to focus their investigations on
how multimedia might hinder or enhance group
interactions based upon gender, and perhaps more
importantly, culture. In particular, it seems essential to investigate global CMC group interactions
in light of cultural and language differences. For
example, community systems like BuddySpace,
that provide the users location could provide cues
to support those engaging in group interactions
within that type of online community space.
Researchers should be encouraged to take advantage of the known CMC issues and focus their
investigations on more global issues that are bound
to emerge as communities expand. While gaining
knowledge of how more robust, multimedia CMC
might impact various individuals it vital, future
research should go beyond and also focus on how
to best educate members of global communities
of practice to provide clear, concise communication within those more advanced electronic communication systems. With continued knowledge
of the advanced CMC systems and the diverse

1565

Computer-Mediated Communication

cultural interactions that happen with them, this

form of information technology will better serve
the expanding global community.
Finally, instant messaging (IM) and text messaging has become widespread, globally. This
form of asynchronous communication is novel in
that it changes the type and, perhaps, the reason
for a particular communiqu. For example, one
might text or IM another because they are
busyworking, in class, or talking with othersand cannot be interrupted with a traditional
phone call or chat. Furthermore, this form of
communication has developed its own lexicon and
syntax. While people may have wanted to communicate in similar ways in the past, this technology has changed the way many communicate, as
evidenced by its overwhelming use. Specifically,
this technology has provided the opportunity for
people to multitask in many areas of life, which
brings up an interesting question. How will this
new form of communication impact or change
more traditional forms of communication? With
that in mind, an interesting opportunity for future
research would be to investigate how technology advances have impacted or changed more
historical forms of human communication and
social interactions.

References
BuddySpace (n.d.). Retrieved March 15, 2007,
from http://www.buddyspace.org
Carli, L. L. (1984). Sex differences in task behaviors, social behaviors and influences as a function
of sex composition of dyads and instructions to
compete or cooperate. Dissertation Abstracts
International, 45(1-B), 401.
Daft, R. L., & Lengel, R. H. (1986). Organizational
information requirements, media richness and
structural design. Management Science, 32(5),
554-571.

1566

Dede, C. (2004). Enabling distributed learning

communities via emerging technologies: Part I.
THE Journal, 32(2), 12-22.
Dennis, A. R., Kinney, S. T., & Hung, Y. C. (1999).
Gender differences in the effects of media richness. Small Group Research, 30(4), 405-437.
Eakins, B. W., & Eakins, R. G. (1978). Sex differences in human communication. Boston:
Houghton Mifflin.
Elluminate (n.d.) Retrieved September 22, 2006,
from http://elluminate.com
Fahy, P. J., Crawford, G., & Ally, M. (2001). Patterns of interaction in a computer conference
transcript. International Review of Research in
Open and Distance Learning, 2(1).
Fahy, P. J. (2003). Indicators of support in online
interaction. International Review of Research in
Open and Distance Learning, 4(1).
Fishman, P. M. (1983). Interaction: The work
women do. In B. Thorne, C. Kramarae, & N.
Henly (Eds.), Language, gender and society.
Rowley, MA: Newbury House.
Grinter, R. E., & Eldridge, M. (2003). Wan2tlk?:
Everyday text messaging. In Proceedings of ACM
Conference on Human Factors in Computing
System (CHI 2003) (pp. 441-448).
Herring, S. (1993). Gender and democracy in
computer-mediated communication. Electronic
Journal of Communication/La Revue Electronique de Communication [On-line serial], 3(2).
Reprinted in R. Kling (Ed.). (1996). Computerization and controversy (2nd ed.). New York:
Academic Press.
Herring, S. (1999). The rhetorical dynamics of
gender harassment on-line. The Information
Society, 15, 151-167.
Hutchins, E. (2000). The cognitive consequences
of patterns of information flow. Intellectica, 1(30),
53-74.

Computer-Mediated Communication

iVisit (n.d.). Retrieved September 22, 2006, from

http://ivisit.com
Jaffe, J. M.,Lee, Y., Huang, L., & Oshagan, H.
(1999). Gender identification, interdependence,
and pseudonyms in CMC: Language patterns in
an electronic conference. Information Society,
15(4), 221-234. (EJ602017)
Kadende-Kaiser, R. M. (2000). Interpreting
language and cultural discourse: Internet communication among Burundians in the Diaspora.
Africa Today, 47(2), 121-148.
Kim, A. J. (2000). Community building on the
web: Secret strategies for successful online communities. Berkley, CA: Peachpit Press.
Maguire, T. (1999). Gender cues impact marketers.
American Demographics, 21(1), 16.
Microsoft Live Messenger (n.d.). Retrieved
October 25, 2006, from http://get.live.com/messenger
Microsoft NetMeeting (n.d.). Retrieved October
25, 2006, from http://www.microsoft.com/windows/NetMeeting
Orkut (n.d.). Retrieved March 15, 2007, from
http://www.orkut.com
Palloff & Pratt (1999). Building learning communities in cyberspace: Effective strategies for the
online classroom. San Francisco: Jossey-Bass.
MySpace (n.d.). Retrieved March 15, 2007, from
http://www.myspace.com/
Rheingold, H. (1993). A slice of my virtual community. In L. M. Harasim (Ed.), Global networks.
Cambridge, MA: MIT Press.
Savicki, V., Kelley, M., & Oesterreich, E. (1999).
Judgments of gender in computer-mediated communication. Computers in Human Behavior, 1(2),
185-194.

Savicki, V., & Kelley, M. (2000). Computer

mediated communication, gender and group
composition. CyberPsychology and Behavior,
3, 817-826.
Savicki, V., Kelley, M., & Ammon, B. (2002).
Effects of training on computer-mediated communication in single or mixed gender small task
groups. Computers in Human Behavior, 18,
257-259.
Skype (n.d.). Retrieved September 22, 2006, from
http://www.skype.com
Soukup, C. (1999). The gendered interactional
patterns of computer-mediated chatrooms: A
critical ethnographic study. The Information
Society, 15, 169-176.
Wenger, E. (1998). Communities of practice:
Learning, meaning and identity. New York:
Cambridge University Press.

Additional Reading
Barnes, S. B. (2001). Online connections: Internet interpersonal relationships. Creskill, NJ:
Hampton Press.
Barrett, M., & Davidson, M. J. (2006). Gender
and communication at work: Gender and organizational theory series. Burlington, VT: Ashgate
Publishing.
Collison, G., Elbaum, B., Haavind, S., & Tinker,
R. (2000). Facilitating online learning: Effective
strategies for moderators. Madison, WI: Atwood
Publishing.
Day, P. (Ed.). (2004). Community practice in the
network society: Local action/global interaction.
New York: Routledge.
Ermann, M. D., & Shauf, M. S. (2003). Computers,
ethics, and society. New York: Oxford Press.

1567

Computer-Mediated Communication

Green, E. (2001). Virtual gender: Technology

consumption and identity matters. New York:
Routledge.
Jonassen, D. H., & Harris, P. (Eds.). (2004).
Handbook of research on educational communications and technology (A project of the Association for Educational Communications and
Technology). Mahway, NJ: Lawrence Erlbaum
Associates, Inc.
Katz, J. E., & Rice, R. E. (2002). Social consequences of Internet use: Access, involvement, and
interaction. Cambridge, MA: The MIT Press.
Kelly, T. B., Berman-Rossi, T., & Palombo,
S. (Eds.). (2001). Group work: Strategies for
strengthening resiliency. Binghamton, NY: The
Haworth Press, Inc.
Kisielnicki, J. (Ed.). (2002). Modern organizations in virtual communities. Hershey, PA: IRM
Press.
Krolokke, C., & Sorensen, C. A. (2006). Gender communication theories & analyses: from
silence to performance. Thousand Oaks, CA:
Sage Publications.
Lau, R. W. H., Li, Q., Cheung, R., & Liu, W. (Eds.).
(2005, July 31-August 3). Advances in Web-based
learning ICWL 2005. 4th International Conference, Hong Kong, China. Heidelberg, Germany:
Springer-Verlag.
LeVine, P., & Scollon, R. (Eds.).
(2004). Discourse and technology: Multimodal discourse
analysis. Washington, DC: Georgetown University Press.
Moggridge, B. (2007). Designing interactions.
Cambridge, MA: The MIT Press.

Saint-Onge, H., & Wallace, D. (2003). Leveraging

communities of practice for strategic advantage.
Burlington, MA: Butterworth-Heinemann (an
imprint of Elsevier Science).
Shane, P. M. (2004). Democracy online: The prospects for political renewal through the Internet.
New York: Routledge.
Stangor, C. (2004). Social groups in action and
interaction. New York: Psychology Press.
Sullivan, N.E., Mesbur, E.S., Lang, N.C., Goodman, D., & Mitchell, L. (Eds.). (2003). Social work
with groups: Social justice through personal,
community and societal change. Binghamton,
NY: The Haworth Press, Inc.
Talbot, M. M. (1998). Language and gender: An
introduction. Malden, MA: Blackwell Publishers Inc.
Tu, C. (2004). Online collaborative learning
communities: Twenty-one designs to building
an online collaborative learning community.
Westport, CT: Libraries Unlimited (an imprint
of Greenwood Publishing Group, Inc.).
Tubbs, S. (2004). A systems approach to small
group interaction (NAU). New York: McGraw
Hill.
Vestal, W. (2006). Communities of practice and
associations. Houston, TX: APQC.
Wenger, E., McDermott, R., & Snyder, W. M.
(2002). Cultivating communities of practice.
Boston: Harvard Business School Publishing.
Wiberg, M. (Ed.). (2005). The interaction society:
Theories, practice and supportive technologies.
Hershey, PA: Information Science Publishing.

Palloff, R. M., & Pratt, K. (2005). Collaborating

online: Learning together in community. San
Francisco: Jossey-Bass.
This work was previously published in Handbook of Research on Global Information Technology Management in the Digital
Economy, edited by M. Raisinghani, pp. 195-214 , copyright 2008 by Information Science Reference, formerly known as Idea
Group Reference (an imprint of IGI Global).
1568

1569

Chapter 4.9

Mobile Phone Use

Across Cultures:

A Comparison Between the

United Kingdom and Sudan
Ishraga Khattab
Brunel University, UK
Steve Love
Brunel University, UK

Abstract
Recently, the ubiquitous use of mobile phones by
people from different cultures has grown enormously. For example, mobile phones are used to
perform both private and business conversations.
In many cases, mobile phone conversations take
place in public places. In this article, we attempt
to understand if cultural differences influence
the way people use their mobile phones in public
places. The material considered here draws on the
existing literature of mobile phones, and quantitative and qualitative work carried out in the UK
(as a mature mobile phone market) and the Sudan
(that is part of Africa and the Middle East culture
with its emerging mobile phone market). Results
indicate that people in the Sudan are less likely
to use their mobile phones on public transport or

whilst walking down the street, in comparison to

their UK counterparts. In addition, the Sudanese
are more willing to switch off their mobile phones
in places of worship, classes, and meetings. Implications are drawn from the study for the design
of mobile phones for different cultures.

Introduction
Economic globalization and the widespread use
of mobile phones have changed the way people
live and manage their lives, and cut down the
virtual distance between countries, regions, and
time zones. New ways of using mobile phones are
constantly emerging (e.g., downloading music to
listen to on the train), and the pervasive use of
mobile phones in public places for private talk

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Mobile Phone Use Across Cultures

(both business- and socially-oriented) is a clear

example of how mobile phones are changing our
economic and social lives. As a result of this,
there is an emergent body of research on the use
of mobile phones in social spaces. For example,
Ling (2004) highlights how their use in public
places has raised questions of what the appropriate
or inappropriate behaviour is in public places. In
this study, he found that people perceived mobile
phone use in places such as restaurants as unacceptable, partly because mobile phone users tend
to talk louder than usual so that people nearby feel
intruded upon, embarrassed, and have a sense of
being coerced into the role of eavesdropper on a
private conversation.
Research has also shown that mobile phones
can occupy concurrent social spaces, spaces with
behavioural norms that sometimes conflict, such
as the space of the mobile phone user, and the
virtual space where the conversation takes place
(Palen, Salzman, & Youngs, 2000). This feeling
of conflict has led researchers in this area to propose that the use of mobile technology in public
places is creating a new mixture of public and
private space that has yet to be accommodated
by for users of mobile technology and bystanders
in terms of what is acceptable or unacceptable
behaviour.
This phenomenon has been analysed predominately using concepts drawn from Goffmans
analysis of social interaction in public places
(Goffman, 1963). In this work, Goffman suggested that people have specific public faces
and personas for different public social locations.
The idea behind this is that individuals have rules
that determine their behaviour in public places,
or what Burns (1992) refers to as the observance
of social propriety. For example, Murtagh (2001)
presented findings from an observational study
of the nonverbal aspects of mobile phone use in a
train carriage. Murtagh found that changing the
direction of ones gazeturning ones head and
upper body away from the other people sitting next

1570

to you in the carriagewas a common feature of

mobile phone behaviour on trains. These behavioural responses were seen as being indicative
of the subtle complexities involved when using
mobile phones in public locations. This study
suggests that mobile phone users are actively
engaged in trying to distance themselves from
their current physical location in order to enter
a virtual environment with the person they are
having a mobile phone conversation. In relation to
this, Love and Perry (2004) used role-play experiments to investigate the behaviour and attitudes
of bystanders to a mobile phone conversation.
They found that participants had strong views on
embarrassment, discomfort, and rudeness. They
also report that the actions of those who were put
in the position of overhearers followed a pattern:
they acted as though they were demonstrably not
attending, even though they were all able to report
accurately on the content of the conversation.
However, to date, most of the research reported
in this area has tended to focus on what is termed
the developed world. Mobile phones are also transforming peoples lives in the developing world.
In Africa, the unreliable and inefficient landline
telecommunication infrastructure has made the
mobile phone the solitary available communication tool for many people (BBC, 2003). However,
as mobile phone use in Africa continues to grow,
there is a need for mobile phone companies who
are entering this market to consider the possible
impact of cross-cultural differences in peoples
attitude towards mobile phone and service applications.
This article first briefly reviews relevant literature about the use of mobile phones in public
places. The concept of culture and cultural models
are explored in th second section. In the third
section, the methods of this study are presented.
Techniques of collecting the data and the procedure of this study are presented in the fourth and
fifth sections, respectively. Some key findings
from the study are presented and discussed in the

Mobile Phone Use Across Cultures

sixth and seventh sections with reference to how

cultural differences might affect mobile phone
use in public places. Finally, the conclusion of
this study is presented in the last section.

