考虑层数信息的XML文档聚类方法

Clustering XML documents by layer information

提出了一种层数敏感的XML文档数据集聚类方法CXLI。首先提出结构表概念,消除XML文档的重复和嵌套结构。然后提出考虑层数信息的XML文档基本编辑操作约束。进一步给出考虑层数信息的XML文档间相似性度量方法。最后使用凝聚型层次聚类方法对XML文档数据集进行聚类。在ACM SIGMOD数据集和人工生成的数据集上进行了实验验证,结果表明:在计算时间基本相同的情况下,CXLI方法具有更好的精确度。

A layer-sensitive XML document collection clustering method CXLI is proposed in this paper. First, a concept of structural table is put forward to clear up the duplication structures in XML documents. Second, the constraints o{ editing operations are established. Third, a testing method of the similarity between XML documents is presented. Finally, the XML documents are clustered using agglomerative hierarchical clustering method. ACM SIMOD data set and synthetic data set are used to test the proposed method. Results show that the proposed CXLI has better precision under similar time cost.