An inductive learning strategy for automated knowledge acquisition based on concept rule

A new approach for an automated knowledge acquisition technique through conceptual clustering of examples and derivation of a concept rule from these clusterings is described. This rule derivation is based on the physical observations that if some attributes of an example set are similar, then there must exist one or more examples consisting of the rest of the attributes of the example set. The concept rules derived by this process from each cluster set together with the initial example set are equivalent to a new example set, plus the original example set. The approach learns by induction and by discovery, as the process generates new rules which infer new facts. This type of learning system provides a means for an improved automated knowledge acquisition process with enhanced inferencing capability and bypasses the knowledge engineer as the facilitator and intermediary in expert system applications