What is Culture?
Culture is a complicated paradigm that is difficult
to accurately define. According to some researchers, culture must be interpreted (van Peursson,
in Evers & Day, 1997). Hofstede (1980) conceptualized culture as programming of the mind,
suggesting that certain reactions were more likely
in certain cultures than in others, based on differences between the basic values of the members
of different cultures (Smith, Dunckley, French,
Minocha, & Chang, 2004). Culture can also be
seen as a collection of attributes people acquire
from their childhood training. These attributes
are associated with their environment, surroundings that influence the responses of people in that
culture to new ideas, and practices and use of new
technology (such as mobile phones). Given that
culture may affect the way people behave and
interact in general, Ciborowski (1979) identified a
close link between knowledge and culture. In the
context of mobile phone communication, it may
be argued that culture influences knowledgeor
the individuals general experiencetherefore
affecting, in this instance, their attitude towards
mobile phone use in public places.
Another explanation of culture has been offered by Hofstede (1980). He produced a cultural
model that focuses on determining the patterns of
thinking, feeling, and acting that form a cultures
mental programming. This model has been
adopted for the study reported in this article,
as researchers in the area of cross-cultural differences and technology use consider it a valid
and useful measure of systematic categorization
(e.g., De Mooij, 2003; Honald, 1999). In addition, it is also considered to be directly related

to the relationship between product design and

user behaviour (De Mooij & Hofstede, 2002). An
explanation of Hofstedes cultural dimensions is
as follows:

Power distance: the extent to which less

powerful members expect and agree to unequal power distribution within a culture.
The two aspects of this dimension are high
and low power distance.
Uncertainty avoidance: discusses the
way people cope with uncertainty and risk.
The two faces of this dimension are high
uncertainty avoidance and low uncertainty
avoidance.
Masculinity vs. femininity: refers to gender
roles, in contrast to physical characteristics,
and is usually regarded by the levels of assertiveness or tenderness in the user. The two
aspects of to this dimension are masculinity
and femininity.
Individualism vs. collectivism: deals with
the role of the individual and the group, and
is defined by the level of ties between an
individual in a society. The two aspects of
this dimension are individualism and collectivism.
Time orientation: deals with the extent to
which people relate to the past, present, and
future. The two aspects of this dimension
are short-term orientation and long-term
orientation.

A number of cross-cultural studies have investigated differences in attitudes towards new

technology. Smith, French, Chang, and McNeill
(2001) carried out a study using Hofstedes model.
They adapted the Taguchi methoda partial
factorial experimental design methodin order
to investigate differences between British and
Chinese users satisfaction and preferences for
Web sites. They found significant differences
between British and Chinese users in their prefer-

1571

Mobile Phone Use Across Cultures

ence of detailed e-finance product information.

For example, Chinese users tended to adopt a
more holistic approach to viewing Web content,
as compared to British users.
In another study, Honald (1999) found that
German mobile phone users preferred clearlywritten and inclusively rich user manuals, whereas
Chinese mobile phone users focused on the quality
of the pictorial information.
Evers and Day (1997) found that there are clear
cultural differences between user acceptance of
interfaces for different cultural groups. In their
study, they found differences between Chinese
and Indonesian users. Indonesians were found to
like soft colours, black and white displays, and
pop-up menus more than Chinese users. Also,
Indonesians seemed to prefer alternative input
and output modes (e.g., sounds, touch screens,
data gloves, and multimedia) in comparison to the
Chinese who preferred the use of many different
colours for the interface design.
Despite the importance and the relevance of
cultural factors and its impact on the use of global
products and services (such as mobile phones),
little research has compared the effect of cultural
differences on issues such as social usability of
mobile phone use in the developing and the developed world. Sun (2003) argues that variation
in cultural states will cause different attitudes or
ways of using mobile phones.
The practice of the missed call is a clear
example of how users from different cultures
develop their own usage style. The missed call
is when the caller places a mobile phone call and
purposely hangs up before the recipient can answer
the call. Donner (2005) investigated the phenomenon in Sub-Saharan Africa where the missed call
is known as Beeping. He found that users have
produced elaborated codes and social messages to
be exchanged over the network without bearing
any costor at least not from those who are in a
less secure financial situation.
Another exclusive mobile phone cultural
attitude is evident in Bangladesh, Uganda, and

1572

Rwanda, where a woman, for example, borrows

money to buy a special mobile phone designed
for multiple user accounts and rural access.
After this, she then buys minutes in bulk and
resells them to customers in her village. This
programme is funded by the Grameen Bank
mainly for use in rural areas (Bayes, von Braun,
& Akhter, 1999).
The same idea has mutated slightly in the big
cities of Egypt and the Sudan where vendors who
own fixed contract mobile phones buy a bulk of
talk minutes and resell them in smaller chunks to
individuals in order to make a profit. This practice
is accommodated by their national phone service
providers and is known as balance transfer.
Obviously, this practice cannot be seen in London
or any of the developed world cities.
If mobile phone users have different usage
patterns, the question that the study in this article
addresses is: can we assume that people from different countries use mobile phones in the same
way? Thus the question arises: are there any roles
for cultural differences in the way people use their
mobile phones in public places? Therefore, the attitude of the British (a mature mobile phone user
market) and the Sudanese (an emerging mobile
phone user market) were examined in relation to
their attitudes towards the use of mobile phones
in public places.

Methodology
Participants
88 participants took part in the study: 43 British
(22 male, 21 female) and 45 Sudanese (20 male,
25 female), ranging in age from 15 to 63 years old,
with the average age of 30 years. All participants
were mobile phone users. The range of mobile
phone use for the Sudanese participants was from
2-5 years, whereas the British participants had
used mobile phones for 4-12 years.

Mobile Phone Use Across Cultures

Data Collection
Data was collected in this study using a questionnaire and an interview. The development of
the questionnaire went through several stages.
First, the generation of the questionnaire was
collated by employing an exhaustive review of
the literature generally on mobile phones, human-computer interaction (HCI), and cultural
issues in mobile phone use. Second, an in-depth
session was conducted with participants from both
countries (the UK and the Sudan) to develop the
questionnaire. Initially, a total of nine Likert-type
questions were developed. The scale was then
tested for content validity, which can be defined
as the extent to which a test actually measures
what it is supposed to measure (Rust & Golombok,
1989). This was undertaken using what is known
as the judgemental approach, with three mobile
HCI experts.
As a result of this process, the questionnaire
was subsequently revised to consist of six Likerttype questions. The six Likert statements focused
on attitudes towards the use of mobile phones in
public places. An example of the Likert statement
used in this study is as follows:
Mobile phones should not be switched off during
meetings:

Strongly agree
Agree
Neutral
Disagree
Strongly disagree

The attitude scale had a combination of positive and negative statements in order to control for
any possible acquiescence effect from participants
when they were completing the attitude questionnaire. This is a phenomenon whereby participants in
a study may unwittingly try to respond positively to
every question in order to help the investigator with

their study. This type of questionnaire format is one

of the most common methods used to elicit attitudes
from users in HCI research (Love, 2005).
In addition to the questionnaire, a semistructured interview was carried out. The interview
questions included open-ended and closed questions, and were designed to gather information
on the use of mobile phones in public places, the
practice of the missed call, and other features such as
the use of mobile phone caller ID. The main points
that were covered in the interview were:
1.
2.

Attitude towards the use of mobile phones

in public places.
The use of the missed calls types in the two
cultures. For example, the type of missed calls
used and the social messages sent through the
missed call, and how recipients differentiate
between these types of missed calls.

Examples of questions covered in the interview

were:
How do you feel about using mobile phones on
public transport?
How do you feel about using mobile phones in
school during classes?
How do you feel about using mobile phones in
restaurants?

Procedure
Participants were chosen from an opportunistic
sample in both the UK and Sudan and asked to
complete the questionnaire and return them to the
researcher once they had completed them.
The questionnaires took approximately 15
minutes to complete. At this point, an arrangement
was made to interview a subset of the participants
who had been selected randomly and volunteered
to answer the interview questions. Participants
were informed from the outset that the results of

1573

Mobile Phone Use Across Cultures

the study would be anonymous, and they would

be able to obtain the results of the study from the
researcher on request.

Results
An independent sample T test was carried out to
compare attitudes towards using mobile phones in

public places in the UK and the Sudan. There was

a significant difference found in the attitudes for
using mobile phones in public transport between
the British and the Sudanese (t=5.99, p<0.001).
The British were more willing to use it on public
transport than the Sudanese.
Another significant difference was noted
between the two countries towards using mobile
phones whilst walking on the street. Again, the

Table 1. Attitudes towards the use of mobile phones in public places in the UK and the Sudan

I would be comfortable
using my mobile phone
in restaurants
I would not be comfortable using my mobile
phone on public transport

I would be comfortable
using my mobile phone
whilst walking down the
street

Mean

Std.
Deviation

Std.
Error
Mean

df

P
Value
Sig 2
tailed

42

2.83

1.146

.177

1.325

70.241

.189

Sudan

45

2.56

.755

.113

British

42

3.29

1.175

.181

5.925

69.046

.000

***

Sudan

45

2.02

.753

.112

British

42

3.69

1.070

.165

3.884

82.171

.000

***

Sudan

45

2.84

.952

.142

42

4.45

.861

.133

3.094

51.314

.003

**

45

4.89

.318

.047

42

3.88

.968

.149

2.316

69.411

.023

45

4.29

.626

.093

42

4.00

1.307

.202

2.552

61.278

.013

45

4.58

.690

.103

COUNTRY

British

Mobile phones should be

switched off in places of British
worship
Sudan
Mobile phones should
not be switched off dur- British
ing meetings
Sudan
Mobile phones should be
switched off in schools
British
during classes
Sudan
*P<0.05; **P<0.01; ***P<0.001
1574

Mobile Phone Use Across Cultures

British were more favourable towards this than

the Sudanese (t=3.884, p<0.001). The Sudanese
were found to be more willing to switch off their
mobile phones in places of worships, meetings,

and in schools during classes. Please see Table 1

for a summary of the main results.
In terms of differences between the attitude of
the British and the Sudanese males, an unrelated

Table 2. Attitude difference between the Sudanese males in using mobile phones in public places and
the British males

Gender
Mobile phones should
Sudanese
be switched off in places
Male
of worship
British
Male
Mobile phones should
Sudanese
be switched off during
Male
meetings
British
Male
Mobile phones not to be
Sudanese
switched on in schools
Male
during classes
British
Male
I would be happy usSudanese
ing mobile phones in
Male
restaurants
British
Male
I would not be comfortable using a moSudanese
bile phone on public
Male
transport
British
Male
I would be comfortable
using a mobile phone
Sudanese
whilst walking on the
Male
street
British
Male

Std.
Deviation

Std.
Error
Mean

20 4.90

.308

.069

23 4.43

.992

.207

20 4.20

.523

.117

23 3.83

1.154

.241

20 4.50

.827

.185

23 4.17

1.403

.293

20 2.50

.688

.154

23 2.70

1.105

.230

20 2.25

.786

.176

23 3.13

1.180

.246

20 3.15

.813

.182

23 4.04

.825

.172

Mean

P
value
sig 2
tailed

df

2.134

26.761 .042

1.397

31.583 .172

.942

36.374 .352

-.706

37.389 .485

-2.912

38.570 .006

**

.869

40.330 .001

**

***

*P<0.05; **P<0.01; ***P<0.001

1575

Mobile Phone Use Across Cultures

T test revealed that the British males are more

willing to use mobile phones on public transport
and when walking on the street than the Sudanese
males (t=-2.912, t=.869, p<.001). Please see Table
2 for a full summary of the results.

Comparing the attitudes of the British and

the Sudanese females towards the use of mobile
phones in public placesan unrelated T test
revealed the British females are more relaxed
using mobile phones in public transport than the

Table 3. Attitude differences between females in the UK and the Sudan

Gender
Mobile phones should be
switched off in places of
worship

Mobile phones should be
switched off during meetings

Mobile phones should not

be switched in schools
during classes
I would be happy to use
my mobile phone in a
restaurant
I would not be comfortable using a mobile phone
on public transport
I would be comfortable
using a mobile phone
whilst walking on the
street

Sudanese
Female
British
Female
Sudanese
Female
British
Female
Sudanese
Female
British
Female
Sudanese
Female
British
Female
Sudanese
Female
British
Female
Sudanese
Female

British
Female
*P<0.05; **P<0.01; ***P<0.001

1576

Std.
Deviation

Std.
Error
Mean

25 2.60

.816

.163

19 3.00

1.202

.276

25 1.84

.688

.138

19 3.47

1.172

.269

25 2.60

1.000

.200

19 3.26

1.195

.274

25 4.88

.332

.066

19 4.47

.697

.160

25 4.36

.700

.140

19 3.95

.705

.162

25 4.64

.569

.114

19 3.79

1.182

.271

Mean

df

P
value
sig 2
tailed

-1.248

30.068

.222

-5.408

27.256

.000

-1.955

34.863

.059

2.348

24.196

.027

1.929

38.758

.061

2.892

24.322

.008

***

**

Mobile Phone Use Across Cultures

Sudanese females (t=2.348,p<.001). Please see

Table 3 for a full summary of the results.

