Database and Expert System Technology - Information Systems Reengineering, Integration and Normalization - page 77

pare descriptions of incoming stimuli with frames in the knowledge base, and
retrieve the class frame that best matches the situation.
• Inheritance
The matching retrieves the relevant frame that contains general information for
the reasoning process and applies inherited information to specific information.
The basic inheritance mechanism uses member links, sub-class links, and pro-
totype descriptions of class members to assert and retrieve the specific informa-
tion.
• InstanceFrameReasoning
The inheritance reasoning infers the frame by using the “is_a” link and the in-
stance frame reasoning infers the frame through their “a_part_of” link. It is a
mechanism to retrieve specific information for a slot with instance frame values.
• ProceduralReasoning
This is a mechanism to retrieve specific information for a slot with procedure
values or to perform constraint and integrity checking by the use of demons. The
technique includes sending a message to an object-oriented method or perform-
ing an external call in order to run a normal routine (e.g., calling standard func-
tions in LISP).
• CardinalityandConstraintChecking

A frame-based system considers cardinality, default, and restriction specifica-
tions as constraints on the legal values of a slot. The system provides constraint
checking procedures for determining whether a slot's value is valid.

Currently most frame-based representation facilities also provide a convenient
rule-based management facility. There are usually two ways to combine rules and
frames. One is to attach a production rule language to the frame-based system,
such as in GoldWorks (Casey 1989 ). The frame facility supplies an expressively
powerful language for describing the objects being reasoned about and automati-
cally performs a useful set of inferences on those descriptions. The other involves
representing a rule as a frame, such as in KEE (Fikes and Kehler 1985 ). KEE allows
production rules to be represented by frames so that they can easily be classified
into taxonomies, created, analyzed, and modified as necessary.
Several advantages have been claimed for frame-based knowledge representa-
tion schemes. Many of these advantages involve the representation of stereotypes
and assertion clustering, which improves access to knowledge by storing associated
representations together. It is expected that this technique will become common in
the future, particularly in large and sophisticated expert systems.

2.7 Summary

Database systems and expert systems are the major components of information
systems. The legacy data models include hierarchical, network, relational, object-
oriented, and XML. The hierarchical model has an inverted tree structure data struc-