Abstract: A fast hybrid clustering algorithm for Web documents clustering is proposed which optimizes the initial center values of K-means algorithm through STC algorithm.Firstly，the initial center values are extracted after the Web document set is clustered by STC algorithm.Secondly，by mapping the each internal node of suffix tree into M-dimensional VSM，each feature term weights is computed using TF-IDF extended with phrases.Finally，the final result is generated by K-means algorithm.The evaluation experiments indicate that the new hybrid algorithm is more effective on clustering documents than ordinary K-means and STC algorithm.Moreover，it is as fast as K-means and STC algorithm.

摘要： 提出了一种使用后缀树聚类算法优化K-means文档聚类初始值的快速混合聚类方法STK-means。该方法首先构建文档集的后缀树模型，使用后缀树聚类算法识别初始聚类、提取K-means聚类算法初始值中心值。然后，把后缀树模型的节点映射到M维向量空间模型中的特征项，利用TF-IDF方案计算基于短语的文档向量特征值。最后，使用K-means算法产生聚类结果。实验结果表明该方法优于传统K-means聚类算法和后缀树聚类算法，并具备了这些算法聚类速度快的优点。