Interview Results
The interview results corresponded with the
questionnaire data, indicating that there is a difference between the British and the Sudanese attitudes towards the use of mobile phones in public
places. Sudanese were found to be less willing
to use mobile phones in public places than their
British counterparts. In the interview, Sudanese
participants revealed various reasons for their
uncomfortable attitude towards the use of mobile
phones in public places. For example, some of the
participants felt that the use of mobile phones in
public transport is unacceptable because it can
be disturbing to other people in close proximity
to the mobile phone user. As one of the Sudanese
interviewees commented:
Using a mobile phone in public places, especially on public transport where you are closely
surrounded by people, is not something that you
can do comfortably. It is viewed as improper and
unacceptable, as it disturbs others.

situation; because most of the people surrounding

you will hear your conversation and this attitude
in itself is not acceptable in our community.
People are not expected to discuss private issues
so publicly.
On the other hand, British participants were
found to be more comfortable using mobile phones
in public places as one of the interviewees commented:
I have no problems using my mobile phone in
public places and especially on public transport,
as I can make use of time while sitting there doing nothing.
Another British interviewee added:
I use my mobile phone in public places all the
time and it does not bother me at all that people
are listening to my mobile phone conversations.
I do not know them and it is unlikely they are
going to know more details about the topic I am
discussing.

The use of mobile phones on public transport may

be considered as a sign of disrespect to others. In
particular, to older passengers who you have to
respect and act quietly around them.

The results of this study also indicated that

Sudanese females were less willing to use mobile
phones in public places than British females. Sudanese females felt that the use of mobile phones
in public places, especially on public transport,
could attract unwanted attention to them in a
society that expects females to keep a low profile.
This was echoed in one of the Sudanese female
interviewees comments:

An added justification that was revealed by

Sudanese participants for not feeling comfortable
using mobile phones in public places was related to
their tight rules in keeping private issues private,
as one of the interviewees commented:

I do not like using my mobile phone in public

places at all as it only magnetizes others attention towards me. If you are on the mobile phone
in a public place, people start gazing at you
unappreciatively.

The use of mobile phones in public places to discuss private matters can put you in an awkward

Another Sudanese female interviewee

added:

Another Sudanese interviewee added:

1577

Mobile Phone Use Across Cultures

Usually, I do not use my mobile phone in public

places, I prefer to keep a low profile. For me,
this attitude is a sign of respect for my self and
others.
British females appeared to have different
viewmost of the interviewees were found to feel
more comfortable using their mobile phones in
public places. As one of the British interviewees
commented:
I prefer to use my mobile phone in public places;
it keeps me busy and in a way safe, for example
when I want to get my car from the car park when
it is dark, I always make sure that I am talking
to one of my friends on the mobile phone just in
case something happens.

Discussion
The results from the study were interpreted in the
light of Hofstedes cultural dimensions to try and
gain some insight into the way culture may influence the use of mobile phones in public places.
It appears from the results that the British
generally are more comfortable using mobile
phones in public places than their Sudanese participants, who are more reluctant to use mobile
phones in contexts such as public transport and
whilst walking along the street.
The collectivistic culture to which the Sudan
belongs to (using Hofstedes criteria) indicates an
inclination toward a tightly-knit social framework
(Hofstede, 1980). The priority is for the groups
needs, rather than the individual wishes. Therefore, perhaps the use of mobile phones in public
places for private talks can be seen as a self-centred
act, and quite impertinent for the group needs. The
group expects the individual to be considerate to
the established social etiquette. The mobile phone
user in public transport is expected to adhere to
the social protocol and to respect other peoples
privacy.

1578

Another reason for the British comfortable

attitude to mobile phone use in public places may
be due to bystanders nonverbal communication
attitude. This concept is highlighted by Goffman
(1963) where he refers to it as civil inattention.
Civil inattention refers to the ways in which people
acknowledge the existence of others without paying them extra attention; he regarded this as a
gesture of respect required from strangers. Lasen
(2002a) found that civil inattention is clearly
present in UK culture: the British tend to avoid
open and straightforward looking at other people,
and keep away from paying direct attention to
others, especially on public transport, such as
the Underground. He suggested that this attitude
may encourage British mobile phone users to talk
more freely outdoors without being concerned
about others watching them.
In contrast, in the Sudan, it was noted that
civil inattention is not clearly evident. Sudanese
people tend to look at each other directly. Lasen
(2002a) suggested that a lack of proper gaze in
certain cultures where civil inattention does
not rule may be viewed as a lack of respect or
ignorance. This lack of civil inattention perhaps
justifies the reason behind the Sudanese unwillingness to use their mobile phones in public places,
as they are influenced by bystanders nonverbal
communication attitude. One can say the more
civil inattention paid to others, the more free and
relaxed they might feel towards using their mobile
phones, and vice versa.
Another justification for not using mobile
phones in public places might be due to the
high score that the Sudan attained on Hofstedes
uncertainly avoidance dimension. According to
Hofstede, cultures with high uncertainty avoidance scores tend to be expressivepeople talk
with their hands, raise their voices, and show
emotions. These characteristics can play a role in
decreasing the need to carry out private conversations in public places because people in these
cultures know that they tend to talk loudly and
expressively, which attract bystanders attention,

Mobile Phone Use Across Cultures

plus there is a high risk of being known to people

around you. Another important point is that as
Sudanese people in general talk loudly and in an
expressive way, this tends to increase the level of
external noise for mobile phone users. Therefore,
people talking on mobile phones need to raise their
voices more to win over competitive speakers.
This loud talking may attract bystanders attention and invite eavesdroppers, which can cause a
feeling of embarrassment on the part of the mobile
phone user. In addition, mobile phone users may
feel that bystanders might disrespect them if they
discuss their private matters publicly.
Additionally, the Sudanese attitude might be
related to the high score obtained on Hofstedes
power distance dimension, where a tight set of social rules are established, and people are expected
to follow and respect these rules. For example,
the social protocol for behaviour in public places
is well recognized in the Sudan, and people are
expected to behave in certain ways and not to
speak loudly in front of others (especially older
people). Private issues should be kept private and
dealt with in a private manner and in private settings. It is considered improper to breach these
norms. Although in the UK, a social protocol
for behaviour in public places also exists, the
maturity of the UK mobile phone market may
have relaxed or altered peoples expectations and
acceptance behaviour in public places. Palen et
al. (2000) found that a persons attitude towards
public mobile phone use changes (becomes more
accepting) as their mobile use increases. In addition, Palen (2002) predicted that as adoption
of mobile phones increases, people will be less
disturbed about proper use, but will still prefer
to have mobile free zones.
In terms of specific gender differences, Sudanese females were found to be more uncomfortable about using mobile phones in public places
in comparison to British females. This attitude
fits in with the feminine attribute of the Sudan
culture suggested by Hofstede (1980), where the
prevailing value is caring for others. The UK,

in contrast, is judged by Hofstede to be more

masculine-oriented, and the dominant values are
achievement and success.
Although the Sudanese females practice all
their rights in terms of education, work, leisure,
and the like, they are looked after and cared for
by the whole society. As a result of this caring
perception towards females in the Sudanese
culture, their attitudes and behaviours are more
controlled and guarded as they are expected to
follow social protocols more than men. For example, Sudanese females are expected to keep a
low profile and deflect attention from themselves
by talking quietlyand preferably avoid talkingin public spaces.
On the other hand, according to the results of
this study, British females are more comfortable
using mobile phones in public places. This may
be due to the feminine attribute of the UK suggested by Hofstede (1980) where women are seen
as equal to men, and they are expected to look
after and guard themselves more autonomously.
In contrast to the Sudanese females, British females can use mobile phones in public places as
symbolic bodyguards (Lasen, 2002b). In this
context, mobile phones are used as a technique
to defend your private space within areas that
are heavily populated with unknown strangers
(Cooper, 2000; Haddon, 2000). As Goffman
(1963) has remarked, women especially do not
like to show themselves alone in public places,
because this may indicate that they are not in
a relationship: a condition which (1) provides
a bad impression of their social status and (2)
leaves them in a vulnerable situation which can
be acted upon by unknown males. To deal with
these situations, the mobile phone is quite useful, as it works as a safety net and indicates that
this person has their social networks and is not
isolated (Plant, 2002).
The other significant result reported in this
study is that the Sudanese are more likely to
switch off their mobile phones in places of worship. Measuring these results against the Hofstede

1579

Mobile Phone Use Across Cultures

typology, the Sudanese score high on uncertainty

avoidance scalereligion is valued and greatly
respected. Peoples attitude towards switching off
mobile phones in places of worship in the Sudan
is therefore expected. It is also related to the high
scores Sudan has on power distance, as roles are
set, and religious men are very much valued and
respected in the society, so both the Muslim and
the Christians in the Sudan tend to be aware of the
importance of switching off their mobile phones
in places of worship. This result could also be
related to the reduced number of people in the
UK attending places of worship.
The Sudanese also appear more willing to
switch off their mobile phones during meetings
than the UK participants. This attitude may be
related to their high score in the power distance
dimension where people are expected to respect
the structure, rules, and the norms of the setting
where they are currently present.
As for the British disinclination to switch off
their mobile phones during meetings, it might be
related to the individualistic feature of the British
society, where time is valued, and there is a push
for making good use of it. It may also be related
to the maturity of British mobile phone adoption
where mobile phones have blurred the borders
between business and social rules. In relation
to this, Churchill (2001) found that the mobile
phones in the UK are used to form and maintain
both work and leisure relationships.

Conclusion
The increased use of mobile phones by people
from different cultural backgrounds has become
an integral part of our world phenomena, yet to
date the impact of cultural differences on the way
people use their mobile phonesand its implications on mobile phone designhas failed to
be investigated comprehensively. As this article
illustrates, mobile phone users with cultural dif-

1580

ferences were found to use their mobile phones in

different ways, and their attitudes may have been
influenced by their cultural norms. Although one
can argue that cultural norms can be reshaped
by technology, results obtained from this study
indicate that cultural heritage would appear to
influence users mobile phone behaviour. The
results obtained from this study also suggest that
mobile phone designers need to develop a richer
understanding of culture in order to develop
mobile phones that satisfy culturally specific
needs, and thus support mobile phone users in
their current and potential future communication
activities. This is an issue we intend to explore in
the next phase of our research.

References
Bayes, A., Von Braun, J., & Akhter, R. (1999). Village pay phones and poverty reduction: Insights
from a Grameen bank initiative in Bangladesh.
Bonn: Center for Development Research, Universitat Bonn.
BBC. (2003). A report by the Worldwatch Institute
in Washington. Mobile phone use grows in Africa.
Retrieved October 9, 2007, from http://news.bbc.
co.uk/1/hi/world/africa/3343467.stm
Burns, T. (1992). Erving Goffman. London:
Routledge.
Churchill, E. (2001). Getting about a bit: Mobile
technologies & mobile conversations in the UK
(FXPL international Tech. Rep. No. FXPAL.
TR.01-009).
Ciborowski, T.J. (1979). Cross-cultural aspects of
cognitive functioning: Culture and knowledge.
In A.J. Marsella, R.G. Tharp, & T.J. Ciborowski
(Eds), Perspectives on cross-cultural psychology.
New York: Academic Press Inc.
Cooper, G. (2000). The mutable mobile: Social
theory in the wireless world. Paper presented

Mobile Phone Use Across Cultures

at the Wireless World Workshop, University of

Surrey.
De Mooij, M. (2003). Consumer behavior and
culture. Consequences for global marketing and
advertising. Thousand Oaks, CA: Sage Publications Inc.

Ling, R. (2004). The mobile connection the cell

phones impact on society. San Francisco: Morgan
Kaufmann.
Love, S. (2005). Understanding mobile human-computer interaction. Elsevier Blueworth
Heinemann: London.

De Mooij, M., & Hofstede, G. (2002). Convergence and divergence in consumer behavior:
Implications for international retailing. Journal
of Retailing, 78, 61-69.

Love, S., & Perry, M. (2004). Dealing with mobile

conversations in public places: Some implications
for the design of socially intrusive technologies.
Proceedings of CHI 2004, Vienna (pp. 24-29).

Donner, J. (2005). The rules of beeping: Exchanging messages using missed calls on mobile phones
in Sub-Saharan Africa. Paper presented at the
55th Annual Conference of the International
Communication Association: Questioning the
Dialogue, New York.

Murtagh, G.M. (2001). Seeing the rules: Preliminary observations of action, interaction and mobile
phone use. In B. Brown, N. Green, & R. Harper
(Eds.), Wireless world. Social and interactional
aspects of the mobile age (pp. 81-91). London:
Springer-Verlag.

Evers, V., & Day, D. (1997). The role of culture

in interface acceptance. In M.S. Howard, J. Hammond, & G. Lindgaard, Proceedings of the Human
Computer Interaction INTERACT97 Conference
(pp. 260-267). Sydney: Chapman and Hall.

Palen, L. (2002). Mobile telephony in a connected

life. Communications of the ACM, 45(3), 78-82.

Goffman, E. (1963). Behaviour in public places.

Notes on the social organization of gatherings.
Free Press.

Palen, L., Salzman, M., & Youngs, E. (2000).

Going wireless: Behaviour and practice of new
mobile phone users. In Proceedings of the Conference on Computer Supported Cooperative Work
(CSCW00) (pp. 201-210).

Haddon, L. (2000). The social consequences

of mobile telephony: Framing questions. Paper
presented at the seminar Sosiale Konsekvenser
av Mobiltelefoni, organised by Telenor, Oslo.

Plant, S. (2002). On the mobile: The effects of

mobile telephones on social and individual life.
Motorola, London. Retrieved October 9, 2007,
from http://motorola.com/mot/doc/0/267_MotDoc.pdf

Hofstede, G. (1980). Cultures consequences:

International differences in work-related values.
Beverly Hills: Sage Publications.

Rust, J., & Golombok, S. (1989). Modern pychometrics: The science of psychological assessment.
New York: Routledge.

Honold, P. (1999). Learning how to use a cellular

phone: Comparison between German and Chinese
users. Jour Soc. Tech. Comm, 46(2), 196-205.

Smith, A., Dunckley, L., French, T., Minocha, S.,

& Chang, Y. (2004). A process model for developing usable cross-cultural Websites. Interacting
with Computers, 16, 63-91.

Lasen A. (2002a). The social shaping of fixed

and mobile networks: A historical comparison.
DWRC, University of Surrey.
Lasen, A. (2002b). A comparative study of mobile
phone use in London, Madrid and Paris.

