An Introduction to Entity Relationship

Diagrams (ERDs)

Written by: Robin Beaumont e-mail: [email protected]

Date last updated: 09/04/2007 14:23

Version: 5
 Health Informatics
Entity Relationship Diagrams (ERDs)

How th i s d oc um en t sh ou l d be u sed:
This document has been designed to be suitable for web based and face-to-face teaching. The text has been made
to be as interactive as possible with exercises, Multiple Choice Questions (MCQs) and web based exercises.
If you are using this document as part of a web-based course you are urged to use the online discussion board to
discuss the issues raised in this document and share your solutions with other students.

W ho thi s do cum e n t i s aim e d a t:

This document is aimed for two types of people:
 Those who wish to become involved in database development but are not interested in the nuts
and bolts of programming, such people are commonly called domain experts and act a bridges
between a professional group (e.g. medics, Solicitors etc) to which they belong and IT experts.
 As an introduction for those just beginning professional computer science courses

I hope you enjoy working through this document.

Robin Beaumont

Robin Beaumont [email protected] 17/07/2010 Page 2 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

Contents
1. Before you start .......................................................................................................................................... 5
1.1 Prerequisites ............................................................................................................................................. 5
1.2 Required Resources.................................................................................................................................. 5
2. Learning Outcomes .................................................................................................................................... 6
3. Introduction ................................................................................................................................................ 7
3.1 Why Learn About ERDs? .......................................................................................................................... 7
3.2 What is an ERD?....................................................................................................................................... 8
4. Entity types, instances and attributes ....................................................................................................... 9
4.1 Showing Attributes in ERDs..................................................................................................................... 13
4.2 Summary ................................................................................................................................................ 13
4.3 Identifying Entity Types ........................................................................................................................... 15
4.3.1 Analysing a Narrative Description of Requirements .............................................................................................. 15
4.3.2 Identifying Appropriate Entity Types ...................................................................................................................... 17
4.3.3 Good and Bad Entity Types ................................................................................................................................... 18
4.3.4 Warning about the „is a Kind of‟ Situation .............................................................................................................. 19
4.3.5 Workshops............................................................................................................................................................. 19
4.3.6 Drawing ERDs ....................................................................................................................................................... 19
4.4 Summary ................................................................................................................................................ 20
5. Relationships? .......................................................................................................................................... 22
5.1 What Does it Mean?................................................................................................................................ 22
5.2 Foreign Keys - How it all Works ............................................................................................................... 23
5.3 What Name do you give a Foreign Key Attribute? .................................................................................... 23
5.4 What are Dependency and Referential Integrity? ..................................................................................... 23
5.5 What is a One to One Relationship? ........................................................................................................ 24
5.6 What is a One to Many Relationship? ...................................................................................................... 25
5.7 Optionality ............................................................................................................................................... 25
5.8 How Many is Many? ................................................................................................................................ 26
5.9 What is a Many to Many Relationship? .................................................................................................... 26
6. Some Common Mistakes in ERDs ........................................................................................................... 28
6.1 Confusing Instances and Types............................................................................................................... 28
6.2 Including Lookup/Descriptor Entities ........................................................................................................ 29
6.3 Unnecessary Complexity ......................................................................................................................... 30
6.4 Poor Aesthetics ....................................................................................................................................... 31
7. The Relationship between Narrative Descriptions and ERDs ................................................................ 32
8. Recursion in ERDs ................................................................................................................................... 38
9. Conceptual, Logical and Physical Data Models ...................................................................................... 39
9.1 Electronic Message Board Example ........................................................................................................ 40
10. Where do ERDs fit Into the Database Design Process? ......................................................................... 41
10.1 The Small Personal Database.............................................................................................................. 41
10.2 Departmental and Hospital Systems .................................................................................................... 42

Robin Beaumont [email protected] 17/07/2010 Page 3 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)
11. Publicly Available Data Models (ERDs) ................................................................................................... 43
12. CASE Tools ............................................................................................................................................... 44
12.1 What Advantage(s) do CASE Tools Offer Over Simple Diagramming Tools?........................................ 44
13. Exercises .................................................................................................................................................. 47
14. MCQs......................................................................................................................................................... 48
15. Summary ................................................................................................................................................... 52
16. References ................................................................................................................................................ 52
17. Links.......................................................................................................................................................... 52
18. Appendix A Relationship Terminology .................................................................................................... 53

Robin Beaumont [email protected] 17/07/2010 Page 4 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

1. Before you start

1.1 Prerequisites
This document assumes that you have the following knowledge and skills:
Skills:
That you have used the following features of a Database Management System (DBMS) such as Access to:
 Create Tables
 Create Relationships (and therefore know about the relationship window)
 Create simple Queries in the query design window
 Create a simple form

Knowledge:
You should also be able to describe what the following concepts mean:
 Tables, indexes and Fields
 Relationships (not a detailed description)
 Forms
 Queries

If you have completed the ECDL (European Computer Driving Licence) you will have covered these topics. If not I
recommend that you do so now. You can take the ECDL either at a local college (in the UK) or as a distance
Learning course. There are also some very good books guiding you through the ECDL.

1.2 Required Resources

This document provides you with the knowledge and skills to be able to develop and draw ERDs using a pen and
paper it does not describe how to create and maintain ERDs using a specialist tool such as a case tool (DeZign,
MagicDraw, System Architect etc.). Please see http://www.robin-beaumont.co.uk/virtualclassroom/contents.htm
section 11 for various CASE tool tutorials.
I do recommend that you either complete either the DeZign or MagicDraw tutorial while, you work through this
document.

Robin Beaumont [email protected] 17/07/2010 Page 5 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

2. Learning Outcomes
This document aims to provide you with the following skills and information. After you have completed it you should
come back to these points, ticking off those you feel happy with.

Learning outcome Tick box

Be aware of the history of ERD diagraming

Be able to list the reasons why the ability to develop ERDs is important for database developers and those involved in database
development

Be able to provide a definition and example of an entity type

Be able to provide a definition and example of an entity (instance)

Be able to provide a definition and example of an attribute

Be able to explain the relationship between entity type, entity instance and attribute

Be able to discuss the importance of context in identifying entity types

Be able to produce a list of candidate entity types from a narrative description

Be able to discuss the various criteria you may use to help refine an initial list of entity types

Be able to suggest criteria that entity types should conform to

Be aware of the use of workshops and informal ERDs to develop ERDs in a group setting

Be able to provide a definition of relationship within the ERD context

Be able to describe the Parent/Child concept

Be able to provide an example of how the relationship concept is implemented within a relational DBMS (e.g. Access)

Be able to explain the existence (mandatory) dependency constraint

Be able to explain the optional dependency constraint

Be able to describe what referential integrity is

Be able to provide examples and explain what a one to one relationship is

Be able to provide examples and explain what a one to many relationship is

Be able to provide examples and explain what a many to many relationship is

Be able to discuss the main issues in the confusion concerning conceptual, logical and physical data models

Be able to evaluate the quality of an ERD from the perspective of undue complexity and aesthetics

Be able to describe the pragmatic approach which is usually adopted when developing successive ERDs

Be able to describe the additional functions a CASE tool offers over drawing programs

Be able to provide examples of various models that already exist and are available to the public for a number of situations

Be able to describe recursion in ERDs

Be able to describe how recursion is often removed from ERDs as they are refined

Be able to indicate that the process of normalising to fifth business normal form removes recursive relationships

Be able to contrast the main differences between small scale and large scale database projects

Be able to describe a process for developing a small database

Be able to demonstrate the central part ERDs play in all database development methods

Robin Beaumont [email protected] 17/07/2010 Page 6 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

3. Introduction
This document will introduce you to a technique used to provide a diagrammatic description of certain aspects of the
data requirements for a database. The technique is called entity relationship modelling and uses Entity Relationship
Diagrams (ERDs) as the main method. It is a particular kind of data modelling and is one of the oldest techniques
around having been developed in the 1970's
by Peter Chen who is very much still alive and
continues to produce important research
(http://www.csc.lsu.edu/~chen/chen.html).
You can read his original paper at
http://www.csc.lsu.edu/~chen/pdf/erd.pdf. One
of my online students was motivated enough
to go and see him when he was working
through a former version of this document (he
did live close by).
You may have come across such diagrams
already using a variety of databases including
Microsoft Access. The example below shows
an example from a database to collect
research data from those who suffer from
diabetes and are also pregnant (don‟t worry
about understanding the diagram at this
stage):

3.1 Why Learn About ERDs?

This document is very much about getting you to learn what ERDs are and how to produce them. You may ask why,
so here are a few reasons I believe it is important for you to do so:
 These diagrams form the basis of most database design methods. If you ever are involved in database
design (i.e. data modelling) you will therefore need to understand them because if they are wrong (i.e. the
blueprint), the database that is built from them will also be wrong!
 These diagrams form the basis for several more trendy techniques that you will come across from systems
developers and learn about in subsequent documents, most notably UML.

Robin Beaumont [email protected] 17/07/2010 Page 7 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)
 These diagrams provide a method of analysing a situation that forces you to adopt a stance of a typical
database modeller. By using them you begin to realise how their minds work and also begin to appreciate
why some databases are so problematic.
These are only a few of the reasons that I personally believe this skill is so important.
Even if you do not intend to become a programmer or systems analyst you may want to become the type of person
who is able to provide a bridge between your professional group – such as doctor, vet or solicitor – and those who
develop or manage information systems. Such people are vitally important in developing usable information
systems and there is a great shortage of them. Such a person is often referred to as a „Domain Expert‟.
Before you start to learn what ERDs are and how to create them yourself there are some Multiple Choice Questions
(MCQs) on the next page to see how much you have taken in so far.

Exercise 1. MCQs
1. From the list below choose two reasons why it is important for a „domain expert‟ such as you to learn about the
ERD method:
a. Provides insight into the mindset of database developers
b. Is the only method available to describe the data requirements
c. Provides credibility to IT personnel
d. Forms the basis of most data modelling techniques
e. Has been proved to be the most cost effective method of specifying data requirements

2. From the list below choose the one option that describes the most desirable „domain expert‟ from the medical
profession:
a. Someone who has developed several databases but knows little of database modelling or current issues in
medicine
b. Someone who has little interest in how information may help the department
c. Someone who has problems working in a collaborative environment
d. Someone who has previously managed IT projects
e. Someone who has knowledge of data modelling techniques and currently works in the appropriate situation

3.2 What is an ERD?

Definition:
An ERD is a graphical description of the data for a particular database = a graphical data model. It represents the
data at a high level of abstraction.
Some more clarification:
The term „database‟ above usually implies a computer-based
database; however, given the complexity of the ERD it might
well be realised („implemented‟) in the form of some type of
paper based clerical system.
„High level of abstraction‟ means that it is not necessary to
show details of the various fields or indexes, just the bare
bones.
Opposite is a very simple ERD that shows three entity types
and two relationships. On the following pages we will
consider each of these elements separately.

Robin Beaumont [email protected] 17/07/2010 Page 8 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

Exercise 2. MCQs
1. ERDs provide a description of (one correct answer):

a. The processes that occur in the model

b. The various entities and their relationships in the model
c. The entities in the model
d. The processes and data requirements of a model
e. The User Interface aspects of the model

2. ERDs provide a (one correct answer):

a. A detailed description of the data within a data model

b. A detailed description of the proposed uses of a database
c. A guide to the training costs
d. A high level description of the data within a data model
e. A graphical picture that is of little use in developing a database

4. Entity types, instances and attributes

The first question is: what is an entity? However, unfortunately this is not an easy question to answer as the term
„entity‟ has been used to mean two different things. I have tried below to present these two different meanings as
„entity instance‟ and „entity type‟, adapted from Reingruber & Gregory 1994.

Some definitions of Entities and Entity Types

“An entity is a „thing‟ or „concept‟ or „object‟. Well, most of the time”.
An entity [type] is not a single „thing‟ but rather a representation of like or similar things that share characteristics
(properties). For example, King Lear and Hamlet are both plays and have properties such as name, author, cast of
actors and lines of verse. The entity [type] describing these might be PLAY, with King Lear and Hamlet as examples
of instances or occurrences of PLAY [i.e. each is an entity [instance]].
Entities involve information. The „things‟ that entities represent are things about which the enterprise wants or needs
to retain information. Therefore, while a data model should be an accurate representation of the business and its
rules, we should never forget that data modelling is generally a precursor to design and development of structures
intended to collect, store and dispense data.” (p64)
Every entity [instance] in an entity type has the same set of attributes (characteristics).
The following are some definitions provided by various writers also listed in Reingruber & Gregory 1994 p63:
Thomas Bruce (1992): “Any distinguishable person, place, thing, event, or concept, about which information is kept”
Peter Chen (1976): “A thing which can be distinctly identified”
CJ Date (1986): “Any distinguishable object that is to be represented in a database”
James Martin (1989): “...anything about which we store information (e.g. supplier, machine tool, employee, utility
pole, airline seat, etc). For each entity type, certain attributes are stored”

The bottom line is that:

 An entity type is a template or blueprint about which you are interested in collecting information (e.g. man,
desk etc).
 An entity instance) is a single example (=instance) of an entity type (e.g. man=Robin Beaumont; desk=my
desk in the upstairs study etc). People often say this is a “real world example”.
 Both of the above consist of a name and a set of attributes.
The above ideas go back to ancient Greece, but I won‟t bore you with such details now.

