Practical Guidelines for Supertype and Subtype Modeling - High-Performance Web Databases: Design, Development, and Deployment - page 322

situation it is dangerous for an analyst to treat the subtype as exhaustive, but
the analyst may be tempted to do so because of the scope of the known list.
As a result, application systems may be developed that are highly oriented
around the subtype and require major enhancements each time a new sub-
type is introduced.
PRACTICAL SOLUTIONS
Even with the existence of diagrammatic complexity, subtype modeling
can be a viable tool in every modeling project. The less committed data
modeler, however, may tend to abbreviate the modeling effort and include
only rudimentary subtype modeling and not the level of detail and preci-
sion that is implied by the data itself. Guidelines are therefore needed to
make necessary subtype modeling a practical option for all modelers and
to promote consistency in the development of subtype schemas.
Goals
To develop practical strategies, subtype modeling must be viewed as a
means to an end so that its goals can be better understood. If the data
administrator understands why the data model should contain subtypes,
then how to develop subtype models becomes clearer. Subtype modeling
does not exist to articulate a list of the classes into which instances of an
entity can be placed. This approach forces analysts to continually invent
subtype schemas that add no real value to the model, yet contribute to the
diagrammatic complexity that may prevent successful subtype modeling.
The purpose of subtype modeling is to define the detailed business con-
straints that are placed on the facts that are represented in the supertype-
based model. This is the means to an end approach. The classes that an
instance of an entity may fall into are listed explicitly. This helps the data
modeler understand the differences in how the data that relates to the
instance is presented and used. If no differences are discerned, the subtype
is not useful. This approach results in a single goal that must be satisfied to
include a subtype entity in a model.
This goal states that a subtype entity may be included in a data model if
and only if it contains at least one property (attribute or outbound relation-
ship) with a cardinality different than that defined when the property is cre-
ated for the supertype entity directly. This goal reduces the number of
superfluous subtype entities that are defined on a typical project and
ensures that the defined entities continue to provide their intended value.
The Entity Constraint Matrix
Another step that can be taken to develop practical subtype modeling
strategies is to remove entity-relationship diagramming complexity as a
central issue. Subtype modeling problems often concern the excessive