Smith, A., French, T., Chang, Y., & McNeill, M.

(2001). E-culture: A comparative study of efinance
Web site usability for Chinese and British users. In
D. Day & L. Duckley (Eds.), Designing for global

1581

Mobile Phone Use Across Cultures

markets. Conference (6th. 2001). Proceedings of

the Third International Workshop on Internationalisation of Products and Systems (pp. 87-100).
Buckinghamshire: The Open University.

Sun, H. (2003). Exploring cultural usability: A

localization study of mobile text messaging use.
Paper presented at CHI 2003, Ft. Lauderdale,
FL.

This work was previously published in the International Journal of Technology and Human Interaction, edited by B. C. Stahl,
Volume 4, Issue 2, pp. 35-51, copyright 2008 by IGI Publishing, formerly known as Idea Group Publishing (an imprint of IGI
Global).

1582

1583

Chapter 4.10

Gender and Computing at

University in the UK
Ruth Woodfield
University of Sussex, UK

INTRODUCTION
In the late 1970s, womens progress and participation in the more traditional scientific and technical fields, such as physics and engineering, was
slow, prompting many feminist commentators to
conclude that these areas had developed a nearunshakeable masculine bias. Although clearly
rooted in the domains of science and technology,
the advent of the computer was initially seen to
challenge this perspective. It was a novel kind
of artefact, a machine that was the subject of
its own newly created field: computer science
(Poster, 1990, p. 147). The fact that it was not
quite subsumed within either of its parent realms
led commentators to argue that computer science
was also somewhat ambiguously positioned in
relation to their identity as masculine. As such,
it was claimed that its future trajectory as equally
masculine could not be assumed, and the field of
computing might offer fewer obstacles and more
opportunities for women than they had experienced before. Early predictions of how womens
role in relation to information technology would

develop were consequently often highly optimistic

in tone. Computing was hailed as sex-blind and
colour-blind (Williams, Cited in Griffths 1988,
p. 145; see also Zientara, 1987) in support of a
belief that women would freely enter the educational field, and subsequently the profession, as
the 1980s advanced.
During this decade, however, it became increasingly difficult to deny that this optimism was
misplaced. The numbers of females undertaking
undergraduate courses in the computer sciences
stabilised at just over a fifth of each cohort through
the 1980s and 1990s, and they were less likely to
take them in the more prestigious or researchbased universities (Woodfield, 2000).
Tracy Camps landmark article The Incredible
Shrinking Pipeline (1997), using data up to 1994,
plotted the fall-off of women in computer science
between one educational level and the next in the
US. It noted that a critical point was the drop-off
before bachelor-level studycritical because the
loss of women was dramatic, but also because a
degree in computer science is often seen as one
of the best preparatory qualifications for working

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Gender and Computing at University in the UK

within a professional IT role1. The main aim of

this article is to examine how the situation has
developed since 1994, and within the UK context.
It will also consider its potential underlying causes,
and possible routes to improvement.

BACKGROUND
In the UK, throughout the 1990s and into the
new millennium, the achievements of secondary school-age girls (11-16 years) progressed
significantly in the more traditional scientific and
technical subjects, and began surpassing those
of boys. Before an age when some curriculum
choice is permitted (14 years old), girls perform
better in science. Furthermore, although fewer
of them take science once they have choice, they
continue to surpass boys achievements in the
area. Higher proportions of girls now gain an
A-C grade pass in their GCSE examinations in
chemistry and biology and physics (Department
of Trade & Industry (hereafter DTI), 2005; Equal
Opportunities Commission (hereafter EOC,
2004)). In terms of A levels, the qualifications
usually taken at the end of the first two-year period
of non-compulsory education (16-18 years), girls
also proportionately achieve more A-C passes in
these subjects (EOC, 2004).
Achievements in computing courses have
followed this trend. Over the last decade, girls
have gained around 40% of GCSE qualifications
in computer studies, and they are now more
far likely to gain an A-C pass than their male
counterparts (EOC, 1994-2004). Nevertheless,
at A level, when students traditionally specialise
in three or four subjects, the trend has been for
the majority of girls to opt out of computing. In
1993, in England, girls only accounted for 13%
of students deciding to further their computing
education to A level standard in England (EOC,
1994). By 2003, this picture had significantly
improved, with girls comprising 26% of those
studying computing or information technology

1584

A level (Joint Council for Qualifications, 2004).

Although this still represents a substantial leak
between one educational level and its successor,
it is noteworthy that girls have recently become
proportionately more likely to gain the top grades
in these qualifications as well (DTI 2005; Joint
Council for Qualifications, 2004).

COMPUTER SCIENCE AT THE

HIGHER EDUCATION LEVEL2
As Figure 1 indicates, the proportion of women
within computer science courses at tertiary levels
remained fairly static between 1994-2003, despite
the improving proportion of them taking computer
science at A Level over the same period. Although
there appears to be a slight increase between 2002
and 2003, this is likely to be due to changes in the
way graduate statistics in the UK were calculated
between these two years3. On average, women
comprised 22% of those completing a degree in
the area over the period.
In the UK, female applicants to computer
science courses differ in key respects from their

Figure 1.

Note. Based on original analysis of Higher Education

Statistics Agency data provided under ad hoc data
enquiry: 23148

Gender and Computing at University in the UK

male counterparts. They are especially likely to

cite their interest in the subject as the main reason
for studying it and are committed to finding the
right course above other considerations, such
as location of university and so forth (Connor,
Burton, Pearson, Pollard, & Regan, 1999; Craig,
Galpin, Paradis, & Turner, 2002b; Millar & Jagger,
2001). This interest would seem to be intrinsic,
as they are less likely to cite future employment
prospects as a motivating factor for their subject
choice than their male counterparts (Millar &
Jagger, 2001). Female computer science applicants
have indicated that, in order to aid course choice,
they review a wide range of information and are
especially influenced by the information provided
in university prospectuses and by experiences of
pre-entry campus visits. Indeed, information deriving from universities is deemed far more helpful
than advice from secondary school teachers and
careers advisers (Connor et al., 1999; Craig et al.,
2002b; Millar & Jagger, 2001).
Women are slightly more likely to be accepted
for a degree within the broad area of mathematical sciences and Informatics, of which computer
science is a subset, probably due to their better
success rates at A-levels. They are also more likely
to complete their degrees (DTI, 2005).
Once on their degree courses, although they
are now over-represented within higher education as a whole (comprising 56% of new graduates), as Figure 1 has illustrated, women remain
significantly outnumbered within each cohort of
computer science undergraduates. This is in a
context where computer science is attracting more
undergraduates year-on-year whereas physics,
chemistry, and engineering are attracting fewer,
and mathematical sciences broadly the same
numbers (DTI, 2005). As Figure 2 chronicles,
between 1994/5-2002/3 the numbers of men
and women completing degrees in this area in
the UK climbed throughout the period, with the
percentage of men climbing slightly faster than
the percentage of women. In 1995, 5.5% of all
male graduates qualified in computer sciences, as

against only 1.5% of all females. By 2003, 11.8%

of males were graduating in this area, as against
2.8% of females. While there are grounds for
optimism insofar as the proportionate increases
during this period for women were comparable
to those of men, it remains of concern that while
more than a tenth of male graduates take their
degrees in this area, less than 3% of females do.
This is significant in the context of the link between
computer science and professional-level IT work,
and the critical role computer science knowledge
plays in our society more generally.
Women are also generally under-represented
among the academic faculty of computer science
at university, although their numbers in Lecturer
(28%) and Researcher (30%) ranks remains better
than we might expect given their undergraduate
representation (DTI, 2005). They are, however,
disproportionately underrepresented in the more
senior academic ranks (DTI, 2005).
Interestingly, however, as Figure 3 shows, of
those taking degrees in this area, women fare no
worse than men in terms of classification achieved,

Figure 2.

Note. Based on original analysis of Higher Education

Statistics Agency data provided under ad hoc data
enquiry: 23148

1585

Gender and Computing at University in the UK

Figure 3.

Note. Based on original analysis of Higher Education

Statistics Agency data provided under ad hoc data
enquiry: 23148

and their performance would seem to be improving. In 1994/5, 9.6% of men achieved the top
grade of degreea First Classas opposed to
7.6% of women. By 2003, however, 11.9% of men
were achieving this grade against 11.6% of men.
Furthermore, women have also been marginally,
but consistently, more likely to obtain a good
degreea First Class or Upper Secondthan
men since 1994, and by 2003 just over 50% of
them achieved this standard as against just under
47% of men.

FUTURE TRENDS and

CONCLUSION
Once this picture is digested, we can see that there
are some grounds for cautious optimism about
the future: an increase in female participation at
A level; improvements in female performance
in relation to both pre-university and university
qualifications; and evidence of above-average

1586

interest in computer science on the part of female

undergraduates. Notwithstanding these changes, a
key question is begged, however, given the context
of the overall increase in the importance of the
subject: Why is the under-representation of women
that established itself in the field in the 1980s and
1990s not correcting itself more quickly?
It is still claimed by some that women are
naturally less predisposed than their male counterparts to be attracted to, or to perform well
within, the sciences in general (McCrum, 1996)
and computer science in particular (Craig et al.,
2002b). This position, however, is increasingly
difficult to sustain in the light of changes in
female participation and performance rates in
educational computing, and the fact that there
are significant local variations in the uptake and
experience of university-level computer science
courses (Cahoon, 2001; Craig et al., 2002b; Millar
& Jagger, 2001). As Cahoon has suggested, it is
much more likely that social factors are at play to
explain female under-representation as women
can and do succeed in Computer Science when
conditions do not deter them (2001, p. 1).
These social factors would seem from the
evidence to fall under two main areas. First,
females, including female computer science
undergraduates, project a far more negative image of IT work than males. As their educational
choices are so closely bound up with their projected
career trajectories (EOC, 2001; Miller, Neathey,
Pollard, & Hill, 2004), it is reasonable to assume
that this tendency has impacted on female recruitment levels to the subject at university. Secondly,
female experiences of computer science degrees
would seem to be less positive than those of their
male counterparts, and it is very likely that the
stories of such experiences become part of the
information young women draw on when making their decisions about what subjects to study
at university.
To take the first point, the available evidence
suggests that women taking computer science
degrees, and courses that might prepare them for

Gender and Computing at University in the UK

undergraduate study, are generally not likely to be

fully aware of the benefits of a career in IT (Millar
& Jagger, 2001; Peters, Lane, Rees, & Samuels,
2003), while simultaneously being more likely to
be alive to its real and perceived disadvantages.
They are more likely to project, for instance, that
it will be solitary work, when they, as a group,
have a preference for team-based and sociallyembedded computing (Craig et al., 2002b). The
problems facing them should they opt for a career
in computing cannot be ignored, substantial as
they are. Female computer science graduates
are less likely than males to find employment at
the end of their degrees, despite their tendency
to have achieved a higher degree classification,
and to be better educated more generally. Once
in employment, they are likely to receive less
pay for comparable work, and many describe a
working ethos that systematically privileges men
over women (Millar & Jagger, 2001; Woodfield,
2000). In a context where female educational
achievements are outstripping those of males at
every level of education in the UK, and female
work aspirations are becoming increasingly
clear and uncompromising, it is almost certainly
the case that the negative perception of IT work
plays a critical role in determining the numbers
of women choosing courses that qualify them
for it. As Millar and Jagger have claimed, problems appear in the relationship between women
and the computer sciences once women begin
to structure their education towards a career
(2001, p. A-iv). There are substantial benefits
to IT work, however, that could be clarified and
communicated to those selecting their GCSE, A
level and university courses: salary levels, notwithstanding the gender gap, tend to be higher
than average; existing female IT professionals
rarely describe themselves as bored or unfulfilled
in terms of the work per se. More fundamentally
than the need to communicate these benfits, the
sector needs to consolidate the beneficial aspects
of IT careers for women, and work to eradicate
the more negative aspects.

The earlier communication about the advantages of IT work takes place, the better. It
is claimed that educational choices, based on
projected career costs and benefits, are already
taking place when UK teenagers face their first
opportunity to specialise, around age 14. This is
also an age when young males and females are
most sensitive to gender-appropriate and genderinappropriate career choices. It is of concern then,
that there seems to be so few reports of positive
representations of professional IT work as a viable
career goal for girls (and their parents) by teachers
or careers advisers within secondary schools, and
that the provision of female-friendly spaces for
computing is still considered cutting edge (Craig
et al., 2002b; Millar & Jagger, 2001; Peters et
al., 2003). Improvements in the knowledge, and
sometimes the attitude, of adults playing a key
role in pre-university course choices, could make
a significant difference to female participation
rates at this level.
As well as the role played by the image of IT
work, the reported experiences of those undertaking a computer science degree must also partially
account for the level of female participation at
university. There is a wealth of evidence to suggest that female students under-representation
on courses creates its own problems in terms
of their confidence, adjustment, enjoyment, and
achievement levels. They report themselves to
feel pressurised to adapt to a male-dominated
and male-oriented educational regime in order to
survive (see, for example: Bjorkman, Christoff,
Palm, & Vallin, 1998; Blum, 2001; Margolis &
Fisher, 2002; Peters et al., 2003). Evidence from
those institutions that have attempted to make
significant changes within their computer science
programmes so that women will feel less alienated
suggests that the have enjoyed far greater female
participation and approval rates as a result (Connor
et al., 1999; Margolis & Fisher, 2002; Millar &
Jagger, 2001). In doing so, they point to a wish list
of best practices for others interested in doing the
same. These include building a curriculum that

1587

Gender and Computing at University in the UK

assumes a minimal level of previous IT experience, providing mentoring systems and femalefriendly learning spaces, and ensuring that faculty
are sensitive to the needs and abilities of female
students (Blum, 2001; Cahoon, 2001; Hefce 2002;
Margolis & Fisher, 2002). One course in the UK,
where such considerations have been central for
some time, reports high rates of female uptake of
degree places, but also that 100% of women on
the course maintained a desire to become an IT
professional after the completion of their degree
(Craig et al., 2002).
The finding that women are heavily reliant
on the information provided in prospectuses is
heartening in this regard. It confirms that there
is a clear window of opportunity for the more
progressive universities through which they
can encourage potential applicants and reduce
reliance on general anecdotal or impressionistic
information that may not apply to their courses.
If more institutions embraced the same ethos, a
critical mass of women entering computer science
undergraduate programmes could be achieved
in the next decade. This, in turn, could impact
significantly on the general culture of educational
computing and beyond into the workplace, which
could lead to improvements in the sectors image.
The beneficiaries of this would not just be the
women who would otherwise have turned away
from the subject at degree level, but also the UK
economy and society which is arguably at present drawing its computing scientists and skilled
IT professionals from an artificially restricted
pool.

understanding of the male dominated computing

culture and its influence on women. SIGCSE Bulletin, 30(2), 64-69.

