AN INFERENCE BROWSER FOR VERIFYING THE KNOWLEDGE BASE IN KNOWLEDGE-BASED SYSTEMS

In developing knowledge-based systems, the process of collecting the knowledge from the experts, representing it in certain formats, and verifying it is required. It is however not easy to verify the formulated knowledge base by checking if a desired conclusion is derived by a sequence of inferring steps. This paper suggests a model of inference browsers by which the knowledge engineers may easily consult a sequence of inferring steps and verify the knowledge base. The suggested inference browser provides the environment in which the knowledge engineers may observe a sequence of inferring steps displayed in the graphical form, access directly the contents of the rules and the facts on the sequence, and observed a newly generated sequence of inferring steps when some of the rules or the facts are changed. Further, based on the graphically displayed inferring sequence, the inference browser itself detects the erroneous inferring step if it exists, analyzes it, and corrects the associated errors in the knowledge base. Finally the suggested inference browser is compared to other similar tools in terms of the facilities they provide.