Robin Beaumont [email protected] 17/07/2010 Page 9 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)
Attributes are basically characteristics. Traditionally these attributes were easily measurable such as weight, height,
name (text used to be represented as a series of numbers – ASCII codes) etc. However there is no reason why an
attribute could not be something like a passport photo or ECG recording.
A number of examples will hopefully make things a little clearer.

Why is this an entity type?

Because it as a name
Because it has a set of unique attributes i.e. surname,
Doctor = Entity type forename, date of birth, GMC number, salary, gender etc.
Why is this a type rather than an instance?
Dr Dodger smyth (GMC number 39200456) born Because it possesses these attributes but they do not
19/12/1956 = entity instance of doctor have specific values

Why is this an entity instance of Doctor Entity type?

Because it has the same set of attributes as the doctor entity type i.e. surname,
forename, date of birth, GMC number, salary, age etc.
And each attribute has a specific value

Another way of thinking about the bond between an entity type and instance is to think of the type as being the
column headings (= attribute names) for a number of rows each of which is an instance of the entity type.

Doctor Doctor entity type with 3

surname forename date of birth GMC number salary gender
entity instances of
Doctor
smyth Dodger 19/12/1956 39200456 89K male
Coates Jill 03/05/1966 5748337 67K female
Worsley Alan 11/07/1970 578493 80K male

Doctor
surname forename date of birth GMC number salary gender Doctor entity type with
no entity instances of
Doctor

Doctor
surname forename date of birth GMC number salary gender Doctor entity type with 4
entity instances of
smyth Dodger 19/12/1956 39200456 89K male
Doctor
Coates Jill 03/05/1966 5748337 67K female
Worsley Alan 11/07/1970 578493 80K Male
Brown John 29/10/1055 4958576 45K Male

Coates Jill 03/05/1966 5748337 67K female

Worsley Alan 11/07/1970 578493 80K Male 3 entity instances of

Brown John 29/10/1055 4958576 45K Male Doctor

Now can you see why an entity type is like a template for entity instances.

Robin Beaumont [email protected] 17/07/2010 Page 10 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)
Below are some examples of entity types and instances for three different contexts:

A typical general practitioner planning on collecting information about patients:

Entity type Entity instance example(s) providing values for some of the attributes

Doctor Angus Wallace living in the UK

Patient Joe Bloggs in Newcastle , Alan Smith London

Visit 12/06/2001 at 4pm Prospect House Newcastle

Prescription Valium 5mg eight times a day

Advice leaflet Chronic back pain leaflet given to Mrs Smith last friday

The university administrator thinking about developing a course database:

Entity type Entity instance example(s) providing values for some of the attributes

Module Coordinator Robin Beaumont, Joe Brand

Student Paul Whatling

Tutor Ruth Brown

Module Introduction to Informatics to run in 2003

A garage collecting detailed information about car components:

Entity type Entity instance example(s) providing values for some of the attributes

Spark Plug Rover, model type 49532

Engine 122 Brake Horsepower

Cooling system 17 pints capacity

Gearbox Second gear ratio 7.55:1

Propeller shaft Hardy Spicer type

Are the above examples correct?

While the above examples probably list a selection of the important entity types for each of the contexts, there is no
absolute way of being able to say they are correct. Rumbaugh et al (1991, p21) makes this important point when
discussing object identification; which for the moment you can assume is the same as an entity instance:
“We define an object as a concept, abstraction, or thing with crisp boundaries and meaning for the problem at hand.
Objects serve two purposes: they promote understanding of the real world and provide a practical basis for
computer implementation. Decomposition of a problem into objects depends on judgment and the nature of the
problem. There is no one correct presentation.”
This is clear from considering the last example concerning car component details. Assume that another garage
owner who was less concerned with collecting detailed information for each component only wanted one item of
information recorded for each component. In this instance we would probably have CAR as the entity type and each
of the entity types listed above relegated to attributes of the CAR entity type. There is no definitive correct answer;
we only know by talking to the actual garage owner. Entities are largely defined by the context.

Robin Beaumont [email protected] 17/07/2010 Page 11 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

Exercise 3. Entity Types and Instances

Spend five minutes filling in the tables below. List some entity types along with examples of their entity
instances. Some of your entity types may have no or several instances. The important thing is that each
instance possesses the same set of attributes as the entity type.

Entity type name:

Attributes names:

Attributes values:

Entity type name:

Attributes names:

Attributes values:

Entity type name:

Attributes names:

Attributes values:

Entity type name:

Attributes names:

Attributes values:

Robin Beaumont [email protected] 17/07/2010 Page 12 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

4.1 Showing Attributes in ERDs

The diagram below shows the ERD displayed in the previous section but with details of several attributes for each of
the entity types. For example, the entity type DOCTOR has four attributes displayed: id, surname, DOB and
gender. Beside each one you will notice a term such as DECIMAL or DATE etc. This is just indicating what type of
data each attribute is.
To work successfully through this document you do not need to know the various data types that each attribute can
take, such as decimal etc. You only need to be aware that you can show an ERD either just with the entity type
names showing or details of the attributes as well; I personally prefer the former layout as it is far less cluttered.

You may have begun to feel that there is some similarity between the concepts we have been discussing above and
various concepts you will have come across when using a database management system such as Access.
In fact there are two very basic similarities:
Table definitions (in a database) = Entity types (in an ERD)
Records in a table (in a database) = Entity instances (in an ERD)

It must be realised that this is only a rough guide as there are numerous exceptions to the rule, but it does help to
think of the concepts in the above manner.

4.2 Summary
 An ERD consists of entity types and relationships.
 An entity type may be bound to zero or more entities instances (you can therefore have an entity type with
no instances).
 An entity type possesses a number of attributes (e.g. a GP possesses name, date of birth, GMC number,
salary, etc).
 Each entity instance possesses the same set of attributes as the associated entity type, where now each
attribute possesses a particular value (e.g. GP = Smith; date of birth = 21/09/78, etc).

We will now look in more depth at the very important task of identifying entity types for a particular context, but
before that here are some MCQs for you to answer.

Robin Beaumont [email protected] 17/07/2010 Page 13 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

Exercise 4. MCQs
1. Which of the following provides the best description of an entity type (select one)?
a. A specific concrete object with a defined set of processes (e.g. John Brown with diabetes)
b. A value given to a particular attribute (e.g. height - 230 cm)
c. A thing that we wish to collect data about where zero or more, possibly
real world examples of it may exist
d. A template for a group of things with the same set of characteristics that may exist in the real world
e. An undefined concept that needs further clarification

2. Which of the following provides the best description of an entity instance (select one)?
a. A specific concrete object with a defined set of processes (e.g. John Brown with diabetes)
b. A value given to a particular attribute (e.g. height - 230 cm)
c. A thing that we wish to collect data about where zero or more, possibly
real world examples of it may exist
d. A template for a group of things with the same set of characteristics that may exist in the real world
e. An undefined concept that needs further clarification

3. From the following list, select two entity instances:

a. Steinway piano model D design template
b. McGill type forceps as used in surgical operations
c. Tony Blair ( British prime minister)
d. PIII type chip that is found in many modern PCs
e. An advice leaflet for chronic back pain issued to Mrs Smith on 02/10/2002

4. Which one of the following statements is true (select one)?

a. An ERD can only display entity type names.
b. Attributes are a rarely considered aspect of entities.
c. Attributes must always be displayed in ERDs.
d. Attributes equate to table names in a database.
e. Attributes can optionally be displayed in ERDs.

5. Select from the following the best list of attributes for the entity SHOE for someone working in a shoe shop (select
one answer):
a. 1, 2, 6
b. 1, 2, 6, 3
c. 1, 2
d. 1, 6
e. 1, 2, 3, 4, 5, 6
1=size
2=number of lace holes
3=propulsion
6. Select from the following the best list of attributes for the 4=cooking time
entity RECIPE for someone following a recipe at home (select 5=number of portions
one): 6=material colour
7=volume (loudness)
a. 4, 5, 7, 8
8=cooking temperature
b. 4, 5
c. 4, 5, 6
d. 4, 5, 8
e. 4, 5, 6, 8

Robin Beaumont [email protected] 17/07/2010 Page 14 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

4.3 Identifying Entity Types

While there is much written concerning ERD theory, far less has been written about actually developing the models.
Two of the most common ways of identifying entity types for developing ERDs (i.e. data models) are:
 Narrative descriptions of requirements
 Workshops
Both of the above methods rely to some extent on expert domain knowledge to identify additional, and reclassify
identified, entity types. We will now look at the first method, using a narrative description of the system requirements
on which to base the development of an ERD.
Rather than give a brief description of each stage and then visit each in turn, I will just take you on a journey and
then reflect upon what we will have done in a summary at the end. So now let‟s start with the narrative description.

4.3.1 Analysing a Narrative Description of Requirements

Frequently the first thing to be produced when someone has the idea of developing a database is a paper document
describing what he or she wants. This often includes something like the following:
We wish to develop an information system (database) for our hospital in which:
“Patients are treated in a single ward by the doctors assigned to them. Usually each patient will be assigned a
single doctor, but in rare cases they will have two. Patients either pay for their treatment directly or through an
insurance company.
Healthcare assistants also attend to the patients, and a number of these are associated with each ward.
Initially the system will be concerned solely with drug treatment. Each patient is required to take a variety of drugs a
certain number of times per day and for varying lengths of time.
The system must record details concerning patient treatment and staff payment. Some staff are paid part time, and
doctors and care assistants work varying amounts of overtime at varying rates (subject to grade).
The system will also need to track what treatments are required for which patients and when, and it should be
capable of calculating the cost of treatment per week for each patient.
When users use the system they will be able to print out as well as view on screen the results. (Although it is
currently unclear to what use this information will be put.) “
(taken from http://www.umsl.edu/~sauter/analysis/er/er_intro.html and expanded)

The first stage is to pick out all the nouns (names) in the above passage; this provides a good baseline from which
to consider possible entity types.

Exercise 5. Marking Nouns

Mark in the above narrative all the nouns (names) you see.

Robin Beaumont [email protected] 17/07/2010 Page 15 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)
You can see in bold below the possible set of nouns (names) that I have come up with:
“Patients are treated in a single ward by the doctors assigned to them. Usually each patient will be assigned a
single doctor, but in rare cases they will have two. Patients either pay for their treatment directly or through an
insurance company.
Healthcare assistants also attend to the patients, a number of these are associated with each ward.
Initially the system will be concerned solely with drug treatment. Each patient is required to take a variety of
drugs a certain number of times per day and for varying lengths of time.
The system must record details concerning patient treatment and staff payment. Some staff are paid part time
and doctors and care assistants work varying amounts of overtime at varying rates (subject to grade).
The system will also need to track what treatments are required for which patients and when and it should be
capable of calculating the cost of treatment per week for each patient.
When the users use the system they will be able to print out as well as view on screen the results. (Although it is
currently unclear to what use this information will be put.)”
(taken from http://www.umsl.edu/~sauter/analysis/er/er_intro.html and expanded)

Gathering together all the above nouns produces the following list, not in any specific order:

 Patients  System
 Doctors  Users
 Healthcare  Cost
assistants
 Care assistants  Time
 Ward  Day
 Drug treatment  Lengths
 Drugs  Details
 Patient treatment  Overtime
 Treatment  Rates
 Staff payment  Grade
 Staff  Week
 Insurance Results
company
 Screen  Use
  Information


The next stage is to consider which of these nouns are appropriate entity types to include in the ERD.

