Users of Web search engines are often forced to sift through the long ordered list of document “snippets” returned by the engines. An interactive partially clustering method is put forword in this paper. First, PCCS uses the clustering algorithm to cluster part of the documents, finds highly accurate cluster digests (partial clusters), gets user feedback to merge and correct these digests, and then uses the Na l ¨ ve-Bayes classification algorithm to classify the rest documents. The incremental classification model can be saved and then be used to help classify future Web query results. In order to improve the efficiency of the method, a hybrid feature selection is also proposed to reduce dimension of document vector: Entropy feature selection and classification model based feature selection. It is shown that the method is faster than other algorithms. PCCS helps users more quickly and efficiently to navigate the results of a query at a more topical level than having to examine each documents text separately.%PCCS是为了帮助Web用户从搜索引擎所返回的大量文档片断中筛选出自己所需要的文档，而使用的一种对Web文档进行快速聚类的部分聚类方法：首先对一部分文档进行聚类，然后根据聚类结果形成分类模型对其余的文档进行分类.采用交互式的一次改进一个聚类摘选的聚类方法快速地创建一个聚类摘选集，将其余的文档使用Nal ¨ ve-Bayes分类器进行划分.为了提高聚类与分类的效率，提出了一种混合特征选取方法以减少文档表示的维数：重新计算文档中各特征的熵，从中选取具有最大熵值的前若干个特征；或者基于持久分类模型中的特征集来进行特征选取.实验证明，部分聚类方法能够快速、准确地根据文档主题内容组织Web文档，使用户在更高的主题层次上来查看搜索引擎返回的结果，从以主题相似的文档所形成的集簇中选择相关文档.
展开▼
