Evaluation of Knowledge Representation Schemes as a Prerequisite toward Development of a Knowledge-Based System

Abstract

In developing a knowledge-based system, it is crucial to use an appropriate knowledge representation scheme (KRS) for efficient working of the developed system. However, it is difficult to select an appropriate KRS due to lack of a generalized method of comparison among the variety of schemes available. The objective of this paper is to discuss a generalized method for comparative evaluation of KRS, based on criteria of expressiveness and performance. The discussed method has been implemented for evaluation of four widely used KRSs: rule-based model, object-based model, relational model, and hybrid model. Evaluation is being carried out using a knowledge base for zonation of landslide hazard. Based on the criteria of expressiveness and performance, it has been found that hybrid-based KRS is best for representation of the domain knowledge available in the Indian Standard Code 14496-1998 (Part-II).