A Method of Characterizing Document Clusters

摘要

A conceptual method of characterizing document clusters is presented for clarifying what the given document clusters intend to express. Keyword frequency based approaches are insufficient in clarifying the document contents, since the keywords themselves usually have some semantic relations. We here try to attack this clarification problem by formulating a method of finding the generalized expressions for the clusters. Computational experiments comparing the proposed method with the frequency-based approaches are also given.