DT: a classification problem solver with tabular-knowledge acquisition

Abstract

An approach is described to integrating a classification problem solving function and an interactive knowledge acquisition function on a tabular knowledge representation. A tabular knowledge base presents concept functions in DNF (disjunctive normal form), which handles disjunctive concepts with multiple-value attributes. Because of the peculiar knowledge-base construction, which represents concept functions with positive and negative disjuncts, the knowledge refinement procedures are kept simple, and changes in the knowledge base are comprehensible to users. Based on the approach, a system called DT was developed.<>