A framework for the comparative analysis and evaluation of knowledge representation schemes

Abstract

Abstract This article sets forth a framework for comparing and evaluating knowledge representation schemes based on the requirements for “good” representation discussed in extant literature. The dimensions of the framework suggest that knowledge representation schemes should possess a suitable mix of four basic considerations: Representational adequacy (Variety of Expressiveness, Modularity, Semantics, and Organization of Knowledge); Inference Methods (Reasoning Strategies, Data, Control and Search Strategies); and Inference Requirements (Computational Efficiency, Transparency of line of control, Completeness, and Consistency). A comparative analysis and evaluation of four popular knowledge representation schemes—Logic, Production Rules, Semantic Nets, Frames—highlighting their strengths and weaknesses in the context of the framework is furnished as evidence of its inherent validity and usefulness. In conclusion, it is submitted that incorporating an appropriate blend of the various dimensions elucidated in this article could be a step towards developing more flexible, dynamic, and valuable knowledge representation schemes.