Robin Beaumont [email protected] 17/07/2010 Page 16 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

4.3.2 Identifying Appropriate Entity Types

Rumbaugh et al (1991 p152-3, repeated in Blaha & Rumbaugh 2005 p 185 -6) provides some guidance as to how
to identify appropriate entity types. He suggests that you consider the following issues:
 Redundant entity types Two entity types may express the same information (i.e. two names for the same
concept, or synonyms). In the above example DOCTORS and HEALTHCARE ASSISTANTS might be
considered to be just entity instances of the entity type called STAFF; however, this is unlikely in the above
example because we know from our own expert domain knowledge that doctors are concerned with
prescribing whereas healthcare assistants are not. In addition the information we plan to collect is
specifically concerned with prescribing. We therefore consider the entity type STAFF to be redundant, at
least for the time being; possibly when the system is developed further such considerations can be re-
visited.
 Irrelevant entity types If an entity type has little or nothing to do with the problem, it should possibly be left
out. In the above example one would need to clarify with the person requesting the system if they need
information stored about INSURANCE COMPANIES. Also if the system is initially only concerned with drug
treatments, is there any point collecting information about other treatments?
 Vague entity types In a narrative description, often words are used indiscriminately. In the above example
the description states that “Initially the system will be concerned solely with drug treatment” yet the following
paragraphs refer to PATIENT TREATMENT and also TREATMENT. Are these three different entity types or
not? Again we can only find out by discussing this issue with the person who requested the database. We
would probably suggest that we have two entity types called DRUG TREATMENT and TREATMENT where
TREATMENT is concerned with information about any non-drug intervention and probably not included in
the initial ERD.
 Entity types that are really attributes Often an initial entity type is an attribute. In the above, COST has
been classified as an entity type yet it is probably an attribute of TREATMENT or DRUG TREATMENT or
PATIENT (renaming it BILL).
 Multiple roles become entity types "The name of a entity type should reflect its intrinsic nature and not a
role that it plays. For example, OWNER would be a poor name for a entity type in a car manufacturer's
database. What if a list of drivers is added latter? What about persons who lease cars? The proper class is
PERSON (or possibly CUSTOMER), which assumes various different roles, such as owner, driver, lessee."
(Blaha & Rumbaugh 2005 p 186).
 Implementation information The ERD is concerned with defining the data that needs to be stored. It is not
concerned with the hardware or processing of the data. Therefore, in the above example SYSTEM,
SCREEN and USER can be ignored. Similarly RESULTS are an outcome of processing data. Perhaps
from your knowledge of Access or other databases, you will realise that such things as RESULTS and
reports are just a process of using a query or report within the database management system.
Another type of problem occurs resulting from the existence of homonyms. This is where two entity types with the
same name actually mean different things. For example the entity TREATMENT may mean very different things to
different healthcare professionals. In this instance it may be necessary to add one or more additional entity types to
express the different concepts more clearly (e.g. DRUG_TREATMENT and ART_THERAPY etc).
So what have we ended up with after all the above deliberations? The following is the revised list of entity types:
 Patients, Doctors, Drug treatment, Drugs

By concentrating on the drug aspect of treatment and considering each of Rumbaugh‟s guidelines we have reduced
the list from 27 to 4 for a possible initial ERD.
Exercise 6. Identifying Entity Types from a Narrative

Time: 60 minutes

Look through the DopeHead scenario, found in the scenarios handout, and:

1. Mark all the nouns and produce a list of them.

2. Using Rumbaugh‟s guidelines revise your list to identify possible entity types.

Robin Beaumont [email protected] 17/07/2010 Page 17 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

4.3.3 Good and Bad Entity Types

While the previous page was concerned with identifying entity types, we will now look at a few of the many
guidelines suggested by Reingruber & Gregory 1994 (p65-77) for what makes good entity types.
Entity type names:
 Should be unique for the particular model. (This does not usually apply to attribute names which only need
to be unique to a particular entity type.)
 Should be a singular noun (e.g. PATIENT not PATIENTS).
 Should be self-explanatory to those reading the ERD.
 Should follow any naming conventions locally defined by the modellers. For example, two common
constraints are that:
 They should be written in UPPER CASE.
 Possible spaces should be represented by the _ character (eg DRUG_TREATMENT
rather than DRUG TREATMENT).
 Should not be a name of an individual object (e.g. Freeman Hospital Newcastle upon Tyne).
 Should not express more than one concept (e.g. EQUIPMENT/BED).
 Should be documented in the system design specification. The description for each entity type should be
clear, unambiguous and supplemented with examples.

When carrying out this process it is a good idea to draw up a table similar to the one below to make sure you carry
out the process, the initial proposed entity type name on the left and the final entity type name on the right.
Initial entity Unique Singular Self- UPPER The _ Not a Not more than Documented Final entity
type name: noun explanatory CASE character individual one concept type name
object
Patients
Doctors
Drug
treatment
Drugs

Exercise 7. Clarifying Entity Types

Time: 60 minutes

1. Looking back at the hospital example, consider the following four possible entities. Use the above
guidelines to edit them appropriately:

Patients

Doctors

Drug treatment

Drugs

2. Revisit the initial set of entity types you defined for the DopeHead scenario and edit them accordingly. It
may help to draw up a table similar to the one above.

Robin Beaumont [email protected] 17/07/2010 Page 18 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

4.3.4 Warning about the ‘is a Kind of’ Situation

To start with, people are often confused about the relationship between entity types and entity instances thinking
that an entity instance “is a kind of” entity type. An example will help to make this clearer. Someone might say that
for the entity type ANIMAL entity instances in this case might be BIRD or HUMAN. This is not usually the case. The
reason for this is because thinking about what information you might want to collect about BIRDs and HUMANs
would probably be different in several respects for each of them. Would you really want to collect information about
the size of wingspan for a HUMAN or the IQ of a particular bird? In other words the set of attributes for each is
different. If you are unsure of this go back to Exercise three on page 11
Unfortunately it is not always as simple as this. Here is the argument for keeping BIRD and HUMAN as entity
instances of HUMAN. Suppose you were an antique dealer and just wished to collect information about the
instances of ANIMAL statue you had. In this case you would probably just want to instance of ANIMAL statues you
had rather than animal specific, information in which the amount and type would vary for each. In this situation you
are happy about keeping the same type of information about each of the statues regardless of what they represent,
probably just species name would do for your attribute name. The context dictates what entity types you will
find.

4.3.5 Workshops
The alternative to using a paper document to partially derive a set of entity types is to hold a series of interactive
workshops. This requires commitment, including the willingness to learn, from the person(s) who have requested
the system (i.e. database). They need to understand to a limited extent what you are learning now!
Often a mixed approach can be taken, using a paper document as the starting point, to a workshop. Running
workshops requires good group working skills, and must be carried out in a non-threatening manner. Often the
modellers need to show considerable tact when people get hung up on what the modellers themselves know to be
irrelevant issues (e.g. font size/colour, type of screen or computer etc) for the particular task at hand.
Now that we have looked at how to create an initial list of entity types we will consider the actual mechanics of
drawing entities.

4.3.6 Drawing ERDs

It must be remembered that database modelling, of which ERDs are a particular tool, is a discipline that has only
been in existence for about a quarter of a century. I was recently looking at a book on systems analysis written in
1984 (Daniels & Yeates) and only managed to find 2 of the 300 pages dedicated to databases, and within those two
pages no mention was made of ERD diagramming.
There are a large number of different ERD diagramming styles. And this document is not concerned with you
knowing any particular one intimately but rather being aware of the general style and meaning of them.
Learning to draw ERD's is usually a pen and paper exercise to begin with and I would strongly encourage you to
practice using this simple technique to begin with - be warned you'll also need a larger rubber! Drawing informal
ERDs and discussing them is a very good way of developing them in a measured and appropriate manner, flip
charts, or electronic white boards are excellent in this respect.
Other documents on my web site provide detailed tutorials on how to use specialised software (called case tools) to
draw ERDs. http://www.robin-beaumont.co.uk/virtualclassroom/contents.htm

Robin Beaumont [email protected] 17/07/2010 Page 19 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

Exercise (8.) MCQ

1. Which of the following types of software (applications) is most suitable for developing ERDs?

a. Desktop drawing package (e.g. Smartdraw)

b. Desktop publishing application to ease the use of textual explanations
c. Anything with drawing capabilities (e.g. Microsoft Word)
d. A CASE tool
e. A spreadsheet

4.4 Summary
Specifically we have gone through the following stages to come up with a list of initial entity types for both the
hospital and Dopehead scenarios and obviously we can also use this method to produce a list of candidate entity
types from any narrative description. The list below should remind you of the steps you went through:
1. Identifying nouns and drawing up a list
2. Removing:
a. Redundant entity types (e.g. synonyms)
b. Irrelevant entity types
c. Vague entity types
d. Entity types that are really attributes
e. Roles amalgamated into entity types
f. Implementation information
3. Adding entity types due to homonyms
4. Considering Reingruber & Gregory 1994‟s guidelines:
a. Initial entity type name
b. Unique
c. Singular noun
d. Self-explanatory
e. UPPER CASE
f. The _ character
g. Not an individual object
h. Not more than one concept
i. Documented
j. Final entity type name
We could have also used a more interactive approach with clients in a workshop environment, and developed
ERD's with them.
After a few MCQs we will move onto the second aspect of ERDs, the relationships. If you feel like one this would be
a good time to take a break.

Robin Beaumont [email protected] 17/07/2010 Page 20 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

Exercise 9. MCQs
1. Which one of the following best describes the technique used to identify entity types in a narrative?

a. Identification of verbs, the application of various criteria and standards (eg naming conventions etc) to
refine the list and then the creation of a list of appropriately named entity types
b. Identification of nouns then the application of ISO standards to create a list of appropriately named entity
types
c. Identification of verbs then the application of standards (eg naming conventions etc) to create a list of
appropriately named entity types
d. Identification of nouns, the application of various criteria (eg Reingruber & Gregory 1994‟s) and standards
(eg naming conventions etc) to refine the list and then the creation of a list of appropriately named entity
types
e. Identification of a few important nouns then the application of standards (eg naming conventions etc) to
create a list of appropriately named entity types

2. Which of the following criteria are true (choose three)?

a. Attribute names must always be unique for a particular model.

b. Entity type names must be unique for a particular model.
c. Attribute names must be unique for a particular entity type within a model.
d. An entity type should only express one concept.
e. Entity type names should be singular or plural nouns.

Space for you to make additional notes about Entity types and Instances:

Robin Beaumont [email protected] 17/07/2010 Page 21 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

5. Relationships?
At the beginning of this document I stated that ERDs consist of entity types and relationships. We have already
looked at entity types and will now move on to consider relationships. Just as with Entities you can consider
Relationships at the Type or Instance level but because I find it easier to think of them at the instance level I will
only discuss them at this level.
Relationships are shown on ERDs by lines, usually between entity types. A relationship is a method of linking
together entity types. Within an ERD, each entity type has one or more
relationships with a number of other entity types. A relationship is optionally
(see Reingruber & Gregory 1994 p174) given a name which is a verb
(action). For example:
DOCTOR has several PATIENT(s)
PATIENT visits a DOCTOR
The ERD opposite shows the two above relationships:

5.1 What Does it Mean?

The presence of a relationship between two entity types implies that one is linked to the other. One is said to be the
parent while the other is the child. Consider the following situation:
PATIENT owns MEDICAL_RECORD
Here the situation is that a PATIENT has a relationship to MEDICAL_RECORD. I have illustrated this in two ways
below (idea taken from Carter 1995 p34). PATIENT instance 5 (e5) is associated with MEDICAL_RECORD instance
9 (e9). You could substitute the words „linked to‟ or „related to‟ for „associated with‟ in the previous sentence.

Index card metaphor

e5

e9

PATIENT MEDICAL_RECORD
Important:
Set notation metaphor Relations
are to do
with entity
Instances
e5

e9

Robin Beaumont [email protected] 17/07/2010 Page 22 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)
The section below is not usually included when discussing relationships within ERDs. This is because it is
considered to be too low level; ERDs are meant to be at a high conceptual level. (Notice I have tried to avoid
discussing attribute characteristics such as integers, text lengths etc) However, I have found when teaching this
topic that students can only grasp the meaning of a relationship in an ERD when they are given a low level
explanation. In other words, what does a relationship mean at the end of the day in a database such as Access?

5.2 Foreign Keys - How it all Works

So how do you actually relate two entity types? It is
Patient
remarkably simple; you add a reference to the parent entity
surname forename date of birth id (instance) in the child entity instance by adding another
jones Amanda 19/12/1956 004
attribute. This reference in the child entity is called a foreign
key. To understand why it has this name, it is necessary to
Oaks gale 03/05/1966 005 understand a little more about attributes.
kowal Andrew 11/07/1970 006
Consider a typical set of attributes for a PATIENT entity
bull Chris 29/10/1055 007 type: name, address, date of birth etc. If we have a large
.... .... ... ... group of PATIENT entity instances, it is unlikely that we
could uniquely identify one from any one of these attributes.
What we need is a unique identifier such as a „patient
Foreign key number‟. Such an attribute that possesses a unique value
for each entity instance for a particular entity type is called a
primary key it is often given the attribute name ID.
Medical_record
id source date Patient_ID
If we copy the particular value of this primary key into the
child entity instances they will be linked. What we have
9 GP 19/12/1956 005 done is to copy the key value into an entity foreign to it,
10 St james 03/05/1966 .... hence the „foreign‟.
hospital
Taking the above example with PATIENT and
11 GP 11/07/1970 ...
MEDICAL_RECORD, the MEDICAL_RECORD entity instance
12 Istanbul 29/10/1055 ... 9 will possess a foreign key (attribute), probably called
something like patient_id with the value 5 to link it back to the
particular PATIENT entity instance with which it has a relationship.

