• Technique for examining the structure of data (ignoring the physical detail).

• Notation for representing Entities and Relationships.

• Entity: an Object of interest, which we wish to record information about. Entities are uniquely identifiable.

• Entity is tangible or conceptual or active or permanent, or volatile.

• Entity has attributes. Attributes have ranges of permitted values.

• Relationships: logical association between two entities

Cardinality

Optionality

Names

Check that you understand the diagrams used to represent each concept.

How to derive the Logical Data Model

1. Put entities with the most relationships in the centre of the diagram.
2. Put entities with no masters at the top.
3. Let hierarchies cascade down the page.
4. Avoid crossing lines as they make the diagram difficult to read.
5. Name both ends of relationships on the model.

• Check to see that all relationships necessary to modelling the area are present.
• Remove redundant relationships. Where there is more than one way to link a set of entities.
• Finally check the model against other models created in analysis. We will see later Data Flow Diagrams, Data Stores and Processes.

Documentation of LDM - headings:

Entity Description:

Name

Description

Synonyms (alias)

Attributes (name, keys)

Relationships (optionality, name, cardinality, end entity)

Occurrences

User roles and access rights (read, write etc.)

Archive and destruction

Security

State indicator values

Relationship Description:

Name

Link Phrase

Description

Synonyms (alias)

Entity

User roles and access rights (read, write etc.)

% optional

Min, max, average

Growth per period

Attribute/data item Description:

Name

Description

Synonyms

Validation/derivation

Owner

User roles and access rights (read, write etc.)

Summary of Logical Data Modelling

1. LDM provides a view of the underlying structure of the data in the system being analysed.
2. It tends to be developed in parallel with the data flow diagrams, which we will study next. One approach can validate the other.
3. LDM is refined later by relational data analysis (normalisation)
4. Diagrams need to be supported by documentation describing entities attributes, domains.