References

EOC. (2001). Women and men in Britain. Sex

stereotyping: From school to work. Retrieved
from http://www.eoc.org.uk

Blum, L. (2001, November). Transforming the

culture of computing at Carnegie Mellon. Computing Research News, 13(5), 2,6,9.
Bjorkman, C., Christoff, I., Palm, F., & Vallin,
A. (1998). Exploring the pipeline: Towards and

1588

Cahoon, J. M. (2001). Towards improving female

retention in the computer science major. Communications of the ACM, 44(5), 108-114.
Camp, T. (1997). The incredible shrinking
pipeline. Communications of the ACM, 40(10),
3-110.
Connor, H., Burton, R., Pearson, R., Pollard,
E., & Regan, J. (1999). Making the right choice:
How students choose Universities and Colleges.
UK: Institute of Employment Studies & Universities.
Craig, A., Galpin, V., Paradis, R., & Turner, E.
(2002b). What is computing? The perceptions of
university computing students. A Panel Presentation. Grace Hopper Celebration of Women in
Computing, British Columbia, Canada.
Craig, A., Paradis, R., & Turner, E. A. (2002a).
A gendered view of computer professionals:
Preliminary results of a survey. SIGCSE Bulletin,
34(2), 101-104.
DTI (Department of Trade and Industry). (2005).
Gender and innovationStatistics. Retrieved
from www.set4women.gov.uk
EOC (Equal Opportunities Commission). (19992004). Facts about women and men in Great
Britain. Retrieved from http://www.eoc.org.uk
EOC. (1994). Some facts about women. Retrieved
from http://www.eoc.org.uk

Hefce (Higher Education Funding Council for

England). (2002). Performance indicators in
higher education. Retrieved from http://www.
hefce.ac.uk

Gender and Computing at University in the UK

Higher Education Statistics Agency. (2004).

Original analysis based on ad hoc data enquiry:
23148.
Joint Council for Qualifications. (2004). Student
performance analysis. National percentage figures
for A level grades. Retrieved from http://www.
bstubbs.co.uk/gender/boys-a.htm
Margolis, J., & Fisher, A. (2002). Unlocking the
clubhouse: Women in computing. Cambridge,
MA: MIT Press.
McCrum, G. N. (1996). Gender and social inequality at Oxford and Cambridge Universities. Oxford
Review of Education, 22(4), 369-397.
Millar, J., & Jagger, N. (2001). Women in ITEC
courses and careers. Brighton, UK: Science and
Technology Policy Research Unit & Institute of
Employment Studies.
Miller, L., Neathey, F., Pollard, E., & Hill, D.
(2004). Occupational segregation, gender gaps
and skill gaps. Brighton, UK: EOC Working
Paper 15.
Peters, J., Lane, N., Rees, T., & Samuels, G.
(2003). SET FAIR: A report on women in science,
engineering, and Technology. HMSO, UK: Office
of Science & Technology.
Poster, M. (1990). The mode of information:
Poststructuralism and social context. Cambridge,
UK: Polity Press.
Williams. (1998). Strong feelings about computers.
In M. Griffiths (Ed.), Womens Studies International Forum, 11(2), 145-154.
Woodfield, R. (2000). Women, work, and computing. Cambridge, UK: Cambridge University
Press.
Zientara, M. (1987). Women, technology, and
power, ten stars and the history they made. New
York: AMACON.

Key Terms
Classification of Degree: UK degrees are
ascribed classifications that usually correspond
to the following percentage results (averaged over
all assessments contributing to the final grade):

First 70%+
Upper Second or 2:1 60-70%
Lower Second or 2:2 50-60%
Third 40-50%
Pass 30-40% (a bachelor degree without
honours)

Equal Opportunities Commission/EOC:

The Equal Opportunities Commission is an independent, non-departmental public body, funded
primarily by the government. It provides free
research reports and statistics chronicling the
position of women and men in all aspects of UK
life, including work and education. Web address:
http://www.eoc.org.uk
GCE A Level: The General Certificate of Education, A level (Advanced Level) is usually taken
2 years after GCSEs and as a non-compulsory,
pre-entry qualification for University-level Study.
Students usually take 3 or 4 specialised subjects
at this level, which they are free to select.
GCSE: The General Certificate of Secondary
Education is usually taken by 15-16 year olds in
a range of subjectssome compulsory (English,
maths, and science) and some optional (including
Computer Studies)at the end of their compulsory schooling. Achieving 5 GCSE grades A-C is
considered the desirable educational benchmark
to be reached by this age.
Good Degree: A term used in academic
literature and common parlance to indicate a
degree which has been classified as an Upper
Second or First, (i.e., in the top two degree
classifications. It is normally assumed that those
going on to postgraduate study will have such a

1589

Gender and Computing at University in the UK

degree, and they are preferred in many employment contexts.

HESA (Higher Educational Statistics
Agency): HESA was set up in 1993 by the UK
Government to act as a central source for higher
education statistics and has become a respected
point of reference. Web address: http://www.
hesa.ac.uk.
IT/Computer Professional: An individual
working within the IT/computer sector within a
complex and skilled role that is classified within
the category Professional occupations (e.g.,
software engineers), or sometimes within Associate professional and technical occupations
(e.g., computer programmers)in the National
Statistics classifications, Standard Occupational
Classification Codes (2004): http://www.statistics.
gov.uk

EndNotes
1

Elsewhere, degrees in mathematical sciences

and/or business and administration degrees
are included in analyses of gender representation within degrees associated with IT and
IT employment (see, for instance, Millar &
Jagger, 2001). Here, unless stated otherwise,
I will be using the narrower group of computer science degrees, defined, according
to Higher Educational Statistics Agency
programme categories as including: computer science, information systems, software
engineering, artificial intelligence, and other
programmes within computer science.
All of the statistics included within this section are based on original analysis of Higher
Education Statistics Agency data provided
under ad hoc data enquiry: 23148.
HESA changed its discipline classification
scheme between the years 2001/2 and 2002/3,
making comparisons between pre-2002/3
data and post-2002/3 data problematic.

This work was previously published in the Encyclopedia of Gender and Information Technology, edited by E. Trauth, pp. 365371, copyright 2006 by Information Science Reference, formerly known as Idea Group Reference (an imprint of IGI Global).

1590

1591

Chapter 4.11

Gender and Information

Technology in
Rural Bangladesh
Lutfor Rahman
Association for Advancement of Information Technology (AAIT), Bangladesh
Nusrat Rahman
University of Dhak and Association for Advancement of Information Technology (AAIT), Bangladesh

INTRODUCTION
Information technology (IT) is transforming our
personal, social, cultural, economic and political
lives. But women in developing countries do not
have equal access to knowledge, due to the fact that
they do not have access to the new technologies
at the same level as western European women.
They need to understand the significance of new
technologies and use them in relevant fields. IT
can offer immense opportunities for virtually all
girls and women in developing countries, including poor women living in rural areas.
Developing countries like Bangladesh are usually seen as problematic hosts for ITs because most
developing regions of the country lack economic
resources and indigenous techno-scientific capabilities to develop and deploy IT infrastructures.

The regions also tend not to make the best use of

opportunities of technology transfer.
The wider adoption of IT in the 1990s promised a globally connected community of equal
participants in electronic networks. The Internet
has enabled people to communicate regardless
of race, sex, physical ability, location and social
background (GVU, 1998). IT now offers access to
a huge pool of information. The Internet enables
worldwide communication that is cheaper, faster
and more flexible than older media like e-mail,
telephone, telex or fax.

BACKGROUND
In early 2000, an initiative was taken to develop
a small river island (char) of Bangladesh through

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Gender and Information Technology in Rural Bangladesh

establishment of a resource centre named Indigenous Science and Technology at Ikrail (ISTI), and
using IT as the main vehicle for development. IT
is a livelihood to many women of the developed
countries but is almost unknown to the women
and girls of the river island.
Women in Bangladesh are seen in the frontlines
to fight against hunger, poverty and environmental
degradation. Because of lack of awareness about
the benefits of IT, they cannot raise their problems
for solution to the proper authority. IT can benefit
women in various ways and can facilitate their
participation in different sectors and regions. It
can provide information women need to improve
their own well-being and that of their families.
The introduction of computers in offices can improve the quality of work and scope for women
in data entry, analysis, programming, clerical
and administrative occupations in Bangladesh.
IT could allow them to exchange views, opinions
and information with women of other regions
and countries.
The situation for rural populations in many
regions of the world is characterized by geographical distance from urban centres that offer employment and income, education, cultural events and
public services. IT bears the potential to improve
the situation of rural people in several ways. For
example, it is agreed that early diagnosis of medical issues can prevent many casualties. Once a
patient comes to the health centre, using various
sensors, information can be collected and communicated to an expert at hospitals in the district
headquarters. The primary healthcare centre could
be located in a rural area or in a mobile centre
installed in a van.
IT offers access to information and extension,
to education and training communication and
networking in online discussions. IT also offers
access to employment and income.

1592

Infrastructural Problems
Poor healthcare can result from a lack of good information. Establishing reliable communications
may be one of the most important priorities for
improving healthcare and education. Many rural
areas in the globe have no or outdated telephone
lines that can transmit Internet-based data. The
lack of infrastructure can be overcome by wireless
technology (e.g., radio modems). Mobile telecentres can be a solution for the target group.

HEALTHCARE IT PROJECT
A significant number of women scientists, researchers and technologists work in rural Bangladesh. They are disadvantaged and traditionally
underrepresented in most respects. Their knowledge and skills are unrecognized, underused and
under valued. As such, they are in greater need
of upgrading their skills, especially in the fast
advancing world of information and communication technologies (ICTs), which might enable
them to connect their global colleagues, sources
of information and global knowledge wherever
they may be located.
A survey conducted in early 2000 among 515
women professionals of various disciplines spanning life sciences, social sciences, physical sciences, mineral sciences, engineering, technologies
and medical sciences identified that they are almost
illiterate in IT, although they are well qualified and
experienced in their respective fields. At the first
step, the survey was conducted over the senior
professors of the public and private universities,
scientists of the Bangladesh Council of Scientific
and Industrial Research (BCSIR), researchers of
institutes of forest research, housing research,
fuel research, jute research and water resource
and geological research.

Gender and Information Technology in Rural Bangladesh

They have very poor knowledge about the

power and benefit of IT. This trend was found
true at all levels of educational backgrounds.
The survey revealed that the illiteracy rate in IT
among the medical doctors are high, because the
doctors are normally resistant to new technology,
particularly IT.
To reach the benefit to the rural communities,
participants for IT training were selected from
among applicants spanning all sectors, giving
priority to the rural community. The representation covered various national economy, especially
health, environment, education, natural resources
(mineral resources), energy, industry, agriculture,
fisheries and forestry. Special care was taken in
selection of participants so that the maximum
number of organizations and individuals would
benefit from the program. Training programs on
IT for the duration of 2 weeks were organized in
different private universities at a certain interval
with 20 participants in each phase. The programs
were initially organized at the capital city, and
later, the programs were arranged at the regional
institutes and universities. The regional programs
were mainly organized for the medical doctors
working in the local hospitals, clinics, medical
colleges and universities. Selection of participants
from the women medical doctors was really a
tough job. Neither the medical doctors nor their
family members wanted to spare 2 weeks for attending a training course that they feel are good
for their children, not for the doctors. Strategies
were adopted to attract the medical doctors. As
soon as they realized the importance of IT, they
invested more of their valuable time to learn
further applications. Ninety-two percent of the
trained medical doctors were found to utilize their
gained knowledge intelligently in their respective
fields. The result was obtained from the follow-up
visit by the project staff.

FUTURE TRENDS
Bangladesh is a country of rivers that create islands (chars) of different shapes and sizes. They
are created because of river erosion. The ISTI
resource centre is located in one such island.
The island is circular in shape and its area varies
from 20 to 30 square kilometers, depending on
seasons. Approximately 15,000 people live in the
island community. The people depend on local
cultivation for survival. Their average monthly
earning is $40 (United States dollars). The area
is fertile and rice, wheat, peanut, jute, sugarcane,
and different types of vegetables and fruits are
common. All work is performed manually. The
people are deprived of basic infrastructure. There
is no electricity, phone system or sanitation. The
only source of water is contaminated with arsenic.
There are no health clinics or doctors in the area.
Many adults are illiterate, and curable diseases,
such as cholera, malaria and tuberculosis, are
pervasive.
The ISTI resource centre has been established
with the objective of developing this isolated
region with proper education, empowering local
people with relevant technology, particularly
IT, and providing them with medical facilities
using modern techniques (such as telemedicine
and telehealth care). The ISTI resource centre,
equipped with basic medical test equipment like
stethoscopes, blood pressure meters and blood
testing chemicals, and connected with a hospital
located in the district town, might save lives of
many women and girls. Just a computer with a
modem connected with wireless local loop (WLL)
help solve most of the serious problems of local
women by the women medical doctors empowered
with IT. The ISTI resource centre is working as role
model for several million disadvantaged women
living in the river islands of Bangladesh.