5.3 What Name do you give a Foreign Key Attribute?

This is basically up to you. As a rule of thumb it is a good idea to give it the same name as the attribute (i.e. the
primary key) from which it was copied. Two different entity types can have the same attribute name(s). In the above
example the attribute called „doctor_id‟ in the DOCTOR entity type can also exist in the MEDICAL_RECORD entity
type. Sometimes it is sensible to add a prefix or suffix to the „foreign‟ attribute indicating that the attribute is a foreign
key such as „FKdoctor_id‟.
Do not worry too much about the foreign key idea at the moment. You will understand it at the end of the subsection
when we have worked through a number of examples.
There are a number of types of relationships, each of which we will now investigate separately. But before that we
will take a quick look at one important aspect of relationships: dependency.

5.4 What are Dependency and Referential Integrity?

Dependency is the concept of having something rely upon something else. For the reliance to be fulfilled the thing
that one depends upon must exist. In ERDs most relationships are of the parent/child type where an instance of a
child entity is dependent upon a particular instance of a parent entity (called existence dependency).
I previously described a relationship between the patient (parent) and medical record (child) entity types shown
below. Each patient has a single medical record.

Robin Beaumont [email protected] 17/07/2010 Page 23 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)
Repeating once again how this works; a value in the key index attribute PATIENT entity instance of 5 is copied into
the appropriate medical record entity instance into a specific attribute called the foreign key. Note this is not an
optional requirement for the link to exist; this must be the case. If you imagine the value being „copied‟ into the
foreign key attribute you can see how the term „dependency‟ is so apt. You can‟t have a non-existent value copied
into the MEDICAL_RECORD for a non-existent patient. Or can you? Let‟s consider some interesting possibilities:
 If you allow the foreign key attribute to take an empty value (technically called a null value – for our
purposes they are identical), a particular MEDICAL_RECORD instance can exist without an associated
PATIENT instance linked to it. Is this sensible? This situation is called an optional relationship. According
to Reingruber & Gregory 1994 (p 185) as well as other writers, such relationships should be avoided.
 If you disallow the foreign key attribute to take an empty value a particular MEDICAL_RECORD instance
can only exist after the appropriate PATIENT instance has been created. (You can‟t copy what you don‟t
have!) This is called a mandatory relationship (also called an existence dependency).
 If you change the value of the PATIENT key field in the relationship the MEDICAL_RECORD foreign key
value will need to be updated appropriately to keep the relationship intact.
 If you delete the associated PATIENT entity instance what becomes of the orphaned child records that once
were related to it? You have three options: delete all the child instances, change their foreign key values to
null or do nothing with them. Clearly the third option is not appropriate, as it would make no sense because
the whole purpose of the value is to indicate a relationship.
The actual database software in which a particular database resides is designed to manage all the above problems.
It basically keeps all the foreign key values („references‟) up to date and stops you from damaging the relationships
once you have set up the database. This process is called maintaining referential integrity.
Now that we have some idea of dependency we will look at the simplest type of relationship, called a one to one
relationship.

5.5 What is a One to One Relationship?

We have already looked at a one to one relationship on the previous page, that concerning a PATIENT and
MEDICAL_RECORD. The important aspect of such a relationship is to ensure that you have established which is
the parent and which is the child entity type by way of making it, if at all possible, a mandatory relationship. For
example, in the above situation we say that a MEDICAL_RECORD entity instance cannot exist without it being
associated with a PATIENT entity instance. However, it is possible for a PATIENT entity instance to exist without an
associated MEDICAL_RECORD
entity instance.
Mandatory one to one
relationships are often indicated
as 1:1 while optional one to one
relationships are shown as 0:1.
The diagram below shows
several different types of ERD
notation. In the original ERD
notation, optional/mandatory
preferences were not indicated.

Robin Beaumont [email protected] 17/07/2010 Page 24 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

5.6 What is a One to Many Relationship?

This type of relationship is the most common within databases. Examples include:
 DOCTOR has many PATIENTs
 STUDENT takes many COURSEs
 PERSON has many ACCOUNTs
 GARDEN contains many PLANTs

Patient
The parent is the „one‟ side of the relationship while the child
surname forename date of birth id is represented on the „many‟ side. In this instance one parent
jones Amanda 19/12/1956 004
entity instance can be associated with possibly many child
entity instances. The „many‟ aspect is shown on an ERD
Oaks gale 03/05/1966 005 diagram by „crows feet‟. See below for an example.
kowal Andrew 11/07/1970 006
When you are drawing a one to many relationship you are
bull Chris 29/10/1055 007 also explicitly stating a parent child relationship. If you delete
.... .... ... ... a particular parent instance and it has child instances
associated with it they will more likely than not be deleted as
well. (This is technically called cascade delete). However you
Foreign key can specify a „restricted deletion‟. But then you could end up
with child instances that are no longer associated with a
parent instance. Lets take our Patent / medical_record
Medical_record
example again and this time assume that a PATIENT can now
id source date Patient_ID have multiple MEDICAL_RECORD(s). Suppose we would
9 GP 19/12/1956 005 now like our patient Oaks to have two medical records this is
easily achieved by linking multiple entity instances in the
10 St james 03/05/1966 005
hospital
Medical record entity to the appropriate patient instance. In
other words now two entity instances have the same foreign
11 GP 11/07/1970 ... key value.
12 Istanbul 29/10/1055 ...

5.7 Optionality
As with the one to one relationship, it is
possible to have either a mandatory one to
many („1:N‟) or optional one to many
(„0:N‟) relationship. Once again, if at all
possible, all optional relationships should
be investigated thoroughly to see if they
can be converted to mandatory
relationships. Reingruber & Gregory 1994
provide details.

Robin Beaumont [email protected] 17/07/2010 Page 25 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

5.8 How Many is Many?

This does not matter initially when designing the ERD. But eventually you might want to consider the following two
aspects:
 What is the maximum likely number of child entity instances going to be for any parent entity instance?
 What is the average number of child instances for each parent instance (e.g. 1:5 or 1:50 etc)?
Exercise 10. Drawing Entity Types using DeZign - optional

Time: 60 minutes

Work through the first DeZign tutorial document. For details of how to obtain this tutorial see the first
section of this document - required resources.

Exercise 11. Drawing Relationships

Time: 240 minutes

1. Work through the second tutorial in the DeZign tutorial document.

2. Define and then draw the various relationships (you have identified the entity types in a previous
exercise!) that you have identified in the DopeHead scenario. You can do this either in pen and paper or
using the DeZign software.

5.9 What is a Many to Many Relationship?

The last page showed a relationship between GARDEN and PLANT_SPECIES as having crows feet at each end of
the relationship line. This is a many to many relationship which is often written „N:M‟. The diagram below provides
an example:
GARDEN PLANT_SPECIES

1
Humulus viridens
2
Cynara cardunculus
3
4 Digitalis purpurea
5
Papaver paeoniforum
6
7 Alcea facifolia

8 Robinia pseudoacacia
9
Acer
10
11
12

Robin Beaumont [email protected] 17/07/2010 Page 26 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)
In this example not only can one PLANT_SPECIES instance be associated with several GARDEN instances but
also a GARDEN instance can be associated with several PLANT_SPECIES instances. For example the
PLANT_SPECIES Humulus viridens might be in garden numbers 3, 7 and 12 while the gardens also contain other
species including Cynara cardunculus, Digitalis purpurea, Papaver paeoniforum, Alcea facifolia, Robinia
pseudoacacia and an Acer. The above situation is complex to say the least!
Because of the far reaching consequences of many to many relationships they are often broken down into two,
separate one to many relationships. In this example you might end up with GARDEN having many PLANTs, each of
which is a type of PLANT SPECIES, which I have renamed NAME.
Now notice that we have two parent entity types, GARDEN and NAME,
which can exist independent of each other. In other words we have a
set of GARDEN details and also a set of NAME details. This is not the end of the story as we will probably want to
be able to model the possibility of a plant having a history of names, as some plants get renamed regularly. It is
always advisable to convert many to many relationships to several one to many relationships after the
initial designs. If you are interested in finding out how many to many relationships are implemented in databases
see http://www.robin-beaumont.co.uk/virtualclassroom/chap7/s3/dbcon2.pdf.

Exercise (12.) MCQs

1. Which of the following statements is true (one correct answer)?
a. A relationship within an ERD is always a link between various entity instances. It is implemented within a
database by the use of a foreign key.
b. A relationship within an ERD is always a link between various entity instances. It is implemented within a
database by the use of a primary key.
c. A relationship within an ERD is always a link between two entity instances within an entity type. It is
implemented within a database by the use of a foreign key.
d. A relationship within an ERD is always a link between two entity instances within an entity type. It is
implemented within a database by the use of a primary key.
e. A relationship within an ERD is always a link between two fields within an entity instance. It is
implemented within a database by the use of a primary key.

2. Which of the following best describes an optional relationship (one correct answer)?
a. An optional relationship is one where a foreign key attribute can only take a specific value. For example, patient 001
may have a medical record 003 but no other.
b. An optional relationship is one where a foreign key attribute can take a null value. For example, a patient may or may
not have a medical record.
c. An optional relationship is one where a foreign key attribute can take any value except null. For example, a p atient
may have a medical record or a dummy record.
d. An optional relationship is one where a foreign key attribute may or may not exist. For example, a patient may have a
medical record.
e. An optional relationship is one where a foreign key attribute can take any value regardless of the values in the
associated primary key. For example, a patient may have a specific medical record, an undefined one or none at all
(depending upon your interpretation).

3. Which of the following statements is correct (one correct answer)?

a. A mandatory relationship is also called external dependency. A mandatory one to one relationship implies
that one instance of entity A is always associated with one instance of B. A mandatory one to many
relationship implies that one instance of entity A is always associated with zero or more instances of
entity B.
b. A mandatory relationship is also called existence dependency. A mandatory one to one relationship
implies that one instance of entity A is always associated with one instance of B. A mandatory one to
many relationship implies that one instance of entity A is always associated with zero or more
instances of entity B.
c. A mandatory relationship is also called existence dependency. A mandatory one to one relationship
implies that one instance of entity A is always associated with one instance of B. A mandatory one to
many relationship implies that one instance of entity A is always associated with one or more instances
of entity B.
d. A mandatory relationship is also called external dependency. A mandatory one to one relationship implies
that one instance of entity A is associated with zero or one instance of B. A mandatory one to many
relationship implies that one instance of entity A is always associated with one or more instances of
entity B.
e. A mandatory relationship is also called existence dependency. A mandatory one to one relationship
implies that one instance of entity A is associated with zero or one instance of B. A mandatory one to
many relationship implies that one instance of entity A is always associated with zero or more
instances of entity B.

Robin Beaumont [email protected] 17/07/2010 Page 27 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

6. Some Common Mistakes in ERDs

6.1 Confusing Instances and Types

In the example at the beginning of this document we ended up with four entity types: DOCTOR, PATIENT,
DRUG_TREATMENT and DRUG. If you carried out the exercises correctly you should now have the ERD shown
below.

It may be worthwhile for you to have a look at the

original narrative as well; we have certainly cut it down!
The important sentence is:
“Initially the system will be concerned solely with drug
treatment. Each patient is required to take a variety of
drugs a certain number of times per day and for
varying lengths of time.”
According to our ERD at left:
 A DRUG_TREATMENT instance is related to
many DRUG instances.
 A DRUG instance is related to a single
DRUG_TREATMENT instance.

