COM 117 Introduction to Application Building Week 2 Entity Relation Diagrams

1
COM 117 Introduction to Application BuildingWeek
2 Entity Relation Diagrams

• Lecture outcomes
• Be able to describe entities
• Be able to establish the relationship between
  entities
• Be able to model and construct ERDs
• Be able to apply the general rules for an E-R
  model

2
Week 2 Entity Relation Diagrams

• Also called Logical Data Structures (LDS)
• A top-down approach
• Start by asking
• Who or what do we need to store data about?
• Results in a useful first attempt but
• may be incomplete model
• may have redundant data
• relationships may be vague

3
Terminology

• entities
• an entity is an object about which the system
  requires to hold data
• e.g. CUSTOMER
• occurrences of entities
• an occurrence describes a particular instance of
  an entity
• e.g. a customer called Mrs Partridge!

4
Entities

• values of entities
• is the value of that attribute for a particular
  entity occurrence
• entity keys
• an attribute or set of attributes that UNIQUELY
  identify an occurrence of an entity
• (Primary key in an Access table!)

5
Relationships between entities

• this is a link between two entities that is
  significant for the system
• VERY IMPORTANT to capture ALL significant
  relationships

6
The Degree of Relationships

• one to one
• A husband has one wife (at a time normally!) and
  vice versa
• one to many
• A GP has many patients but the rules of the
  practice dictate that a patient is registered
  with only one doctor
• many to many
• A musician plays in many concerts and a concert
  is performed by many musicians

7
Notation
8
(No Transcript)
9
Entity Relationship Diagrams

• Further Reading
• Britton Doake
• Software System Documentation, A Gentle
  Introduction
• Pages 59-65

10
E-R Models - Practical Guidelines

• A data model should exhibit the following
  characteristics
• data grouped logically
• each group should have a primary key
• no redundant data

11
No redundant data

• data never used by system
• a data item stored in more than one place
• data in one place can be derived from data held
  elsewhere

12
To summarise

• put data items in logical groups that go together
• ensure that a data item needs to be stored, i.e.
  cannot be derived
• ensure that each data item is stored only once
• ensure that there is enough information to
  identify a particular instance of a data group

13
How to Build an E-R Model

• Look at system and identify data objects
  (entities)
• For each entity, list its attributes
• For each entity, check the attributes for
  completeness
• Investigate and record relationships
• Check entity descriptions against data dictionary
  descriptions of data stores, inputs, outputs,
  data flows etc.

14
In short -

• Establish every entity and possible
  relationship
• Rationalise many-to-many relationships
• Delete duplicated
• relationships

15
General Rules for an E-R Model

• Each entity is represented by a box
• A line connecting boxes represents an entity
  relationship
• A crows foot identifies which is the detail
  (many) entity
• Each occurrence of a master can own several
  occurrences of detail
• For any given relationship, a detail may only
  have one master
• A master may exist without details

16
More rules

• Unless information is lost by removing them,
  entities should not be connected by more than one
  relationship route
• Relationship lines must not cross
• Avoid dead crows!

17
Stage 1
18
Stage 2
19
Stage 3