Generalization Hierarchies

Up to this point, we have discussed describing an object, the entity, by its shared characteristics, the attributes. For example, we can characterize an employee by their employee id, name, job title, and skill set.

Another method of characterizing entities is by both similarities and differences. For example, suppose an organization categorizes the work it does into internal and external projects. Internal projects are done on behalf of some unit within the organization. External projects are done for entities outside of the organization. We can recognize that both types of projects are similar in that each involves work done by employees of the organization within a given schedule. Yet we also recognize that there are differences between them. External projects have unique attributes, such as a customer identifier and the fee charged to the customer. This process of categorizing entities by their similarities and differences is known as generalization.

Description

A generalization hierarchy is a structured grouping of entities that share common attributes. It is a powerful and widely used method for representing common characteristics among entities while preserving their differences. It is the relationship between an entity and one or more refined versions. The entity being refined is called the supertype and each refined version is called the subtype. The general form for a generalization hierarchy is shown in Figure 8.1.

Generalization hierarchies should be used when (1) a large number of entities appear to be of the same type, (2) attributes are repeated for multiple entities, or (3) the model is continually evolving. Generalization hierarchies improve the stability of the model by allowing changes to be made only to those entities germane to the change and simplify the model by reducing the number of entities in the model.

Creating a Generalization Hierarchy

To construct a generalization hierarchy, all common attributes are assigned to the supertype. The supertype is also assigned an attribute, called a discriminator, whose values identify the categories of the subtypes. Attributes unique to a category, are assigned to the appropriate subtype. Each subtype also inherits the primary key of the supertype. Subtypes that have only a primary key should be eliminated. Subtypes are related to the supertypes through a one-to-one relationship.

Types of Hierarchies

A generalization hierarchy can either be overlapping or disjoint. In an overlapping hierarchy an entity instance can be part of multiple subtypes. For example, to represent people at a university you have identified the supertype entity PERSON which has three subtypes, FACULTY, STAFF, and STUDENT. It is quite possible for an individual to be in more than one subtype, a staff member who is also registered as a student, for example.

In a disjoint hierarchy, an entity instance can be in only one subtype. For example, the entity EMPLOYEE, may have two subtypes, CLASSIFIED and WAGES. An employee may be one type or the other but not both. Figure 1 shows A) overlapping and B) disjoint generalization hierarchy.

Figure 1: Examples of Generalization Hierarchies

Rules

The primary rule of generalization hierarchies is that each instance of the supertype entity must appear in at least one subtype; likewise, an instance of the subtype must appear in the supertype.

Subtypes can be a part of only one generalization hierarchy. That is, a subtype can not be related to more than one supertype. However, generalization hierarchies may be nested by having the subtype of one hierarchy be the supertype for another.

Subtypes may be the parent entity in a relationship but not the child. If this were allowed, the subtype would inherit two primary keys.

Summary

Generalization hierarchies are a structure that enables the modeler to represent entities that share common characteristics but also have differences.

The next and final step in the modeling process is to Add Data Integrity Rules.