Information Visualization Technologies

Definition:
• Information Visualization involves presenting complex data in graphical
formats to enhance understanding and decision-making.
• The theories associated with information visualization are being applied
in commercial and experimental systems to determine the best way to
improve the user interface, facilitating the localization of information.
They have been applied to many different situations and environments
Importance:
• Helps in discovering insights, patterns, and correlations in data.
• Critical in environments such as weather forecasting, architectural
design, and more.
The ones focused on Information Retrieval Systems are investigating how
best to display the results of searches, structured data from DBMSs and
 Goals of IRS linked to Information
Visualization Techniques
Key Areas of Focus of information visualization techniques:
• Displaying search results and structured data from databases.
• Correlating data through link analysis.
The goals for displaying the result from searches fall into two major classes:
• Document Clustering:The goal of document clustering is to present the user
with a visual representation of the document space constrained by the search
criteria.
• Ex:Visualization tools in this area attempt to display the clusters, with an
indication of their size and topic, as a basis for users to navigate to items of
interest. This is equivalent to searching the index at a library and then
pursuing all the books on the different shelf locations that are retrieved by the
search. The second goal is to assist the user in understanding why items were
retrieved, thereby providing information needed to refine the query.
• Search Statement Analysis:Modern search algorithms and ranking techniques
make it difficult to understand query impacts. Search Statement Analysis
 Structured Databases and Link Analysis
Structured Databases:
• Organizes citation and semantic data that describe
documents.
Link Analysis:
• Correlates documents by analyzing relationships between
events and timelines.
• For example, a time/event link analysis correlates multiple
documents discussing a oil spill caused by a tanker. Even if all
of the items retrieved on the topic are relevant, displaying
the documents correlated by time may show dependencies
of events that are of information importance and are not
 Different Ways Of Organizing the Information
Hierarchical Information Organization
• Tree structures are used to represent information over time or across larger
units.
• A two-dimensional representation becomes difficult for a user to
understand as the
hierarchy becomes large.
So we use Cone-tree and Tree
Maps techniques.
• Cone-Tree: 3D representation
of hierarchical data.The squares
at the leaf nodes in tree are the
actual documents. Higher level
nodes can be considered centroids
representing the semantic of the
 Tree Maps
• Use screen space to display parent-child data structures efficiently.This
technique maximizes screen space and shows topic relationships by
using rectangular boxes. Each box represents parent-child relationships
within the data. The size of the boxes indicates the number of items in
a topic, and their location shows topic relationships.
Ex: A particular information work space focused on articles on computers
may appear as
shown below:
The CPU, OS, Memory, and
Network management articles
are all related to a general
category of computer operating
systems versus computer
applications which are shown
in the rest of the figure.
 Semantic Scatterplots
• When the information has network-type relationships, an approach
using clusters can be shown via a semantic scatterplot.
• A graphical technique to display relationships between data points.
• Vineta and Bead Systems use 3D scatterplots for data clustering.
• Semantic Landscape: Use elevation to indicate the frequency and
importance of concepts.
• The difficulty of representing all of the axis is overcome by projecting
the space onto a plane and using elevation to indicate the frequency
of occurrence and importance of a theme (concept) creating a
semantic landscape.The detailed relationships between items and
their composite themes can be seen by the valleys, cliffs and ranges
shown on the terrain map.
• Different ways of overcoming the multidimensional space
representation in a hierarchical environment is suggested by
Kohonen, linked trees or graphs or a non-Euclidean landscape.
 Visualization for Search Refinement
Challenges:
• Modern search algorithms and ranking systems complicate the
understanding of query results.
• The goal of many visualization techniques is to show the semantic
relationships between individual items to assist the user in locating those
groups of items of interest. Another objective of visualization is in
assisting the users in refining their search statements. It is difficult for
users in systems using similarity measures to determine what are the
primary causes for the selection and ranking of items in a Hit file.
• Visualization tools need to assist the user in understanding the effects of
his search statement even to the level of identifying important terms that
are not contributing to the Envision Interface search process. One solution
is a graphical display of the characteristics of the retrieved items which
contributed to their selection.
• Envision System: Displays index terms, relevance rank, and estimated
relevance of search results interactively. The Envision system not only
Ex: Circles represent single
items with the relevance
weights displayed below them.
Ellipses represent clusters of
multiple items that are located
at the same point in the
scatterplot with the
number of items in the ellipse
and their weights below
the ellipse. In this example,
estimated relevance is on the
X-axis and author’s name is on
the Y-axis. This type of interface
provides a very user friendly
environment but encounters
problems when the number of
relevant items and entries for an
axis becomes very large. Envision
plans to address this issue by a
“zoom” feature that will allow
seeing larger areas of the
scatterplot at lesser detail.
• Veerasamy and Belkin’s System: They use a series of vertical columns of bars. The columns of bars
represent documents, and the rows represent index terms. The height of the bar corresponds to the
weight of the corresponding term (row) in the corresponding item (column). In addition to the query
terms, the system shows the additional words added to the system by relevance feedback.
 Document Content Analysis and Retrieval
DCARS:
System Representations
• Displays query results
as histograms showing
term contributions.
• Enhances understanding
of how search statements
influence retrieval.
• They display the query
results as a histogram with
the items as rows and each
term’s contribution to the
selection indicated by the width of a tile bar on the row
• DCARS provides a friendly user interface that indicates why a particular
item was found, but it is much harder to use the information in
 Cityscape
• A metaphor that uses buildings
to represent themes or concepts
in 3D space.
• This is similar to extending bar
charts to three dimensions.Buildings
can be connected by lines which can
vary in representation to describe
interrelationships between themes.
• An easily understood metaphor for users is that of a library.
Information content can be represented as areas within a library that
the user can navigate through. Once in a particular “information
room” the user can view the virtual “books” available within that
space as if they are silting on a bookshelf. Once the book is accessed,
the user can scan a group of related items with each item
 VISUALIZING SPECIFIC TEXT
• Another task in information systems is the visualization of
specific text within an item versus between items. In some
situations, items are allowed to change over time via editing.
Thus, there is both the static representation and a time
varying representation. Text changing representations are
very important when the text being represented is a software
program of millions of lines of code.
Ex: AT&T Bell laboratories created the SeeSoft system which
uses columns and color codes to show when different lines of
code have been changed. This technique was used as a basis for
a similar code visualization tool, DEC FUSE/SoftVis.Here a user
can quickly see the relative structure of all of the code files
composing a system along with the complexity of each of the
 Visualizing Structured Databases
Although information retrieval focuses on the unstructured text, another aspect of informational
items is the citation data and structured aspects of indexing items. This data structure can be
manipulated via structured databases as well as traditional information systems.
Query By Example (QBE) and Information Visualization and Exploration Environment (IVEE) use
visual representations of structured databases for efficient search and exploration.
Query By Example (QBE): Developed by IBM, QBE is the first visual query tool that allows users to
create queries by providing examples, making it easier to retrieve data from structured databases.
How it Works:
Users create a query by filling out a template or form with example data they want to
find. The system then translates this example into a database query.
Example: Imagine you have a database of books with the following fields: Title, Author,
and Publication Year.
Scenario: You want to find books written by "J.K. Rowling".
INPUT:
Title: [blank]
Author: "J.K. Rowling"
Publication Year: [blank]
The system retrieves all books authored by J.K. Rowling, showing you relevant entries without
needing to write a SQL query like SELECT * FROM books WHERE Author = 'J.K. Rowling'.
 Visualizing Structured Databases
Information Visualization and Exploration Environment (IVEE)
What is IVEE?

