Applying conceptual clustering to knowledge-bases construction

摘要

Concept hierarchy is a common approach to organizing structural knowledge in expert systems because of its efficient mechanism to store and generalize a large body of interrelated concepts. Construction of a concept hierarchy, however, is still an unstructured manual process. This laborious process, which is usually conducted by interviewing with human experts, is generally regarded as the bottleneck in developing an expert system. In this paper, a conceptual clustering approach to automatically generating a concept hierarchy is presented. Details of the automated process are illustrated with an experimental system. Initial findings obtained from this study indicate that: (1) construction of concept hierarchy can be automated; (2) inference making can be accomplished by partial matching; (3) conceptual clustering can be used as a general modeling tool in other management disciplines.