| 摘要 |
<P>PROBLEM TO BE SOLVED: To achieve an appropriate clustering result. Ž<P>SOLUTION: The program for document clustering first clusters a collection by using the keyword extracted from the document collection as a keyword whose appearance frequency is relatively high, and computes the evaluation value of each cluster. After the computation of the evaluation value, the selection of the keyword used for clustering is changed. Following that, the collection is clustered by using the keyword after the change, and the evaluation value of each cluster is computed. Next, the evaluation value after the change is compared with the evaluation value before the change. If the evaluation value after the change is higher than that before the change, the selection of the keyword used for clustering is changed. Ž<P>COPYRIGHT: (C)2010,JPO&INPIT Ž
|
