Detecting ambiguity: an example in knowledge evaluation

Expert systems have been developed around one expert partly because the expert has been totally responsible for the soundness of the knowledge base. Without strong aids to help ensure soundness in building expert systems, we must rely on the soundness of the mature cohesiveness of a human expert. As knowledge bases grow this mature expertise will not be adequate. We propose a new method (for expert systems) to aid the expert in knowledge evaluation within the rule-based system setting. We consider the problem of ambiguity within a classification system as an example of the proposed technique.