1593

Gender and Information Technology in Rural Bangladesh

CONCLUSION
This training program has been particularly
suitable and effective for relatively marginalized
groups and remote communities in a developing
country because of its low cost and the use of
local organizers and trainers, courses especially
designed to meet the participants needs and in situ
follow up. The direct involvement in the courses of
senior policy makers, employers and local experts
has proven to be absolutely crucial in gaining
support for the continuation and expansion of the
course and, perhaps more importantly, effecting
attitude change to womens roles and capabilities
in science and medicine. This was underlined by
the increasing cooperation extended by them to the
project team during preparation and execution of
the project. The outcome of the whole training has
been impressive. Participants value to the work
place has increased, some attained promotions
and some changed careers.
The final evaluation report shows that most
women medical doctors and other health-related
professionals empowered with IT have expressed
their opinions that every medical doctor must
attend relevant technologyparticularly IT
courses if they want to enhance their services
and maintain their personal secrecies. Some mentioned that IT is a magic and all-pervasive tool.

References
Babul, R. (2005). Empowering women through
ICT.
Desai, U. B. (2000). COMMSPHERE, Communication sector for health care.

Johnson, C. J., & Rahman, L. (2001). In Proceedings of the Community Technology Conference at
Murdoc University, Australia (Vol. 1).
Rahman, L. (1999). Paths to prosperityCommonwealth science and technology for development.
Rahman, L. (2004). Leading role in ICTenabled
human development in Bangladesh. In Proceedings of the AISECT 2004 International Conference
on ICT in Education and Development.
Tomasevic, R. (2000). Putting the Web to work
for womenthe nature of gender.

KEY TERMS
Information and Communication Technology (ICT): ICT covers computing, electronics
and telecommunications.
Medical Professionals: Medical doctors and
scientists in health-related subjects, such as nutrition, bio-chemistry, psychology and so forth.
Researchers: Graduates and post graduates
engaged in research activities.
Scientists: Science graduate with some research experience in scientific fields.
Technologists: Engineers and scientists with
experience in hardware.
University: A university is characterized by a
wide variety of teaching and research, especially at
a higher level, that maintains, advances, disseminates and assists the application of knowledge,
develops intellectual independence and promotes
community learning.

This work was previously published in the Encyclopedia of Gender and Information Technology, edited by E. Trauth, pp. 423425, copyright 2006 by Information Science Reference, formerly known as Idea Group Reference (an imprint of IGI Global).

1594

1595

Chapter 4.12

Chinese POS Disambiguation

and Unknown Word Guessing
with Lexicalized HMMs
Guohong Fu
The University of Hong Kong, Hong Kong
Kang-Kwong Luke
The University of Hong Kong, Hong Kong

ABSTRACT
This article presents a lexicalized HMM-based
approach to Chinese part-of-speech (POS) disambiguation and unknown word guessing (UWG).
In order to explore word-internal morphological
features for Chinese POS tagging, four types of
pattern tags are defined to indicate the way lexicon
words are used in a segmented sentence. Such patterns are combined further with POS tags. Thus,
Chinese POS disambiguation and UWG can be
unified as a single task of assigning each known
word to input a proper hybrid tag. Furthermore, a
uniformly lexicalized HMM-based tagger also is
developed to perform this task, which can incorporate both internal word-formation patterns and
surrounding contextual information for Chinese
POS tagging under the framework of HMMs.
Experiments on the Peking University Corpus

indicate that the tagging precision can be improved

with efficiency by the proposed approach.

INTRODUCTION
While a number of successful part-of-speech
(POS) tagging systems have been reported for
English and many other languages over the past
years, it is still a challenge to develop a practical
Chinese POS tagger due to the language-specific
issues in Chinese POS tagging. First, there is not
a strict one-to-one correspondence for a Chinese
word between its POS and its function in a sentence. Second, an ambiguous Chinese word can
act as different POS categories in different contexts without changing its form. Third, there are
many unknown words in real Chinese text whose
POS categories are not defined in the dictionary

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

Chinese POS Disambiguation and Unknown Word Guessing with Lexicalized HMMs

used. Furthermore, there are almost no explicit

morphological features in Chinese words, such as
inflexion, derivation, and capitalization in English,
that can be used directly for POS disambiguation
and unknown word guessing (UWG). All these
factors make it much more difficult to achieve a
high-performance POS tagger for Chinese.
Recent study on POS tagging has focused
on machine-learning approaches, including
hidden Markov models (HMMs) (Brants, 2000;
Weichedel et al., 1993), transformation-based
error-driven learning (TBL) (Brill, 1995), maximum entropy model (Ratnaparkhi, 1996), support vector machines (SVMs) (Nakagawa et al.,
2001), and loglinear models (Fu & Wang, 2002).
Machine-learning approaches have the advantage
of robustness. However, it is difficult for most
machine-learning techniques to keep a balance
between their capacity and their computational
cost (Nakawa et al., 2002). Pla and Molina (2004)
have proved that HMM-based taggers have the
highest training and tagging speed. However, they
usually achieve relatively low tagging precision,
because standard HMMs only take into account
contextual POS information for tagging. On the
contrary, some learning models such as ME and
SVMs are capable of handling much richer lexical information for POS tagging. However, they
usually require higher computational cost, which
inevitably will result in reduction of efficiency
in training and tagging. This will be a serious
problem in processing a large amount of data or
in some online applications. In order to tackle
this problem, some recent work, such as Lee, et
al. (2000) and Pla and Molina (2004) suggested
the use of lexicalization techniques to enhance
the standard HMMs. Their experiments have
shown that the lexicalized models can improve
POS tagging precision without increasing much
computational cost in training and processing.
POS disambiguation and unknown word
guessing are two key issues in developing a
Chinese POS tagger for practical applications.

1596

On the one hand, Chinese is highly ambiguous

with respect to part of speech. Consequently, the
first task of a POS tagger is how to find a proper
POS for each ambiguous word in a sentence. On
the other hand, there are many unknown or outof-vocabulary (OOV) words in real Chinese texts
whose POS categories are not defined in advance
in the system dictionary. So, a practical tagger
should be capable of predicting or guessing with
accuracy the POS categories for these unknown
words in an open-ended text.
Following the line of Lee et al. (2000) and Pla
and Molina (2004), we propose in this article a
unified approach to Chinese POS disambiguation
and unknown word guessing. In order to explore
word-internal morphological features for Chinese
POS tagging, we introduce four types of pattern
tags that indicate the way of a lexicon word to
present itself in a real segmented sentence. Further, we define a hybrid tag set by merging these
patterns with POS tags, with which Chinese POS
disambiguation and unknown word guessing can
be unified as a single process of assigning each
known word (KW) in input a proper hybrid tag.
Moreover, a statistical tagger is developed based
on the uniformly lexicalized HMMs. In this way,
three kinds of features contextual tag information, surrounding lexical information, and
word-internal word-formation patterns can
be incorporated for Chinese POS tagging under
an efficient HMM-based framework. As a consequence, the taggers performance can be improved
with efficiency in training and tagging.
The rest of this article is organized as follows:
The second section presents a novel formulation for Chinese POS tagging. Next a uniform
lexicalization technique is introduced to enhance
the standard HMMs for POS tagging. The fourth
section gives a brief description of the tagging
algorithm. Finally, we report our experimental
results on the Peking University corpus in the
fifth section and give our concluding remarks on
this work in the final section.

Chinese POS Disambiguation and Unknown Word Guessing with Lexicalized HMMs

UNIFYING POS DISAMBIGUATION

AND UNKNOWN WORD GUESSING
In this section, Chinese POS disambiguation and
unknown word guessing is reformulated as one
unified process of known word tagging, based on
the introduction of word formation patterns.

Word Formation Patterns

The input of our current task is a segmented
sentence or a sequence of words1 that might be
either known words listed in the system lexicon
or unknown words unlisted in the system lexicon.
In general, a known word consists of only one
lexicon word, while an unknown word is composed of several lexicon words. As discussed in
Fu and Luke (2005), a lexicon word w has four
possible patterns to form a word in a segmented
sentence: (1) w is segmented as an independent
known word; (2) w is the beginning component
of an unknown word; (3) w is a middle component of an unknown word; (4) w is the ending
component of an unknown word. In this article,
four different tags ISW, BOW, MOW and EOW
are used to denote these patterns, respectively.
Thus, a segmented sentence can be represented
as a sequence of lexicon words accompanied with
their word-formation pattern tags. For example,
theoriginallysegmentedsentence
(zhang1 xiao3hua2 zai4 xiang1gang3 da4xue2
gong1zuo4, Zhang Xiaohua is working at the
University of Hong Kong) can be represented as
/ISW/MOW/EOW/ISW/ISW/ISW
/ISW.

its POS

i
n addition to the

contextual information
surrounding it
. For

example, if a word ends with

the character (xing4, nature), then it is more
likely to be a noun. On the contrary, if a word
ends with the character (hua4, change or a suffix such as ize and -ify), then it should be tagged
as a verb in most cases. In order to explore these
important position features for POS disambiguation and unknown word guessing, a hybrid tag
set is thus designed by merging the above four
patterns with POS tags. In our work, a hybrid
tag has the following format: T1-T2. Where T1
denotes a POS tag and T2 denotes a word-formation pattern tag.
With this hybrid tag set, an original POStagged sentence can be represented equivalently as
a sequence of known words accompanied by their
relevant hybrid tags. For example, an equivalent
expression of the POS-tagged sentence
/nr/nr/p/ns/n/v(ZhangXiaohua is working at the University of Hong Kong)
is /nr-ISW /nr-MOW /nr-EOW /p-ISW
/ns-ISW /n-ISW /v-ISW.
Obviously, the POS category of an unknown word
in a sentence will be determined once each of
its components is assigned a proper hybrid tag.
At this point, the two subtasks in Chinese POS
tagging (i.e., POS disambiguation and unknown
word guessing) thus can be unified as a single
task of tagging each known words in input2, with
a proper hybrid tag indicating its POS category
and its word-formation pattern.

POS Disambiguation and

Unknown Word Guessing
as Known Word Tagging

LEXICALIZED HMMs

Given a real word in a Chinese sentence,

the
position of its

component
lexicon word
s is another
important evidence

that

can be used to determin

e

Given a sequence of known words W = w1w2 ...wn,

a statistical tagger aims to find an appropriate sequence of hybrid tags T = t1t 2 t n that maximizes
the conditional probability P(TW); namely,

Uniformly Lexicalized HMMs

1597

Chinese POS Disambiguation and Unknown Word Guessing with Lexicalized HMMs

in most current POS tagging systems (i.e., N =

1 or N = 2).

At present, standard HMMs have been used

(1)
widely in many NLP applications such as automatic speech recognition (ASR) and POS tagging,
where the probability P(W) can be dropped in
due to their efficiency in training and processing.
that it is fixed for a specific word sequence W.
Furthermore, a variety of theories and algorithms
Thus, we have a general model for POS tagging
have been proposed for building HMMs with
as follows:
efficiency. However, a standard HMM-based
taggern only can take into account contextual
T = arg max P(W | T ) P(T ) = arg max P( w1, n , t1, n ) = arg m
acategory
x P(information
wi | w1,i 1 , t1,ifor
) P(tagging.
t i | w1,i 1 ,The
t1,i 1important
)
T
T
T
i =1
n
contextual lexical features often are ignored. As a
max P( w1, n , t1, n ) = arg m
a x P( wi | w1,i 1 , t1,i ) P(t i | w1,i 1 , t1,i 1 )
consequence, HMM-based taggers usually have
T
T
i =1
relatively low tagging performance. In order to

(2)
enhance tagging performance and give consideration to efficiency in training and tagging at the
In theory, the model in Equation (2) can provide
same time, we employ the uniform lexicalization
the tagging system with more powerful capacity
technique (Lee et al., 2000) to enrich the standard
for disambiguation and unknown word guessing.
HMMs with contextual lexical information.
However, it involves too many parameters to be
In approximating the general model in Equacomputable in practice. To address this problem,
tion (2), the uniformly lexicalized modeling adopts
two modelsthe standard HMM and the lexicaltwo assumptions: the appearance of current word
ized HMMare usually employed to approximate
wi is assumed to depend not only on current tag ti
the general model in Equation (2).
but also its previous N words wiN,i1 and tags tiN,i1,
Actually, the standard hidden Markov modeland the assignment of current tag ti is supposed
ing follows two well-known independent hypothto depend on both its previous N words wiN,i1
eses: The appearance of current word wi depends
and the relevant N tags tiN,i1. Thus, a uniformly
only on current tag ti during known word tagging,
lexicalized HMM can be formulated as
and the assignment of current tag ti depends only
on its N previous tags tiN,i1. Thus, Equation (2)
n
can be rewritten as
T = arg max P( wi | wi N ,i 1 , t i N ,i ) P(t i | wi N ,i 1 , t i N ,i 1 )
P(W | T ) P(T )
T = arg max P(T | W ) = arg max
P(W )
T
T

i =1

( w | wi N ,i 1 , t i N ,i ) P(t i | wi N ,i 1 , t i N ,i 1 )
T = arg max PT( w=i arg
| t i )max
P(t i
| tiP
N ,i 1 )i

T

i =1

i =1

(3)

Equation (3) is a general form of the N-order

standard HMMs, where P(witi) denotes the socalled lexical probability, and P(titiN,i1) denotes
the contextual tag probability. Due to the serious
data sparseness in real applications, only the
first-order or second-order models are employed

1598

(4)

In comparison with standard HMMs, uniformly lexicalized HMMs can handle both contextual
words and contextual tags for the assignment of
hybrid tags to known words, which will result in
an improvement of tagging precision. In view of
the serious data sparseness in higher-order models,
we employ the first-order lexicalized HMMs in
our system.

Chinese POS Disambiguation and Unknown Word Guessing with Lexicalized HMMs

Estimation and Data Smoothing

If a large POS-tagged corpus is available, the
lexicalized HMMs easily can be estimated using the maximum likelihood estimation (MLE)
technique. In principle, MLE estimates parameters
using their relative frequencies that can be directly
extracted from the training data. Obviously, MLE
has the advantage of simplicity. However, it will
yield zero probabilities for some cases that are not
observed in the training data. Relatively speaking,
a lexicalized HMM-based tagger will be faced
with a more serious problem of data sparseness
than a standard HMM-based tagger. In order to
tackle the problem of zero probabilities in MLE,
we employ the linear interpolation smoothing
technique to smooth the lexicalized parameters
using the relevant non-lexicalized probabilities.
This process can be formulated with Equation
(5):

classical Viterbi algorithm to perform this task,

which works in three main phases as follows:
1.