For a variety of reasons this is the best way of modelling the relationship. However, let‟s consider another way the
relationships may have been modelled.
A different modeller argues that while a particular DRUG_TREATMENT
has one or more DRUGs a particular DRUG is associated with one or
more DRUG_TREATMENTs. While this may seem more logical it should
be realised that now if we change the particular value in an entity
instance in DRUG we may (depending upon how many DRUG_TREATMENT instances it
is related to) be changing several DRUG_TREATMENT entity instances at the same time!
Why has the confusion occurred? The problem is the result of two things:
1. The modeller getting confused between entity types and entity instances. The modeller
is correct in thinking that a particular DRUG_TREATMENT has one or more DRUGs. S/he
is working at the instance level (good) and thinking about a particular DRUG_TREATMENT
for a particular patient. However, s/he then states that a particular DRUG is associated with
one or more DRUG_TREATMENTs. Here s/he has forgotten that s/he is dealing with instances.
While we are talking about a particular patient‟s particular drug treatment, s/he has now moved to the level of
thinking about all patients‟ relationships to drugs. In this case the following is correct, A PATIENT takes many
DRUG(s) and a DRUG is associated with many patients thus:
Where we are thinking of DRUG as just a name of the drug, for example
Mr Jones and Miss Smith take Penicillin. This is not the situation in the
scenario where the DRUG is associated with a particular PATIENT and
DRUG_TREATMENT DRUG here refers to a specific dose given to a
PATIENT. It makes no sense for more than one PATIENT to share a single dose of something!
One method of avoiding this confusion is to create fake instances with attributes and values. For example:

PATIENT: mr jones DRUG_TREATMENT:May 2nd etc

Bed 5 etc 2003
Dr Smith
DRUG: Morphine batch
no:29485. Ampoule no: 21

2. Another reason why the modeller may have become confused is concerned with the concept of the lookup table,
which we will discuss next.

Robin Beaumont [email protected] 17/07/2010 Page 28 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

6.2 Including Lookup/Descriptor Entities

While the topic of lookup tables is very much concerned with the low level database design, the concept seems
often to impinge upon people‟s decision making when they are just beginning to draw ERDs. This is probably
because we are all familiar with the concept and therefore feel that it must be represented somewhere in the ERD.
A lookup table is something used by the users of databases to provide a pick list of possible entries for a particular
attribute. It might be something like a disease code or name of a local doctor.
Consider the ERD on the previous page PATIENT -> DRUG_TREATMENT -> DRUG. Let‟s assume a particular
doctor wishes to have the ability to select a drug from a pre-defined drug list, which seems entirely reasonable.
We could add another entity type to the ERD, DRUG_DICTIONARY, which would provide certain values to any
particular DRUG entity instance. In other words certain attribute values for any DRUG entity instance would be
obtained by looking up and copying the values from the DRUG_DICTIONARY entity. You might think that this is just
another one to many relationship, but look below and think about it:
In this situation, however, we do not want
real referential integrity as we don‟t want
Usually not a good the deletion of a particular
idea when you start DRUG_DICTIONARY instance at any time
considering the to delete any of the DRUG entity instances
that may possess the value copied from
model!!
the DRUG_DICTIONARY.
This process of copying values from
another table but not having referential
integrity constraints is usually how a
lookup table works. Such details as lookup
tables are not usually shown in ERDs, at
least not the first few you develop. Thinking
about it this makes sense because just looking at the above ERD there are several other places we could have
lookup tables: one to show the doctors, one for the drug treatments and even one for selecting patients. The ERD
would start to look very cluttered.
Another reason why it makes sense not to show the lookup tables is because frequently these tables are simply
queries on the attribute you are currently working with. For example you may want to add a new drug attribute value
for a new instance of a DRUG based upon all the values that exist in the instances that already exist. You may
remember that when we started to consider which where appropriate entity types, one of the exclusion criteria
where those possible entity types we came up with that were simply the result of processing information from other
entity types.
Just one final warning: although I have indicated that it is not a good idea to show lookup tables in an ERD, there
may be occasion to show the odd one. For example, in the above the DRUG_DICTIONARY might be something
important in your model such as the British National Formulary or something similar. You would probably then
include it and add some text in the ERD providing additional information.
You should begin to realise now that there is not one correct or perfect ERD. Much depends upon the unique
situation in which each is developed.
Lookup tables are also called descriptor tables by modellers who specialise in developing a special kind of database
called a data warehouse. (Nelson G S 2001 Implementing a Dimensional Data Warehouse with the SAS System
SUGI26 proceedings. See: http://www2.sas.com/proceedings/sugi22/DATAWARE/PAPER129.PDF)
Hernandez 1997 p344-349 Calls them Validation tables and recommends that such entity types be distinguished in
the ERD from other entity types by having a distinctive marking.
Some software tools such as System Architect (SA) differentiate between the display of identifying and non-
identifying relation lines by drawing an identifying relation line as a solid line and a non-identifying relation line as a
dash line. An identifying relation indicates that the parent entity identifies the child entity. A non-identifying relation
indicates that the child entity does not rely on the parent entity for its identification in this instance the parent can be
specified as mandatory or optional when the relation is non-identifying. (Much of the information from this last
paragraph has been taken from the excellent SA help file).

Robin Beaumont [email protected] 17/07/2010 Page 29 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

6.3 Unnecessary Complexity

An historical aside – Occam’s Razor
Entia non sunt multiplicanda præter necessitatem – entities ought not to be multiplied except from necessity
Which means that all unnecessary facts or constituents in the subject being analysed are to be eliminated. Occam dissected
every question as with a razor. William of Occam, the Doctor Singularis et Invincibilis (d. 1349), the great Franciscan scholastic
philosopher, was probably born at Ockham, Surrey, UK. Occam being the latinized form of the name.
From Ebenezzer Cobham Brewers Dictionary of Phase & Fable (Wordsworth edition revised by Ivor Evans 1993)

One of the main problems with ERDs is the desire to introduce a multitude of relationships and entities when
thoughtfulness (along with a large amount of scrap paper!) can usually produce a model with much more clarity.
Here are some examples.
Unnecessary entities and relationships
Consider the following:
 A COURIER delivers one or more SAMPLEs.
 Each sample can have zero or more TESTs carried out
on it.
 Each TEST produces zero or more FINDINGs.
 Each sample is signed for by the
RECEIVING_TECHNICIAN.
 The sample is then managed by a
SUPERVISOR_TECHNICIAN who also confirms any
FINDINGs.
 The TESTs are carried out by JUNIOR_TECHNICIANs.
From the above information one possible ERD is shown
opposite. This all looks very complex. The entity types
COURIER, SAMPLE, TEST and FINDING seem sufficiently
different. However, how different are the
RECEIVING_TECHNICIAN, SUPERVISING_TECHNICIAN and
JUNIOR_TECHNICIAN? I even wonder that on a Sunday
afternoon they could in real life be the same person.
Obviously the modeller has tried to present the information that is collected at
each stage concerning the TECHNICIAN input, but I wonder if the following
might be just as suitable.
To indicate that a particular grade of technician should only carry out certain
tasks a small amount of text could be added to the ERD and it is important to
realise that a ERD can not express all the complexities you may wish to model.
Looking again at the ERD there still appears to be a large number of
relationships from the TECHNICIAN entity to various others. Do we need all
these? To help answer this question I would like to digress slightly and consider
two different aspects:
 Independent versus sequential relationships
 Multiple level relationships

Avoid loops of relationships - one may be redundant

Carter 1995 p.49 recommends that if there are a number of relationships which form a loop in your ERD if possible
the loop should be broken. For example in the SAMPLE -> TECHNICIAN -> TEST situation in the ERD above you
should consider very carefully if you need all three relationships. In the above ERD you do but there are situations
where you do not. For example think about the three entity types CUSTOMER, INVOICE and PAYMENT and
assume we have relationships between all three. However in this situation the relationship between Customer and
PAYMENT is no really necessary as any payment information can be obtained by going through the Invoice which
obviously will exist before a payment is made.

Robin Beaumont [email protected] 17/07/2010 Page 30 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)
Independent versus sequential relationships
Consider the following two ERDs:

Can enter VISIT

instances only if a
suitable PATIENT
instance exists
DEPENDENT

Can enter VISIT

instances
independent of
PATIENT instances No “correct” answer
INDEPENDENT
depends on the situation

Multiple level relationships

This is the situation in the COURIER-> SAMPLE ->TEST-> FINDING ERD where the
TECHNICIAN entity has a relationship at three levels. Considering the information above
about dependency, you should begin to realise that if we had a mandatory relationship (1:n)
rather than an optional one (0:n) the situation would be made easier. Why? Because with a
mandatory relationship we know that a parent instance would always possess at least one
child instance. Let‟s presume the ERD opposite was changed so that the relationships
between SAMPLE ->TEST and TEST and FINDING were mandatory. If this were the case
we would know that every SAMPLE would be related to at least one TEST and FINDING
instance. Given this information we might then revise the ERD to have only one relationship
from TECHNICIAN which would then probably be to FINDING.
While the above is one possible method of making the ERD much clearer, it might not be
appropriate to the situation. The question is: what is the real situation and once again what is one trying to achieve?

6.4 Poor Aesthetics

The diagram below shows a poor version of the above ERD. It is unsatisfactory because the entities and
relationships are poorly positioned.

Lines crossed

Entity
inappropriately
positioned

End of relationship
lines overlapping.
Which is mandatory
which is optional?

Robin Beaumont [email protected] 17/07/2010 Page 31 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

7. The Relationship between Narrative Descriptions and ERDs

Throughout this document I have used narrative descriptions as the basis for designing ERDs. This is usually the
case. Importantly the resulting ERDs, after discussion with those who wrote the narratives, are clearer than the
original textual documents, at least I always find them to be so. This is probably because I‟m very much a graphical
rather than a textual person. I always find myself drawing family trees when reading a Russian novel to keep track
of the characters!
We spent some time at the beginning of this document identifying entity types and relationships in narratives. A brief
summary (partially derived from the DeZign help file) of the information presented on those is given below.

One-to-One (1:1) Relationship

This is when at most one instance of an entity A is associated with one instance of entity B. For example:
“Employees in the company are each assigned their own office.” For each employee, there exists a unique office,
and for each office there exists a unique employee.

One-to-Many (1:N) Relationship

In this type of relationship, for one instance of entity A there are zero, one or many instances of entity B, but for one
instance of entity B, there is only one instance of entity A. For example: “A department has many employees but
each employee is only assigned to one department.”

Many-to-Many (M:N) Relationship

This relationship, sometimes called non-specific, is when for one instance of entity A there are zero, one or many
instances of entity B, and for one instance of entity B there are zero, one or many instances of entity A. For
example: “Employees can be assigned to no more than two projects at the same time; projects must have assigned
at least three employees.” A single employee can be assigned to many projects; conversely, a single project can
have assigned to it many employees.

Optionality
The existence of an entity in a relationship is defined as either mandatory or optional. If an instance of an entity
must always occur for an entity to be included in a relationship, then it is mandatory. An example of mandatory
existence is the statement “Every project must be managed by a single department.”

Mandatory = must
If the instance of the entity is not required, it is optional. An example of optional existence is the statement
“Employees may be assigned to work on projects.”

Optional = maybe

The following exercise gives you some practice at working between narrative descriptions and ERDs.

Robin Beaumont [email protected] 17/07/2010 Page 32 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

Exercise 13. Developing ERDs from Narratives

Time: 60 minutes

These excellent examples have been taken from David McLean‟s ([email protected]) home
pages. He is a senior lecturer in computing at Manchester Metropolitan University
(http://www.doc.mmu.ac.uk/online/SAD/T06/erdexes.htm no longer active June 2006).

A) Produce an ERD for the following scenario:

This scenario describes the activities of the Science and Technology faculty in a university in northwest
England. Students take courses in the university. The courses are run by departments. It is possible for a
course to be run by more than one department. Each course consists of a number of units. Lecturers teach
the units, and in some circumstances more than one lecturer will teach a unit. Lecturers of course teach on
more than one unit. Since the introduction of 'modular' courses, it is possible for one unit to be included on
more than one course. A lecturer may be a course leader for a particular course.

B) Produce an ERD for the following scenario:

A farmer wishes to keep computerised records on the milk and calf production of the dairy herd. All calves
produced are sold and not added to the dairy herd. Each cow has a name and date of birth, and will
produce milk for a lactation period after the birth of a calf or calves. Milk recordings for each cow in terms of
litres are taken each day. The information required for each pregnancy of a cow are the bull's name, date of
mating, date of birth of calf or calves and each calf's sex and birth weight. The system is to provide the
following information to the farmer:

• Details of all births of calves attributed to each bull

• Milk yield of a cow over a particular pregnancy (note: a lactation period is associated with each
pregnancy)

C) Produce an ERD for the following scenario:

Olde Worlde Homes (OWH) is a regional building company employing both permanent and contract staff.
The main activity of the company is building houses that recapture the feeling of bygone days on land it
acquires cheaply. All matters of policy (eg what land should be purchased) are determined by
management. Once the land has been purchased, the land is known as a 'site' and details of it such as
area, cost, vendor etc are kept so that management decisions can be assessed.

Several house are built on the site; the number and style are determined by the location, the land size,
target market and the physical assets of the land (eg trees or ponds). It follows from this that all houses are
built to an individual design ('architect designed' according to the sales literature). Information on each
house such as number of bedrooms, style, sale price etc is recorded, again so that management can
review their decision making and also monitor costs.