The Information Visualization and Exploration Environment (IVEE) is a tool that uses visual
representations, like 3D graphics and starfield displays, to help users explore structured
databases. It allows users to see data in an interactive way based on their search criteria.
How it Works:
Users input search criteria, and IVEE generates a visual representation of the data, making it
easier to identify patterns, relationships, and relevant information.
Example: Consider a structured database of houses for sale with attributes like Price,
Location, and Number of Bedrooms.

Scenario: You want to find houses in "New York" under $500,000 with at least 3 bedrooms.
IVEE Input: You set your criteria in the IVEE interface.
Result: IVEE displays the houses in a 3D environment, where each house is represented as a
point in space. You can rotate and zoom in/out to explore various options visually. The
houses that meet your criteria are highlighted, making it easy to compare them at a glance.
 Specialized Visualization Tools
Another area in information visualization is the representation of
pattern and linkage analysis. A system that incorporates many
information visualization techniques including those used to
represent linkage analysis is the Pathfinder. It contains the
Document Browser, Counts, CrossField Matrix, OILSTOCK and
SPIRE tools.
Document Correlation:
• SPIRE Tool: Uses scattergraphs to cluster items based on term
concurrence.
• Pathfinder Project: Combines several visualization techniques
for linkage analysis.
• OILSTOCKPlaces data on geographic maps, illustrating how
information relates to specific locations.Text Representation:
 Conclusion
• Information visualization is essential for simplifying
complex data.
• Tools like Cone-Trees, Tree Maps, and scatterplots
are vital for clustering and understanding search
results.
• Specialized systems like DCARS, SPIRE, and Envision
provide enhanced user interaction with data.
Future Directions:
• Continuous innovation in visualization technologies
is needed to deal with increasing data complexity
and volume.