SYSTEM FOR EXTRACTING KNOWLEDGE OF TYPICALITY AND EXCEPTIONALITY FROM A DATABASE OF CASE RECORDS

摘要

A knowledge tree building system iteratively partitions a database(28a-28n) of case records into a tree of conceptually meaningful clusters(74).Each cluster (74) is automatically assigned a unique conceptual meaning inaccordance with its unique pattern of typicality and exceptionality within theknowledge tree (28); no prior domain-dependent knowledge is required. Thesystem fully utilizes all available quantitative and qualitative case recordda-ta. Knowledge trees built by the system are particularly well suited forartifi-cial intelligence applications such as pattern classification and nonmonoton-ic reasoning.