In order to maintain quality, only building materials which have been approved by the quality assurance
manager can be used to build the houses. Records are kept of the approved materials. All the materials
which are used to build a particular house are itemised and then purchased from the cheapest supplier.
The items (bricks, cement, kitchen units etc) are noted on a list which also contains details of the supplier
and cost. A list of suppliers containing nature of business, telephone number and address is maintained.
Once the houses have been built, they are marketed directly by OWH.

Houses are sold directly to the purchaser (or in some cases joint purchasers) without involving estate
agents. Details of purchasers are kept on file.

Robin Beaumont [email protected] 17/07/2010 Page 33 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

D) The following is a brief description of a library.

Borrowers take books out on loan. A borrower may take out up to five books at any one time. Several
copies of the same book are held for books which are continually in demand. Borrowers may make
reservations for titles which are out on loan.

Items which need to be stored include borrower number, borrower name, date borrowed, ISBN, acquisition
number (allocated when a book is purchased), date acquired, title, author, borrower making reservation,
and date reserved.

The library needs to be able to do the following:

1. check whether a book returned has been reserved
2. check which borrower has a particular book out on loan
3. check which books a borrower has out on loan
4. check whether a copy of a particular title is in the library

Draw an ERD to represent the library. Resolve the many-to-many relationships and suggest identifiers and
the main attributes for each entity.

State any assumptions you make.

Do not turn the page until you have tried the

above exercises!!

Robin Beaumont [email protected] 17/07/2010 Page 34 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

Discussion of Above Exercises

David McLean thoughtfully does not only provide solutions to the above exercises but also information as to how
he came to the solutions. I have taken the liberty of expanding his comments, as clearly his material would be
supplemented by face to face explanations.
A) University course scenario
The main facts abstracted from the scenario are given
1. Students take courses in the university. (see text) opposite.
2. The courses are run by departments. Firstly notice that the list does not include “the Science and
3. It is possible for a course to be run by more than one Technology faculty in a university in northwest England” (ie
department. the University). This is because there is no need to include
4. Each course consists of a number of units. an entity type that only has one instance and includes all
the others.
5. Lecturers teach the units,
6. in some circumstances more than one lecturer will Considering the facts, we come out with the ERD below;
teach a unit. this is only one of several possible solutions based upon
7. Lecturers teach on more than one unit.
how you interpreted the narrative. You may have even
added some extra „domain specific‟ knowledge such as a
8. Since the introduction of 'modular' courses, it is relationship between DEPARTMENT and LECTURER.:
possible for one unit to be included on more than one
course. You will notice that there are several many to many
9. A lecturer may be a course leader for a particular relationships.
course.

There are several many to many relationships in the

ERD which we could develop by adding additional
entities:

Robin Beaumont [email protected] 17/07/2010 Page 35 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

B) Dairy farmer scenario

1. All calves produced are sold and not added to the Initially we have a many to many relationship between
dairy herd.
Cow and Bull. This is easily resolved to Cow has zero or
2. Each cow has a name and date of birth, and will more MATING(s) and Bull has zero or more MATING(s).
produce milk for a lactation period after the birth of a All the other relationships appear to be of the one to many
calf or calves.
variety; each MATING can result in a pregnancy which will
3. Milk recordings for each cow in terms of litres are result in zero or more CALVES and CALFS. There will
taken each day. also be a LACTATION period.
4. The information required for each pregnancy of a cow
are the bull's name, date of mating, date of birth of
calf or calves and each calf's sex and birth weight.
5. The system is to provide the following information to
the farmer:

 Details of all births of calves attributed to each bull

 Milk yield of a cow over a particular pregnancy

(note: a lactation period is associated with each
pregnancy).

A lecturer may be a course leader for a particular course.

The diagram below shows the attributes that are described in the scenario:
David McLean provides the following possible
solution which is simpler than mine and probably
does the job as well. Quoting David, “by reading the
question carefully, the key features are calves, bulls,
milk yield and cows”:

Robin Beaumont [email protected] 17/07/2010 Page 36 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

C) Olde Worlde Homes (OWH) scenario

I have just reproduced David‟s solution below. In the final version it would have been appropriate to convert the four
many to many relationships to several one to many relationships by adding intermediate entity types such as
HOUSE_ARCHITECT between HOUSE and ARCHITECT etc.

D) Library scenario
Once again I have just reproduced David‟s solution below.

You now have a good grasp of the basics of ERDs. But before we finish all the practical aspects of ERDs I would
like to introduce the idea of relationships that occur within a single entity type.

Robin Beaumont [email protected] 17/07/2010 Page 37 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

8. Recursion in ERDs
Recursive relationships are also called unary or involuted relationships.
So far we have discussed examples of relationships between entity types; however, it is possible to have links
within an entity. Such a relationship is called recursive.
The usual example given is that of an EMPLOYEE entity type that has a relationship with itself called supervisor. In
other words an EMPLOYEE entity instance can relate to another EMPLOYEE entity instance in a supervisor role.

Clearly a „one to many‟ or „many to many‟ recursive relationship may also exist. (see
Carter 1995 p61 - 68, Elmasri & Navathe 1989 p. 49) Most writers suggest that these
recursive relationships are acceptable in early models but should be converted to „one
to many‟ or „one to one‟ relationships, as the model is refined. For example the above
recursive relationship could be modelled by introducing a new entity type called
ROLE.

Finkelstein (1989), amongst others, has recommended

a process to remove recursive relationships basically
by creating a „role‟ entity type as in the above example.
Furthermore he specified that models where such
relationships had been removed were defined as being in fifth business normal form:
“An entity is in fifth business normal form if its dependencies on occurrences of the same entity or entity type have
been moved into a STRUCTURE entity”
(Finkelstein 1989 quoted in Reingruber & Gregory 1994 p207)

Very briefly, when designing a database one aims for a highly normalised (basically structured) design but how you
achieve this is a matter of dispute. Each normal form represents a greater degree of data structure. This usually
involves breaking entity types up from those that have a large number of attributes to several with a smaller number
of attributes.
I have not discussed the process of normalisation in this document as there are two conflicting schools of thought
about how you go about developing these eventually highly normalised ERDs. One school of thought recommends
the method I have described in this document, you take a narrative and abstract the nouns etc and constantly refine
your model. A conflicting more formal approach suggests that you subject an initial model to a formal process of
normalisation to obtain the highly normalised ERD. If you wish to find out more about normalisation see
http://www.robin-beaumont.co.uk/virtualclassroom/chap7/s7/index.htm
Whichever method you use it should now be clear to you that developing ERDs is an iterative process, starting with
very rough models and gradually refining them. The topic of the next section firstly considers how this process has
been refined into a number of stages and then secondly where ERDs fit in the whole database development
process, but first a quick exercise.

Exercise 14. MCQ

Which of the following are correct statements about recursion (choose two)?
a. Recursion is a valuable modelling technique in detailed data models.
b. Recursion is added when making a data model comply to fifth business normal form.
c. Recursion is added at the end of the modelling process.
d. Recursion is modelled by introducing „structure‟ entity types in the latter data models.
e. Recursion is something that is allowed at the beginning of the modelling process.

Robin Beaumont [email protected] 17/07/2010 Page 38 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

Exercise 15. Recursion

Time: 10 minutes

Consider this ERD. Think

where recursion is implied and
suggest a solution.

9. Conceptual, Logical and Physical Data Models

Much has been written about „conceptual‟, „logical‟ and „physical‟ data models. Unfortunately most of it is utter
dribble! The problem is often made even worse because of this unsatisfactory situation, by well meaning writers
who continually develop other categories of models such as Cook & Daniels‟ (1994 p10) essential, specification and
implementation models.
The division of logical and conceptual data models was primarily done to help those developing databases.
Many years ago, in 1978 to be exact, the ANSI/X3/SPARC committee produced a standards document specifying
what a database management system (DBMS) should look like. Within it they specified, very sensibly, that there
should be three levels:
 External (‘User’) views (ie what the doctor sees of the database is a different view from that of a secretary)
 Conceptual schema (the ERD, etc)
 Internal schema (the actual 0 and 1 digits which comprise files in the computer)
The idea was that most people worked at the top two levels, which is now taken for granted; nowadays you write a
database using a DBMS. No one in his or her right mind would write a database using basic C++ or any other
programming language.
However, it soon became clear that within the conceptual level there was a great deal of variation. This level then
became (by various writers and database software companies) divided into conceptual, logical and physical
levels. Unfortunately there is no standard definition or consensus as to what this actually means. Take for example
the following situations:
 One person creates an ERD and provides a list of attributes for each entity type. While the foreign key
attributes are not defined yet they are clearly marked by the relationship lines on the ERD. (You can usually
work them out from the diagram.) In contrast, another person adds all the foreign key attributes in the
narrative description of the data model.
 One person creates an ERD and provides lists of attributes, which specify detail data types (integer, string
etc) while another does not.
 One person creates an ERD with „many to many‟ and recursive relationships in it. Another „corrects‟ all
these types of relationships to „one to many‟ or „one to one‟ types.
 One person ignores the various lookup tables in the ERD while another adds the majority of them in.

Robin Beaumont [email protected] 17/07/2010 Page 39 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)
If you took a straw poll of database developers they would say that a conceptual model does not have foreign keys
explicitly defined as attributes, whereas for the second and third situations given above it would be much more
difficult to gain any consensus about what level the model was at. My advice is not to worry. You always start with a
conceptual model and end up with a physical one.
The important thing to realise is that the conceptual model (the ERD) will need tweaking into a physical model
before it can be translated into a particular DBMS. All these models sit within the conceptual level of the
ANSI/SPARC framework.
Obviously, if you have created an ERD within a DBMS such as Access, it must be a physical model.
In summary:
 If the ERD has foreign key attributes defined, it is probably at the physical level.
 If the ERD does not have complete attribute definitions, it is not at the physical level.
 If the ERD has been created within a DBMS, it usually must be at the physical level.
 Most models do not include foreign key attributes

My advice to you is to be like Reingruber & Gregory, and the vast majority of actual database designers, by
totally ignoring the arbitrary divisions. Just say you start off with a high level model and gradually add
detail until it is possible to map the model directly onto the DBMS of your choice.

The problems described above are related to the history of the entity concept. For details see
http://www.aisintl.com/case/method.html.

9.1 Electronic Message Board Example

Barry Williams has been generous enough to put a collection of data models on the web at
http://www.databaseanswers.com/data_models/. The following is one of them and considers the development of an
electronic message board. The author provides examples of what he considers to be the „conceptual‟ and „physical‟
data models as well as Access code to create the database.
Conceptual model Logical model

In this instance both models have the same number of entity types; however, this is not always the case as you will
have already realised with the situation concerning lookup tables. Notice also the addition of the foreign keys in the
physical model. You can also download the SQL script to make this into an Access database at
http://www.databaseanswers.com/data_models/msg_board_physical.htm.

Robin Beaumont [email protected] 17/07/2010 Page 40 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

Exercise 16. MCQ

Which of the following statements best describes conceptual, logical and physical data models?

a. A loosely defined set of terms, often used inappropriately, indicating the gradual progression from a high level
data model to one which provides all the detail required to implement it in a specific database system.
b. A clearly defined set of terms, often used inappropriately, indicating the gradual progression fr om a data model
containing attribute descriptions to one which provides all the detail required to implement it in a specific
database system.
c. A loosely defined set of terms, often used inappropriately, indicating the gradual progression from a data model
containing attribute descriptions to one which provides all the detail required to implement it in a specific
database system.
d. A clearly defined set of terms, often used inappropriately, indicating the gradual progression from a high level
data model to one which provides all the detail required to implement it in a Microsoft compliant database
system.
e. A clearly defined set of terms, often used inappropriately, indicating the gradual progression from a high level
data model to one which provides all the detail required to implement it in a referential database system.

10. Where do ERDs fit Into the Database Design Process?

At the beginning of this document we mentioned that ERDs play a central role in the design of databases. You
should now see why this is the case; they define the actual database‟s structure. If this structure is wrong, no
amount of tinkering around at the edges will solve the problem.
Obviously the way the database looks to the user (the user interface or UI) is also very important, and it is a very
difficult balancing act to calculate the relative amount of attention that should be given to UI and data structure
aspects in the real world with limited resources.
I would argue that the development of a good quality ERD (or a number of them for large databases) and
associated data dictionary is a fundamental prerequisite for success.
Where exactly ERD modelling fits into the database design process depends upon the size of the project. We will
consider two scenarios: the small personal database project and the large departmental database.

