Database and Expert System Technology - Information Systems Reengineering, Integration and Normalization - page 49

2.3.3 Extended Entity Relationship (EER) Model

The ER Model (Chen 1976 ) is a special diagram technique used as a tool for logical
database design. It serves as an informal representation to model the real world by
adopting the more natural view such that the real world consists of entities and rela-
tionships; it also incorporates some important semantic information into the model.
The model can achieve a high degree of data independence and is based on set
theory and relation theory. It can be used as a basis for a unified view of data and a
prelude to designing a conceptual database.
The components of an ER model are:

1. Entity set—An entity set (i.e. entity type) or an entity (i.e., entity instance) is an
important, distinguishable object for an application, e.g., a regular entity, a weak
entity.
2. Entity key—An entity attribute that can uniquely identify an entity instance.
3. Entity attribute—Fields that describe an entity (i.e., properties of an entity).
4. Degree of relationship—The number of entity sets that are related to each other.
(For example, unary means one entity, binary means two entities, ternary means
three entities, and n-ary means n entities related to each other).
5. Cardinality—The connectivity of two entities, i.e., one-to-one, one-to-many, and
many-to-many.
6. Relationship membership—The insertion rules of relationship.
(For example, mandatory means compulsory relationship, optional means not
compulsory relationship).
7. (Minimum, maximum) occurrence—The minimum and maximum instances of
cardinality. (For example, zero minimum occurrence means partial participation
in an optional relationship).

The ER model has been widely used but does have some shortcomings. It is dif-
ficult to represent cases where an entity may have varying attributes dependant
upon some property. For example, one might want to store different information for
different employees dependent upon their role, although there will still be certain
data such as name, job title, and department that remain common to all employees.
Employees who are engineers may require professional qualifications to be stored.
We may need to know the typing speed of typist employees and store the language
spoken by each translator employee.
Due to these limitations, the EER model has been proposed by several authors
(Kozacynski and Lilien 1988 ), although there is no general agreement on what con-
stitutes such a model. Here, we will include in our model the following additions
to the ER model:

• Generalization(Elmasri 1989 )—More than one isa relationship can form data
abstraction (superclass/subclass) among entities. A subclass entity is a subset of
its superclass entity. There are two kinds of generalization. The first is disjoint
generalization such that subclass entities are mutually exclusive, which can be
differentiated by a field in the superclass entity. The second is overlap general-