Explaining control strategies in second generation expert systems

Abstract

Explaining control strategies is an important aspect in second generation expert system explanation. With explicit representation of control, it is possible to construct explanations at different levels of abstraction so as to satisfy different needs. This paper presents the authors, work on explaining control strategies within the context of the system tool, CARMEN. CARMEN's methodology for modeling control knowledge (MMCK) uses problem-solving entities (i.e., tasks, meta-knowledge sources (MKSs), basic-knowledge sources (BKSs) and engines) to integrate different kinds of knowledge and reasoning. These entities reflect the ways experts organize their problem-solving activities. Explanations about control strategies can be clearly given according to the roles these entities play in problem solving. >