10.1 The Small Personal Database

This might be a small database for just you or perhaps

for a research assistant as well. The method described
below would not be suitable if you were planning on
giving it to a team of people to use or if it were a „mission
critical application‟ as the IT people would say. To the
right is the development method suggested by John
Carter, a well respected database educator in the UK, in
his book Database Design and Programming with
Access, SQL and Visual Basic, McGraw Hill 2000 (p2).

Robin Beaumont [email protected] 17/07/2010 Page 41 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

10.2 Departmental and Hospital Systems

In this situation you are likely to go through a formal process in which you initiate either a formal procurement
process for an existing system which might be adapted to your needs or the wholesale development of a new
system. In either case a much more rigorous process will be followed than that shown above. There are numerous
database method(ologie)s available for the management, procurement, specification and implementation of large
systems. Most large computer systems companies have their own methods (e.g. Oracle, ICL, IBM etc) which are
frequently based upon fairly standard techniques such as IDEF1X, SSADM or Star. All these techniques rely to a
great extent upon ERD modelling; although the lines on the diagram might be a different shape for each, there is
really very little difference between them.
In the UK, SSADM and “Structured Systems Analysis and Design” are registered trademarks of the Government
Centre for Information Systems (CCTA) in the UK. The CCTA is responsible for the development of SSADM.
IDEFiX is the USA equivalent.
Obviously these various method(ologies) are based upon a set of implicit assumptions which „colour‟ their approach
significantly. I will not discuss this here; the document concerned with the theories (http://www.robin-
beaumont.co.uk/virtualclassroom/chap11/s4/sa1.pdf) behind systems methods will revisit this very important topic in
more depth.

Exercise 17. Database Methodologies

Time: 30 minutes

1. Please visit some of the sites below. I do not want you to get any detailed knowledge but rather to
become aware of such methods and how they differ from that described by Carter above.

IDEF1X (USA) National Institute of Standards and Technology (NIST): http://www.idef.com/idef1x.html

General list of different approaches:

http://en.wikipedia.org/wiki/List_of_software_development_philosophies

SSADM academic notes: http://www.comp.glam.ac.uk/pages/staff/tdhutchings/chapter4.html

2. Add a few of your own links below or if you are working through this document as part of a web based
course why not share any good or bad links with your colleagues via the discussion board.

Exercise 18. MCQ

ERD modelling is which of the following (choose two answers)?
a. A highly objective method of obtaining data requirements
b. A technique developed over the last two decades
c. A process that requires knowledge of the context of the model to be developed
d. A very subjective method that produces highly personalised models
e. A technique only used in small-scale database development

Robin Beaumont [email protected] 17/07/2010 Page 42 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

11. Publicly Available Data Models (ERDs)

There are several publicly available data models. A few are given below.

You can see a health centre data model, along with numerous others at:
http://www.databaseanswers.com/data_models/health_centre/index.htm
The NHS Data Dictionary & Manual (July 2006): http://www.connectingforhealth.nhs.uk/datastandards/
The NHS Healthcare model: http://www.virtualtravelog.net/projects/Ontology/CBS/hier.htm
An excellent article describing why high level data modelling approaches sometimes fail (specifically the above
models):
http://www.virtualtravelog.net/entries/2004/01/ontology_review_1_the_nhs_common_basic_specification_why_top_l
evel_ontologies_dont_work.html
UNITED KINGDOM: Max Jones‟ PhD thesis on „Formal generic Modelling‟ (no longer active July 2006)
EUROPE: The standards work of CEN/TC251
AUSTRALIA: http://www.aihw.gov.au/publications/index.cfm/title/9813
CANADA: http://secure.cihi.ca/cihiweb/splash.html
UNITED STATES:
CORBAMed, http://www.hipaanet.com/upin7-10.htm the healthcare domain taskforce of the Object Management
Group
HL7
The US Military Medical Technology site http://www.military-medical-technology.com/
One of my Msc students (Luisa Whitelaw) kindly suggested the following additional links on the module discussion
board:
http://support.esri.com/index.cfm?fa=downloads.dataModels.gateway
http://www.who.int/healthmetrics/library/south_africa_05apr.doc
http://www.nisc.co.za/databases
http://www.tama-sa.gov.za/terminology_sa_home.htm
http://www.doh.gov.za/nhis/docs/nchmis.htm
http://folk.uio.no/leopoldo/Publications/Papers/HCIS-1465JN_Finland.pdf
http://www.sita.co.za/documents/invitations/504_2006/RFB%20504EHR%20Strategy.doc
http://groups.csail.mit.edu/medg/courses/6872/2003/slides/lecture2-print.pdf
http://www.avert.org/safricastats.htm
http://www.tac.org.za/aidsstats.html (see links under section “further reading”)
http://iri.columbia.edu/outreach/training/course/bamako1999/html/slide/ZAMO.ppt#263,10,Proposed outputs/results
http://www.musc.edu/infoservices/lanvision/Regulations_Policies_Guidelines/ASTMEMRGuide.pdf
Exercise 19. Public Data Models

Time: 30 minutes

1. Please visit some of the sites listed above. I do not want you to get any detailed knowledge but rather to
become aware of such resources. Make sure you visit the NHS data dictionary and NHS Healthcare model
sites, but don‟t attempt to understand it all fully!

2. Add a few of your own links below or if you are working through this document as part of a web based
course why not share any good or bad links with your colleagues via the discussion board.

Robin Beaumont [email protected] 17/07/2010 Page 43 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

12. CASE Tools

You will have completed many of the tasks in this handout by drawing ERDS either with a pen and paper or using a
CASE tool, the introduction to this document links you to some CASE tool tutorials.
CASE tools began to be developed in the 1980s, and many thought they would replace programmers. The idea was
that you entered all the design details into the tool and it magically produced a finished system. Unfortunately this
has never quite materialised. Good programmers are just as sought after now, if not more so, than they were 20
years ago. A new approach called „Extreme Programming‟ has taken on a few of the good aspects of the CASE
tool approach, by bringing the programmer and the person(s) requesting the system much closer. You could say
that the traditional massive design documents have now been replaced with fluid micro experiences where people
work together in a problem solving manner, using a CASE tool as just one method to document their experiences.

12.1 What Advantage(s) do CASE Tools Offer Over Simple

Diagramming Tools?
A diagramming tool, such as the picture drawing facilities in Word 2000, allows you to draw a diagram easily
enough but that is where it stops. On the other hand a CASE tool is a repository (sometimes called an
encyclopaedia) of information. This repository is formed from the actual diagrams and other data collection
facilities contained in the CASE tool. This repository is in the form of a language (often in the form of a database
itself) which allows the tool to do, amongst other things, the following:
 Rule checking - It checks that you have obeyed various rules while drawing the diagrams.
 Consistency - It checks that there is consistency between diagrams (entities with the same names across
diagrams have the same relationships and attributes etc).
 Reporting - It can create various tabular reports from the information in the repository (eg the automatic
creation of a data dictionary from a number of ERDs).
 Database translation - It allows the information (ERDs etc) to be converted to a number of database
definition languages (eg Access 2000).
 Reverse engineer - It translates databases already developed into sets of ERDs and reports etc.
Most CASE tools provide support for several different database design method(ologie)s. Therefore a variety of
diagramming styles are provided.
A list is provided below of some database modelling CASE tools. It was taken from
http://www.databaseanswers.com/modelling_tools.htm:

PRODUCT VENDOR PRICE COMMENTS

Argo UML http://www.gentleware.co Free - Open Source For Modelling UML Diagrams - but not for the faint-hearted. Has a good
now called m. export facility for diagrams including svg. Requires Java installed on the
Poseidon computer.

Case Studio CharonWare Free demo Download Looks impressive - if you try it, please let me have your comments. Here‟s
Maximum 6 entities), a “5-Star” Review from Dan Horn on 27th. January 2002.
then $149 or $299

CASEWise CASEWise Aimed at Business Process Modelling, but has links to DataArchitect.

DataArchitect Sybase $2,000 If you are buying for the company, buy this one.
DataArchitect is part of Power Designer, which is described below.

Database from Jo Janssens Free A French version is also available. I found this link did not work on October
st
Design Tool 21 , 2002, but if you try it and find it works, please let me know.
(DDT)

Data Design Chilli Source $99 downloaded or DDS supports the ERD Modelling and reverse engineering is being added.
Studio $189 on CD

Robin Beaumont [email protected] 17/07/2010 Page 44 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

PRODUCT VENDOR PRICE COMMENTS

Dezign Datanamic Offers a free download If you are buying for yourself, buy this one - now offers a free Tutorial.
or $139 to buy.

Designer Oracle $200 Available from the Oracle Technology Network (free to join), offering
unbeatable value at a price of $200

Dia Gnome Office Free - Open Source Designed to be like Visio

Enterprise Sparx Systems $99 UML Analysis & Design Tool and very affordable.
Architect

ER Creator Model Creator (Danish) About $100 Includes a Tutorial and a Trial version. Nice, easy to use, simple tool.

ER/Studio Embarcadero A comparative Study of Database Design Tools is available (requires

Trial Version available
registration).

ERWin Computer Associates One of the Market Leaders

ImportER Datanamic Reverse Engineering for MySQL

Excellent value at $55
MySQL

MagicDraw No Magic inc. Free educational versions of the Personal edition for educational
Varies
institutions

Object Oracle Combines Object and Relational Modelling facilities

Database POA
Designer

OR-Compass Logic Works Object-Relational Modelling.

I saw this mentioned in a User Group somewhere, but the only reference I
found is in this 1998 article in DBMS magazine.
POA
Logic Works was taken over by Platinum, which was taken over by CA,
and the CA Product Listing does not include OR-Compass.
If anyone can give me a URL, I‟d be very grateful.

Pacestar A diagrammer rather than a CASE tool

Trial Version available
UML PAcestar
$239
Diagrammer

Platinum This is a Repository, rather than a Modelling Tool, with a „Where-used‟

Repository Computer Associates facility, and it lets you add entity types, such as Business Function,
Business Role and Script.

Power Sybase $2,000 If you are buying for the company, buy this one.
Designer The Data Modelling Tool in this composite product is called
“DataArchitect”.
Unfortunately, the design of the web site has changed recently and
information about Data Architect is difficult to find. But the product is
excellent, and I have found its powerful Reverse Engineering facilities to
be very useful.

ProVision “A Repository-based integrated business process and object modeling

WorkBench Proforma Corporation. Prices on Application toolset.”
An interesting range of Data Model Exchange facilities.

QDesigner Quest. From the „Toad Company‟ - looks good and is well-recommended by at
Pricing difficult to find
least one satisfied user - a downloadable Trial is available(rather
on the Quest Web Site
substantial at 63MB).

SmartDraw SmartDraw $50 Strictly a drawing package, so no SQL creation.

Robin Beaumont [email protected] 17/07/2010 Page 45 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

PRODUCT VENDOR PRICE COMMENTS

Telelogic Formerly Popkin now $3,000-5,000 Up until early 2007 a student edition of System Architect
System Telelogic
Architect
was available on CD-Rom which cost around $34 (ISBN: 0-
2007 07-293278-3). Unfortunately Telelogic, the company that
have taken over Popkins, have no plans to develop the
student edition and are singularly unhelpful. The alternative
tool they recommend, Telelogic modeler at
http://www.ilogix.com/sublevel.aspx?id=1756 can be freely
downloaded, however this is a far inferior piece of software
compared to the student edition of system architect. After
registering you can still download an evaluation copy of
Telelogic System Architect at the Telelogic web site.

Togethersoft TogetherSoft ? Large application, designed for large corporations

Toolkit for University of Twente, Free and Very interesting combination of Tools but not for the PC as runs under
Conceptual Holland Downloadable various Unix platforms
Modeling
(TCM)

Visio Microsoft Trial CD available Many versions - Standard ($199), Technical ($399), Professional ($399)
and Enterprise ($999)

Visual UML Visual Object Modelers Trial time limited Many versions
download

Visible Visible ? Designed for large corporations?

analyst

I agree with the anonymous author above about System Architect being the best if your company can afford it
however my note above detailing my dissatisfaction with Telelogic means that I would now recommend MagicDraw.
Exercise 20. CASE Tools

Time: 15 minutes

Visit the MagicDraw site listed above and then some of the others.

Exercise 21. MCQ

Which of the following are true statements about CASE tools (choose three)?
a. They provide rule checking.
b. They provide a method of automatically generating a user interface for a data model.
c. They are relatively cheap to purchase.
d. They provide a repository for information about the model from which reports can be generated.
e. Some provide reverse engineering capabilities.

Robin Beaumont [email protected] 17/07/2010 Page 46 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

13. Exercises

Time: 360 minutes (6 hrs)

It is very important to practice the skill of developing ERDs, and to help you I have provided a number of
more complex exercises than those you have encountered as you have worked through this document.

