Automated Construction of Classifications: Conceptual Clustering Versus Numerical Taxonomy

摘要: A method for automated construction of classifications called conceptual clustering is described and compared to methods used in numerical taxonomy. This arranges objects into classes representing certain descriptive concepts, rather than defined solely by a similarity metric some priori attribute space. specific form the conjunctive clustering, which concepts are statements involving relations on selected object attributes optimized according an assumed global criterion quality. The method, implemented program CLUSTER/2, tested together with 18 taxonomy two exemplary problems: 1) classification popular microcomputers 2) reconstruction plant disease categories. In both experiments, majority (14 out 18) produced results were difficult interpret seemed be arbitrary. contrast this, that had simple interpretation corresponded well solutions preferred people.