Abstract: Clustering technique is based on the conception that never replacing the cluster representatives selected randomly,partitioning the objects based on the semantic correlation principle, merging the clusters having much big semantic correlation,decomposing less semantic correlation clusters, reinventing the wheel for the objects when a round of clustering processs finished ,they have no end-result.

摘要： 采用的聚类思想是,不替换随机选取的聚类代表,按语义相关的原则界定对象,合并相似度较大的聚类,分解稀疏聚类,对未有归宿的对象再给机会聚类。

关键词: Web挖掘, 模糊聚类, 聚类合并, 聚类分解

CLC Number: 

• TP393.07