Go to my main website and download the “Scenarios for practicing modelling techniques” document:
http://www.robin-beaumont.co.uk/virtualclassroom/contents.htm

This document contains about ten scenarios.

Select two or three scenarios to:

• Develop an ERD for each scenario

• Provide a description of each entity type

• List a number of attributes for each of the entities plus a description of each attribute

• List a set of constraints / assumptions - You do not have the luxury of being able to question the client?

Hints:

Remember to use your own „expert domain knowledge‟ where appropriate. But always state clearly any
assumptions / constraints you have made for each of the models.

By the time you have finished with the model you should have:

• Removed all recursive relationships as described in the page concerned with recursion.

• Converted any many to many relationships to multiple one to many relationships.

Robin Beaumont [email protected] 17/07/2010 Page 47 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

14. MCQs
I have repeated all the MCQs together as a revision exercise for you.

1. ERDs provide a description of:

a. The processes that occur in the model

b. The various entities and their relationships in the model
c. The entities in the model
d. The processes and data requirements of a model
e. The User Interface aspects of the model

2. ERDs provide a (one correct):

a. Detailed description of the data within a data model

b. Detailed description of the proposed uses of a database
c. Guide to the training costs
d. High level description of the data within a data model
e. Graphical picture which is of little use in developing a database

3. From the list below choose two reasons why it is important for a „domain expert‟ such as you to learn about the
ERD method:

a. Provides insight into the mindset of database developers

b. Is the only method available to describe the data requirements
c. Provides credibility to IT personnel
d. Forms the basis of most data modelling techniques
e. Has been proved to be the most cost effective method of specifying data requirements

4. From the list below choose the one option that describes the most desirable „domain expert‟ from the medical
profession:

a. Someone who has developed several databases but knows little of database modelling or current issues
in medicine
b. Someone who has little interest in how information may help the department
c. Someone who has problems working in a collaborative environment
d. Someone who has previously managed IT projects
e. Someone who has knowledge of data modelling techniques and currently works in the appropriate
situation

5. Which of the following provides the best description of an entity type (select one)?
a. A specific concrete object with a defined set of processes (e.g. John Brown with diabetes)
b. A value given to a particular attribute (e.g. height - 230 cm)
c. A thing that we wish to collect data about where zero or more, possibly
real world examples of it may exist
d. A template for a group of things with the same set of characteristics that may exist in the real world
e. An undefined concept that needs further clarification

6. Which of the following provides the best description of an entity instance (select one)?
a. A specific concrete object with a defined set of processes (e.g. John Brown with diabetes)
b. A value given to a particular attribute (e.g. height - 230 cm)
c. A thing that we wish to collect data about where zero or more, possibly
real world examples of it may exist
d. A template for a group of things with the same set of characteristics that may exist in the real world
e. An undefined concept that needs further clarification

7. From the following list select two entity instances:

a. Steinway piano model D design
b. McGill type forceps as used in surgical operations
c. Tony Blair ( British prime minister)
d. PIII type chip that is found in many modern PCs
e. An advice leaflet for chronic back pain issued to Mrs Smith on 02/10/2002

Robin Beaumont [email protected] 17/07/2010 Page 48 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)
8. Which one of the following statements is true (select one)?

a. An ERD can only display entity type names.

b. Attributes are a rarely considered aspect of entities.
c. Attributes must always be displayed in ERDs.
d. Attributes equate to table names in a database.
e. Attributes can optionally be displayed in ERDs.

9. From the following, select the best list of attributes for the entity SHOE for someone working in a shoe shop
(select one):

a. 1, 2, 6
b. 1, 2, 6, 3
c. 1, 2
d. 1, 6
e. 1, 2, 3, 4, 5, 6
1=size
10. From the following, select the best list of attributes for the 2=number of lace holes
3=propulsion
entity RECIPE for someone following a recipe at home (select 4=cooking time
one): 5=number of portions
6=material colour
a. 4, 5, 7, 8 7=volume (loudness)
8=cooking Temperature
b. 4, 5
c. 4, 5, 6
d. 4, 5, 8
e. 4, 5, 6, 8

11. Which of the following best describes the technique used to identify entity types in a narrative:

a. Identification of verbs, the application of various criteria and standards (eg naming conventions etc) to
refine the list and then the creation of a list of appropriately named entity types
b. Identification of nouns then the application of ISO standards to create a list of appropriately named entity
types
c. Identification of verbs then the application of standards (eg naming conventions etc) to create a list of
appropriately named entity types
d. Identification of nouns, the application of various criteria (eg Reingruber & Gregory 1994‟s) and
standards (eg naming conventions etc) to refine the list and then the creation of a list of appropriately
named entity types
e. Identification of a few important nouns then the application of standards (eg naming conventions etc) to
create a list of appropriately named entity types

12. Which of the following criteria are true (choose three)?

a. Attribute names must always be unique for a particular model.

b. Entity type names must be unique for a particular model.
c. Attribute names must be unique for a particular entity type within a model.
d. An entity type should only express one concept.
e. Entity type names should be singular or plural nouns.

13. Which of the following types of software (applications) is most suitable for developing ERDs?

a. Desktop drawing package (eg Smartdraw)

b. Desktop publishing application to ease the use of textual explanations
c. Anything with drawing capabilities (eg Microsoft Word)
d. A CASE tool
e. A spreadsheet

Robin Beaumont [email protected] 17/07/2010 Page 49 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)
14. Which of the following statements is true?

a. A relationship within an ERD is always a link between various entity instances. It is implemented within a
database by the use of a foreign key.
b. A relationship within an ERD is always a link between various entity instances. It is implemented within a
database by the use of a primary key.
c. A relationship within an ERD is always a link between two entity instances within an entity type. It is
implemented within a database by the use of a foreign key.
d. A relationship within an ERD is always a link between two entity instances within an entity type. It is
implemented within a database by the use of a primary key.
e. A relationship within an ERD is always a link between two fields within an entity instance. It is
implemented within a database by the use of a primary key.

15. Which of the following best describes an optional relationship?

a. An optional relationship is one where a foreign key attribute can only take a specific value. For example,
patient 001 may have a medical record 003 but no other.
b. An optional relationship is one where a foreign key attribute can take a null value. For example, a patient
may or may not have a medical record.
c. An optional relationship is one where a foreign key attribute can take any value exc ept null. For example,
a patient may have a medical record or a dummy record.
d. An optional relationship is one where a foreign key attribute may or may not exist. For example, a patient
may have a medical record.
e. An optional relationship is one where a foreign key attribute can take any value regardless of the values
in the associated primary key. For example, a patient may have a specific medical record, an undefined one
or none at all (depending upon your interpretation).

16. Which of the following statements is correct (one only)?

a. A mandatory relationship is also called external dependency. A mandatory one to one relationship
implies that one instance of entity A is always associated with one instance of B. A mandatory one to many
relationship implies that one instance of entity A is always associated with zero or more instances of entity
B.
b. A mandatory relationship is also called existence dependency. A mandatory one to one relationship
implies that one instance of entity A is always associated with one instance of B. A mandatory one to many
relationship implies that one instance of entity A is always associated with zero or more instances of entity
B.
c. A mandatory relationship is also called existence dependency. A mandatory one to one relationship
implies that one instance of entity A is always associated with one instance of B. A mandatory one to many
relationship implies that one instance of entity A is always associated with one or more instances of entity
B.
d. A mandatory relationship is also called external dependency. A mandatory one to one relationship
implies that one instance of entity A is associated with zero or one instance of B. A mandatory one to many
relationship implies that one instance of entity A is always associated with one or more instances of entity
B.
e. A mandatory relationship is also called existence dependency. A mandatory one to one relationship
implies that one instance of entity A is associated with zero or one instance of B. A mandatory one to many
relationship implies that one instance of entity A is always associated with zero or more instances of entity
B.

Robin Beaumont [email protected] 17/07/2010 Page 50 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)
17. Which of the following statements best describes conceptual, logical and physical data models?

a. A loosely defined set of terms, often used inappropriately, indicating the gradual progression from a high
level data model to one which provides all the detail required to implement it in a specific database system
b. A clearly defined set of terms, often used inappropriately, indicating the gradual progression from a data
model containing attribute descriptions to one which provides all the detail required to implement it in a
specific database system
c. A loosely defined set of terms, often used inappropriately, indicating the gradual progression from a data
model containing attribute descriptions to one which provides all the detail required to implement it in a
specific database system
d. A clearly defined set of terms, often used inappropriately, indicating the gradual progression from a high
level data model to one which provides all the detail required to implement it in a Microsoft compliant
database system
e. A clearly defined set of terms, often used inappropriately, indicating the gradual progression from a high
level data model to one which provides all the detail required to implement it in a referential database
system

18. Which of the following are true statements about CASE tools (choose three)?

a. They provide rule checking.

b. They provide a method of automatically generating a user interface for a data model.
c. They are relatively cheap to purchase.
d. They provide a repository for information about the model from which reports can be generated.
e. Some provide reverse engineering capabilities.

19. Which of the following are correct statements about recursion (choose two)?

a. Recursion is a valuable modelling technique in detailed data models.

b. Recursion is added when making a data model comply to fifth business normal form.
c. Recursion is added at the end of the modelling process.
d. Recursion is modelled by introducing „structure‟ entity types in the latter data models.
e. Recursion is something that is allowed at the beginning of the modelling process.

20. ERD modelling is which of the following (choose two)?

a. A highly objective method of obtaining data requirements

b. A technique developed over the last two decades
c. A process that requires knowledge of the context of the model to be developed
d. A very subjective method that produces highly personalised models
e. A technique only used in small-scale database development

Robin Beaumont [email protected] 17/07/2010 Page 51 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

15. Summary
This document has introduced you to a number of new concepts and provided you with the skills to use them
effectively to produce ERDs. You have seen how to define and refine a set of entity types for a given scenario. You
have also considered in detail the various types of relationship that can exist between entity types.
The CASE tool concept has been introduced to you by way of seeing what a number of suppliers have to offer.
Other documents provide practical exercises in using CASE tools.
I would now recommend that you return to the learning outcomes at the beginning of the document and see how
much you have learnt!

16. References
Carter John 2000 Database design & programming with Access, SQL and visual basic McGraw Hill
Carter John 1995 The Relational Database. Chapman and Hall [An excellent introductory book]
Date C J. 1995 (6th ed.) An introduction to database systems. Addison-Wesley.
Elmasri R Navathe S B 1989 Fundamentals of database systems. Benjamin Cummings Wokingham UK.
Everest G 1986 Database management. McGraw-Hill. London.
Finkelstein Clive 1989 An introduction to information engineering. Addision-Wesley. Wokingham UK.
Finkelstein Clive 1992 Information engineering. Addision-Wesley. Wokingham UK.
Hernandez M J 1997 Database design for mere Mortals. Addison - Wesley
Martin James 1981 An end user‟s guide to data base. Prentice Hall [ISBN 0-13-277129-2]
Reingruber Michael C. Gregory William W 1994 The Data Modelling Handbook John Wiley & Sons Chichester
Rumbaugh J Blaha M Premerlani W et al 1991 Object-Oriented Modelling and design. Prentice Hall.
Blaha M Rumbaugh 2005 (2nd ed.) Object-Oriented Modelling and design with UML. [basically this is the seocnd
edition of Rumbaugh et al 1991]Prentice Hall.

17. Links
University of Texas data modelling notes:
http://www.utexas.edu/cc/database/datamodeling/dm/objects.html
Links to database articles (applied information science web site):
http://www.aisintl.com/case/reading.html
Wikipedia has a good article about ERDs:
http://en.wikipedia.org/wiki/Entity-relationship_model
Additional ones for you to add:

Robin Beaumont [email protected] 17/07/2010 Page 52 of 53

 Health Informatics
Entity Relationship Diagrams (ERDs)

18. Appendix A Relationship Terminology

The following table is taken from Carter 1995 p40. Different writers use different words to describe the minimum and
maximum constraint on a relationship; the synonyms are listed below. I have tried to avoid using most of the terms
in this document to avoid confusion. I tend to refer to relationships as being “optional” (where the minimum number
of participants is zero) or “mandatory” (where the minimum is one).

Source (writer) Number of entities in Minimum number of Maximum number of

relationship participants participants
Date Degree
IEW Optionality Cardinality
D.C.C Optionality Degree
Ashworth Degree
Eva Optionality Cardinality and degree
Kroenke Degree Minimal Cardinality Maximal Cardinality
Bamford Degree

Optionality

Cardinality

Document details:
C:\HIcourseweb new\chap11\s9\erds_1.doc
End of document

Robin Beaumont [email protected] 17/07/2010 Page 53 of 53