2.

(5)

where, and refer to the smoothing coefficients.

THE ALGORITHM
This section describes an algorithm for Chinese
POS tagging. The problem of inconsistent tagging
also is discussed in this section.

Viterbi Tagging
Given a sequence of words, there might be more
than one possible sequence of POS tags. The task
of a tagging algorithm is to find the best one that
has the highest score according to the models in
Equation (3) or (4). In our system, we apply the

3.

Decomposing of unknown words. As discussed previously, lexicon words (i.e., known

words) are taken as the basic tagging units.
In this phase, all unknown words in input
are decomposed to a sequence of known
words using the forward maximum match
(FMM) method (Liang, 1987), and each
decomposed known word then is assigned
a proper pattern tag in terms of its position
in the relevant unknown word. At the same
time, each known word in the input is labeled
a pattern tag ISW.
Generation of candidate POS tags. In
this phase, all possible POS tags first are
generated for each word in the input: For a
known word, its POS candidates are generated by looking up the system lexicon; For
an unknown word, its POS candidates are
created by searching the lexical probability
library in terms of its components and their
word-formation patterns. In accordance with
the format defined in the third section, these
POS candidates are then merged with the
relevant pattern tags created in phase (1). All
these combined candidate tags are stored in
a lattice structure.
Decoding of the best tag sequence. In this
step, the Viterbi algorithm is employed to
score all candidate tags using the language
models given in the fourth section and then
to search the best path through the candidate
lattice that maximizes the score. This path
contains the most probable sequence of POS
tags for the input. It should be noted that the
direct output of our tagger is represented as
a sequence of known words together with
a sequence of hybrid tags. For evaluation
purposes, such a format will be converted
further into the normal tagging format by

1599

Chinese POS Disambiguation and Unknown Word Guessing with Lexicalized HMMs

removing the pattern tags and reconstructing

unknown words using their components.

Inconsistent Tagging
Inconsistent tagging will arise when the components of an unknown word are labelled with
different POS tags. For example, the personal
nameinthesegmentedsentence
(Zhang Xiaohua is working at the University of Hong Kong) might be inconsistently
tagged as/Vg-MOW /nr-EOW. In this case,
the system will be unable to make its decision in
determining the POS category of the full word
. Consequently, how to avoid inconsistent tagging is a fundamental issue that relates to the
practicability of our system.
To achieve a complete consistent tagging, we
develop a rule-based module in our system, which
can prevent inconsistent tags from entering into
the lattice for storing candidates during the generation of potential POS tags for unknown words.
Given an unknown word and its components, our
system first creates a set of POS candidates for
each of its components in terms of its relevant
pattern and its lexical probability. Then, it will
continue to generate the POS candidate set for
the unknown word using one of the following
four rules: (1) if the intersection of all the POS
candidate sets for its components is not null, then
the intersection is taken as its POS candidate set;
or (2) if the intersection of the two candidate sets
for its beginning and ending components is not
null, then the intersection is its POS candidate set;
or (3) if the union of the two sets of POS candidates for its beginning and ending components
is not null, then the union set is taken as its POS
candidate set; and (4) if the three previous rules
do not work, then the top frequent POS tag (i.e.,
noun) is taken as its POS candidates. It should
be noted that once the POS candidate set for the
unknown word is determined, all its components
should share this set as their POS candidates.

1600

Rule (2) stresses the importance of the beginning

component and the ending component, because
according to our intuition, the two components
might be more informative than other components
in determining the POS category of a Chinese
unknown word in most cases.

EXPERIMENTS AND DISCUSSIONS

In order to evaluate the effectiveness of our system,
we conducted a number of experiments using the
Peking University corpus and lexicon. This section
reports the results of these experiments.

The Experimental Data

The Peking University (PKU) corpus (Yu et al.,
2003) contains six months (January to June, 1998)
of Chinese newspaper text from the Peoples Daily.
As shown in Table 1, the PKU corpus has been
segmented manually and contains approximately
7.3 million words. Each word in this corpus also
has been tagged with one of 45 POS tag in Appendix A. In our experiments, the first month of
data (i.e., January 1998) was used for testing, and
the remaining five were for training. It should be
noted that an original POS-tagged corpus should
be converted to the format shown in the second
section before using is for training (i.e., a sequence
of known words together with their hybrid tags).
In addition to the PKU corpus, we employed a
POS dictionary of about 75,000 entries in our
system, which were mainly from the Grammatical Knowledge-Base of Contemporary Chinese
(Yu et al., 1998). In order to make this lexicon
complete, a number of non-Hanzi characters and
GBK Hanzi were also added to it.
Table 1 also gives a survey of Chinese POS
ambiguities and unknown words in the PKU
corpus. It can be observed that over 55% of
known words are ambiguous in POS, and some
7% of words in the PFR corpus are unseen in the

Chinese POS Disambiguation and Unknown Word Guessing with Lexicalized HMMs

dictionary used in our system. This shows that

high-performance disambiguation and unknown
word guessing are very important for a practical
Chinese POS tagging system.
Table 2 shows the top 10 frequent types of
unknown words in the PKU corpus. It can be
observed that noun is the most frequent type of
unknown words. We can see that more than 20%
of unknown words are nouns in the PKU corpus.
As can be seen from Table 2, the top 10 frequent
types account for about 93.80% of unknown words

in the training corpus and 94.20% of unknown

words in the testing corpus.

The Experimental Results

In our evaluation, we conducted two experiments
on the PKU corpus in Table 1:
1.

Since a lexicalized HMM-based approach

can handle richer contextual information
for tagging, particularly the contextual lexi-

Table 1. Distribution of words in the PKU corpus

Corpora

All words

Corpus for training

Corpus for testing
Total

6,164,061
1,120,374
7,284,435

Known words
Total
Amb. known words
5,731,303
3,157,763 (55.10%)
1,042,285
569,859 (54.67%)
6,773,588
3,727,622 (55.03%)

Unknown words
432,758 (7.02%)
78,089 (6.97%)
510,847 (7.01%)

Table 2. Top 10 frequent types of unknown words in the PKU corpus

Rank
1
2
3
4
5
6
7
8
9
10
Total

POS
n
nr
m
t
ns
v
j
nz
l
vn
-

Training corpus
Frequency
91,972
79,149
78,395
56,809
31,907
19,460
15,115
14,399
11,955
6,770
405,931

%
21.15
18.20
18.02
13.06
7.34
4.48
3.48
3.31
2.75
1.55
93.80

POS
n
nr
m
t
ns
v
j
nz
l
r
-

Testing corpus
Frequency
16,071
15,991
13,983
10,614
5,306
3,365
2,496
2,390
2,388
959
73,563

%
20.58
20.48
17.91
13.59
6.79
4.31
3.20
3.06
3.06
1.23
94.20

1601

Chinese POS Disambiguation and Unknown Word Guessing with Lexicalized HMMs

2.

cal information for POS tagging, it should

be more powerful in disambiguation and
unknown word guessing in principle than
the standard HMM-based method. Consequently, our first experiment was conducted
to examine how the use of the lexicalization
technique improves the performance of our
system in POS disambiguation and unknown
word guessing.
Due to the larger number of parameters
of the lexicalized HMMs, they are poorly
estimated with MLE for the same training data. That is to say that the problem of
data sparseness will be more serious for
the uniformly lexicalized HMMs than for
the corresponding standard HMMs. Our
second experiment, therefore, was conducted to investigate how much data were
sufficient to achieve a reliable estimation
for the uniformly lexicalized HMMs. In
this experiment, we increased the size of
the training data each time by one month
of corpus and then evaluated the output of
our system on the test corpus.

In our experiments, we calculated tagging

precision for all words, known words, ambiguous known words, and unknown words. We also
computed tagging speed (words per second) to
evaluate the efficiency of our system in tagging.
All the experiments were run on a PC with a
Pentium1.5GHz and a 512MB RAM. The results
are presented next.

Table 3 shows the results of our first experiment. We can see that the lexicalized HMMs
perform better than the standard HMMs as a
whole. In comparison with the standard HMMbased tagger, the lexicalized HMM-based tagger
can improve the tagging precision respectively by
2.06% for all words, 1.88% for all known words,
3.37% for ambiguous known words, and 4.42%
for unknown words. At the same time, the use
of lexicalized models does not lead to a rapid
increase of tagging time. This indicates that the
lexicalized HMM-based method is able to keep
a relative balance between tagging precision and
tagging efficiency.
Figures 1, 2, and 3 present the curves of the
tagging precision for all words, ambiguous known
words, and unknown words, respectively, vs.
the size of the data for training. We can see that
in comparison with the standard HMM-based
tagger, the tagging precision of the lexicalized
HMM-based tagger is changing in a sharper
ascent curve for all three cases as the size of the
training corpus increases, which indicates that
the uniformly lexicalized HMMs need much
more training data than the corresponding nonlexicalized models in order to achieve a reliable
estimation. In other words, the performance of a
lexicalized HMM-based tagging system is more
sensitive to the size of the training data than that
of a standard HMM-based tagging system. Furthermore, it can be observed from Figure 2 and
Figure 3 that a larger improvement of tagging
precision can be achieved for unknown word

Table 3. Comparison of the standard HMMs and the lexicalized HMMs

Method
Standard HMMs
Lexicalised HMMs

1602

Overall
94.08
96.14

Tagging accuracy (%)

KWs
Amb. KWs
94.52
90.79
96.40
94.16

UWs
88.24
92.66

Tagging speed
(w/s)
3,295
2,661

Chinese POS Disambiguation and Unknown Word Guessing with Lexicalized HMMs

Figure 1. Overall tagging precision vs. the size of training data

Overall tagging precision

98
97
96
95
94
LHMM

93

HMM

92
20

40

60

80

100

Size of data for training

Figue 2. Precision for Amb. KWs vs. the size of training data

Precision for Amb. KWs

95
94
93
LHMM

92

HMM

91
90
89
20

40

60

80

100

Size of training data

1603

Chinese POS Disambiguation and Unknown Word Guessing with Lexicalized HMMs

Figure 3. Precision for UWs vs. the size of training data

94

Precision for UWs

92
90
88
86
LHMM

84

HMM

82
20

40

60
Size of data for training

guessing than for the disambiguation of known

words. To our knowledge, the reason might be
that unknown words in real text form an open set,
while known words are a close set. In general, more
training data will provide the system with much
richer contextual information and word-formation cues for the guessing of different unknown
words in an open-ended text. As a consequence,
an increase of training data will result in a steady
improvement of tagging precision in unknown
word guessing.

CONCLUSION
We presented a unified resolution to Chinese POS
disambiguation and unknown word guessing. In
order to explore word-internal cues for Chinese
POS tagging, we introduced four types of wordformation patterns that indicate whether a lexicon

1604

80

100

word is an independent word or a component word

in a specific position of an unknown word in a
sentence. We further defined a hybrid tag set by
merging these patterns with POS tags, with which
we can reformulate Chinese POS disambiguation
and unknown word guessing as one single process of assigning each known word by inputting
a proper hybrid tag. To do this, we developed a
statistical tagger based on the uniformly lexicalized HMMs. In this way, three types of features
word-internal word-formation cues, contextual
tags, and contextual words can be incorporated
to resolve different ambiguous words or unknown
words in Chinese text. We also evaluated our
system using the Peking University corpus and
dictionary. The experiments have shown that our
system can yield correct results for most ambiguous words and unknown words in question. Our
experimental results also indicated that the use of
the lexicalization technique is able to improve the

Chinese POS Disambiguation and Unknown Word Guessing with Lexicalized HMMs

tagging precision without losing much efficiency

in tagging.

ACKNOWLEDGMENTS
We would like to thank the Institute of Computational Linguistics, the Peking University for their
corpus and lexicon. We also would like to thank
the two anonymous reviewers for their helpful
and valuable comments.

REFERENCES
Brants, T. (2000). TnT A statistical part-ofspeech tagger. In Proceedings of the 6th Applied
NLP Conference (ANLP-2000) (pp. 224-231).
Brill, E. (1995). Transformation-based errordriven learning and natural language processing:
A case study in part-of-speech tagging. Computational Linguistics, 21(4), 543-565.
Fu, G., & Luke, K.-K. (2003). Integrated approaches to prosodic word prediction for Chinese TTS.
In Proceedings of the 2003 IEEE International
Conference on Natural Language Processing
and Knowledge Engineering (NLP-KE03) (pp.
413-418).
Fu, G., & Luke, K.-K. (2005). Chinese unknown
word identification using class-based LM. In
Lecture Notes in Computer Science Vol. 3248,
(IJCNLP 2004) (pp. 262-269).
Fu, G., & Wang, X. (2002). A hybrid approach
to Chinese POS tagging: Integrating HMM with
log-linear model. Journal of Chinese Language
and Computing, 12(2), 109-126.
Lee, S.-Z., Tsujii, J., & Rim, H.-C. (2000). Lexicalised hidden Markov models for part-of-speech
tagging. In Proceedings of the 18th Conference

on Computational Linguistics (COLING 2000)

(pp. 481-487).
Liang, N. (1987). CDWS A written Chinese
automatic word segmentation system. Journal of
Chinese Information Processing, 1(2), 44-52.
Nakagawa, T., Kudo, T., & Matsumoto, Y. (2001).
Unknown word guessing and part-of-speech
tagging using support vector machines. In Proceedings of the 6th Natural Language Processing Pacific Rim Symposium (NLPRS 2001) (pp.
325-331).
Nakagawa, T., Kudo, T., & Matsumoto, Y. (2002).
Revision learning and its application to part-ofspeech tagging. In Proceedings of the 40th Annual
Meeting of the Association for Computational
Linguistics (ACL 2002) (pp. 497-504).
Pla, F., & Molina, A. (2004). Improving part-ofspeech tagging using lexicalised HMMs. Natural
Language Engineering, 10(2), 167-189.
Ratnaparkhi, A. (1996). A maximum entropy
model for part-of-speech tagging. In Proceedings of Conference on Empirical Methods in
Natural Language Processing (EMNLP-1) (pp.
133-142).
Weischedel, R., Meteer, M., Schwartz, R., Ramshaw, L., & Palmucci, J. (1993). Coping with
ambiguity and unknown words through probabilistic methods. Computational Linguistics, 19(2),
359-382.
Yu, S., Duan, H., Zhu, X., Swen, B., & Chang,
B. (2003). Specification for corpus processing at
Peking University: Word segmentation, POS tagging and phonetic notation. Journal of Chinese
Language and Computing, 13(2), 121-158.
Yu, S., Zhu, X., Wang, H., & Zhang Y. (1998).
The grammatical knowledge-base of contemporary Chinese: A complete specification. Beijing:
Tsinghua University Press.

1605

Chinese POS Disambiguation and Unknown Word Guessing with Lexicalized HMMs

Endnotes

1606

Unlike other languages like English, Chinese text has no explicit delimiters to mark
word boundaries except for some punctuation marks. Therefore, word segmentation
is usually the first step in performing POS
tagging on a text in Chinese. An introduction
of Chinese word segmentation can be seen

in Fu and Luke (2005). This article focuses

on Chinese POS tagging.
It should be noted that a normal input word
sequence might contain known words
and unknown words. In this article, each
unknown word will be decomposed into a
sequence of known words using the efficient
forward maximum match technique (Liang,
1987) before POS tagging.

Chinese POS Disambiguation and Unknown Word Guessing with Lexicalized HMMs

APPENDIX A.
The PKU Part-of-Speech Tagset
Tag

Chinese

Definition
English

Tag

adjective

Rg

ad

a
dverb-adjective

Tg

an

a
dnoun

Ug

b
c
d
e
f
g
Ag

distinguish word
Vg

conjunction
Yg

a
dverb
h

exclamation
i

position word
j

morpheme
k

adjective morpheme l
distinguish
m

morpheme
adverb morpheme n

numeral morpheme nr

noun morpheme
ns

quantifier
nt

morpheme

Bg
Dg
Mg
Ng
Qg

Definition
English
p
ronoun

morpheme
time morpheme

a
uxiliary

morpheme
verb morpheme

modal morpheme
heading element

i
diom

a
bbreviation

tail element

habitual word

Tag

Chinese

Chinese

Definition
English

nz

other proper noun

onomatopoeia

preposition

q
r
s
t
u
v
vd

quantifier
pronoun
location word
time
auxiliary
verb
adverb-verb

n
umeral

vn

gerund

n
oun
persons name
t
oponym

w
x
y

punctuation
foreign character
modal word

o
rganization noun z

state word

This work was previously published in the International Journal of Technology and Human Interaction, edited by B. C. Stahl,
Volume 2, Issue 1, pp. 39-50, copyright 2006 by IGI Publishing, formerly known as Idea Group Publishing (an imprint of IGI
Global).

1607

1608

Chapter 4.13

IT Implementation in a
Developing Country
Municipality:
A Sociocognitive Analysis
Clive Sanford
Aalborg University, Denmark
Anol Bhattacherjee
University of South Florida, USA

ABSTRACT
This article presents an interpretive analysis of
the key problems and challenges to technology
implementation in developing countries, based
on a three-year case analysis of an IT project in
a city government in Ukraine. We employ the
concept of technological frames of reference as
an analytical tool for articulating the group-level
structures related to the implementation context
from the perspectives of key stakeholders and
examine the degree of conflict between these
frames using a Fishbone diagram. We report
that conflict between technological frames held
by key stakeholders in large-scale system implementation projects often create an unexpected,
dysfunctional, and politically charged implementation environment, ultimately leading to project

failures, even if the project enjoys a high level of

financial and management support. This, in turn,
creates unique challenges for technology implementation projects in developing countries that
are often overlooked in the traditional academic
and practitioner literatures based on experiences
from developed countries.

Introduction
Information technology (IT) has long been viewed
by central planners in the developing world as an
important tool for achieving rapid economic and
wage growth, improving operational efficiency
and effectiveness, and enhancing political participation and transparency. However, achievement of these objectives is often thwarted due

Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

IT Implementation in a Developing Country Municipality

to incentive structures that are based on existing

rules and organizational structures. Improvements in administrative capacity can only be
expected when there is a sound institutional base
that is supported by operational, technical, and
infrastructural facilities. Therefore, planning for
the diffusion of IT into a developing countrys
administrative strategies presents a challenge that
is significantly different from that encountered by
developed countries.
The UN Secretary-General has stated, information technologies can give developing countries
the chance to leapfrog some of the long and painful
stages of development that other countries had
to go through (Annan, 2002). In other words,
IT has the potential to support the development
strategy of bypassing some of the processes
of the accumulation of human capabilities and
fixed investment in order to narrow the gaps in
productivity and output that separate industrialized and developing countries (Steinmueller,
2001). Recently, there is also the indication that
the creation of an information society and, in
particular, e-government implementation would
lead to better (or good) governance (Ciborra & Navarra, 2005). However, the public and businesses
in developing countries often have a mistrust of
governments that are entrenched in hierarchical
structures that are often politicized and corrupt,
and are rife with cumbersome processes that stunt
economic growth and discourage investment in
productive business activities. Efforts to restructure government are either in direct conflict with
institutional archetypes or have proven difficult
for countries with unstable governments to adopt.
In addition, government agencies in developing
and transitional countries often try to install
IT infrastructures that have been designed for
private sector firms and/or for governments in
other countries.
Globalization, democratization, and economic
liberalization have prompted initiatives with IT as
the primary lever to stimulate dramatic changes in
the role of the state. New freedoms and opportuni-

ties, especially prevalent in developing countries,

have raised the expectations of individual citizens,
and emerging and increasingly vocal and proactive
civil societies now do not hesitate to call governments to account. This is somewhat mitigated by
the fact that IT in many public sectors has not
delivered the value expected (Bellamy & Taylor,
1994; Lenk, 1990; Margetts, 1999; Margetts &
Willcocks, 1993; Willcocks, 1994). The literature
on IT in developing countries generally reports
on planned systems, pilot studies, and failures,
and seldom discusses the outcome of projects
that are even partially functional (Cecchini &
Raina, 2005). Further, even when systems are
operational, sustaining operations in an ongoing
manner is difficult (Frasheri, 2003).
This article presents a longitudinal case
analysis of an IT initiative in the city government
of Lviv, a municipality in western Ukraine, to
improve governance and service quality for its
citizens. The specific system under investigation
is a Lotus Notes based document management
system (DMS), intended to automate and streamline citizen services such as building construction
permits, business license applications, and zoning
clarifications. A similar project was rolled out in
the Malaysian government Generic Office Environment (GOE) (Karim, 2003). Grant and Chau
(2005) used the Malaysian GOE as a case example
of service delivery in their strategic focus areas of
their e-government framework. While a governmental DMS has an objective that is congruent
with e-government to enable effective constituent
relationship management (Grant & Chau, 2005,
p. 9), and in general may qualify as a specific
example of e-government, we prefer to orient
the discussion of a DMS as an effort to improve
e-administration. E-administrative initiatives
deal particularly with projects that involve the reengineering of internal structures and organizational activities. These initiatives include cutting
process costs, managing process performance,
making strategic connections in government, and
creating empowerment by transferring power,

1609

IT Implementation in a Developing Country Municipality

authority and resources for processes from their

existing locus to new locations (Heeks, 2002b,
p. 99). This requires a fundamental restructuring
of the public administrations organization and
knowledge management practices by integrating
work processes across departments and agencies
in order to build a system that simplifies interactions and improves services.
We describe three years of the deployment history of the DMS project (2002-2005), by drawing
on the experiences of one of the studys authors,
who was a participant observer in this project
at this time. The author was commissioned by
the City Mayor as a consultant to provide user
training for the DMS and related systems. Based
on numerous personal interactions with people
involved at all levels of this project, such as the
Mayors office, end users, the citys IT department,
and external consultants, and direct involvement
with the DMS and prior Lviv city projects, this
was a unique opportunity to unravel the complex
dynamics of the DMS project as it evolved over
time. Our interpretive analysis of interview, observational, and documentary evidence revealed
several interesting findings and implications about
the politics of IT implementation projects in developing countries. Given the unprecedented nature
of political, sociocultural and economic change
witnessed in Ukraine and other new democracies of eastern Europe over the last decade, this
context provided us an ideal testbed for studying
the processes and impacts of e-administrative
government projects in countries that are seeking
to establish western-style democratic reforms.
The structure of the article proceeds as follows.
The next section discusses the rationales behind
the introduction of IT in developing countries. The
third section describes a sociocognitive view of
IT implementation, which serves as the theoretical backdrop for our research. The fourth section
presents a background of the context of this case
study, followed by a historical description of the
events in the case. The fifth section describes our
research methods, including data collection and

1610

analysis approaches. The sixth section presents our

research findings, based on interpretive analysis
of the case data. The seventh section provides a
discussion of the key issues in the case and the final
section proposes directions for future research
and presents concluding remarks.

The Diffusion of IT to
Developing Countries
Information technology and associated organizational models, which form the basis of information systems and implementation methodologies,
have originated and continue to be developed
principally in western, developed countries. The
unidirectional transfer of IT knowledge and skills
from developed to developing countries is a process that characterizes most recent efforts at public
sector reform (Minogue, 2001). However, in order
to take advantage of technological knowledge of
developed countries, developing countries must
have acquired sufficient technological capabilities
and institutional capacities to identify suitable
technologies and to adapt, absorb, and improve
the technologies imported from abroad (Ahrens,
2002). Often, they are not appropriate for use in
countries that are in transition and have their
own realities on the design of technologies that
are used to automate and inform their internal
and citizen-focused processes. Fountain (2001)
addressed this issue in her differentiation between
the deployment of already invented technologies
that are available to practitioners and the need
to recognize the importance of customizing the
design to a specific locale.
A rationale that is often used for the introduction of IT is to automate, increase efficiencies, and
cut costs. The cost of labor in developed countries
is substantially higher than that in developing
countries (Dewan & Kraemer, 2000), while the
reverse is true with the cost of the IT (Heeks,
2002b). Therefore, replacing inexpensive civil
servants with more expensive IT would not be jus-

IT Implementation in a Developing Country Municipality

tifiable for financial reasons alone. Consequently,

the choice of projects and their ultimate success
is dependent on careful consideration of the embedded web of relations and interactions within a
particular socioeconomic context, and their design
and implementation requires an understanding of
this context (Sein & Harindranath, 2004).
Governmental and international development
agencies (IDAs) in the developed world, such as the
World Bank, the United Nations, and United States
Agency for International Development (USAID),
have spearheaded and provided resources for
deploying governmental IT technologies in the
developing world. Since nearly half the people
of the world today are under 25 years old and
90% of these young people live in developing
countries (World Bank, 2005), IDAs are seeking
to optimize the deployment of their resources
to parts of the world where they would have the
potential to have the most enduring and significant
impact. Their initiatives fit well with the western
ideals of globalization, economic liberation, and
democratization, with improved governmental
functioning in developing countries helping to
secure important sources for raw materials or
low-cost labor for western businesses and opening up new or growing existing markets in these
countries for western goods and services.
IDAs and others believe that the scarcity of
IT in developing countries is one of the reasons
for decreased opportunities for economic growth
and social development in developing countries
(Madon, 2004). However, such initiatives are
challenged by the lack of integrated information
and communication technology infrastructures,
antiquated work practices, and perhaps most
importantly, organizational obstacles such as
conflicting objectives, lack of widespread support, and internal resistance to change within
governmental agencies. Madon (2004) explored
an alternative perspective of development where
conditions are created to encourage individuals
to realize their full potential by freely selecting

their path to improvements in material, social,

and spiritual lifestyles (Sen, 1999). The World
Bank (2005) has encouraged such investments
that empower people to participate in decisions
about development projects that affect their lives
and the lives of their families and have an enduring impact on their quality of life.

A Sociocognitive Perspective of
Implementation
The sociocognitive perspective is based on the
idea that reality is socially constructed through
human beings interpretation of social reality
and negotiation of meaning among social entities
(Berger & Luckmann, 1967). Like the cognitive
perspective, the sociocognitive perspective emphasizes the role of ones personal knowledge
structures in shaping her cognitive interpretation
of the world and consequent actions. However, it
differs from the cognitive perspective in the way
it views the formation of individual knowledge
structures. The cognitive perspective postulates
information processing and learning as the key
processes shaping individual knowledge structures, while sociocognitive research holds that
shared knowledge and beliefs (group-level structures), imbibed through the processes of interaction, socialization, and negotiation within a larger
social group, influence individual sense-making
and interpretation. These shared knowledge
structures, also called frames (Goffman, 1974)
or interpretive schemes (Giddens, 1984), serve
as templates or filters for individual sense-making
and problem solving, filling information gaps with
information that conforms with existing knowledge structures, and systematically rejecting
inconsistent information (Fiske & Taylor, 1984).
Frames also provide a vehicle for organizing and
interpreting complex and sometimes ambiguous
social phenomena, reducing uncertainty under
conditions of complexity and change, and justifying social actions.

1611

IT Implementation in a Developing Country Municipality

The sociocognitive perspective has seen some

applications in prior information systems research.
In the context of information systems development, at least three studies (Bansler & Bodkar,
1993; Davidson, 2002; Newman & Nobel, 1990)
have posited that requirements for IT applications
do not exist a priori but are socially constructed
through interactions among system developers,
user groups, and other key stakeholders. The
language used by participants and the nature of
their interaction influences which requirements
are identified and legitimized (Boland, 1978).
Drawing from the above research, Orlikowski
and Gash (1994) proposed the concept of technological frames of reference as an analytical tool
for understanding and articulating group-level
structures that guide sense-making and action.
Technological frames are defined as assumptions, knowledge, and expectations, expressed
symbolically through language, visual images,
metaphors, a