基于聚簇的XML文档近似连接方法

XML文档近似连接操作是在两个XML文档集合中发现近似的XML文档,其在基于XML数据的信息集成、XML数据清洗等系统中有着广泛的应用.然而,目前XML文档近似连接操作的一个显著问题在于:当文档之间存在较大差异时,存在大量的重复计算,降低了处理效率.对于这个问题,提出了基于聚类的XML文档近似连接方法,基本思想是为每个XML文档建立一个索引,如果两个数据集中若干文档的索引较相似,可以把它们组成一簇,然后在每一簇中执行近似连接.而不在任何簇中的文档,则无需对其进行任何计算.实验结果表明,提出的方法在保证正确率的前提下具有高效性.

Clustering-Based Approximate Join Method on XML Documents

Approximate join on XML document sets is to find the similar documents between two document sets.It is widely used in many applications,such as XML-based data integration,XML data cleaning.Currently,some techniques have been proposed to perform approximate join on XML documents.However,a significant problem in current techniques is that the large differences among the documents result in many redundant computations.In this paper,a clustering-based approximate join method on XML documents is presented.Firstly,an index is built for each XML document tree.If the indecies of some documents are similar,these document trees are clustered into one cluster.Then the approximate join is performed on the docments in each cluster.If a tree is not in any cluster,the computation of its windowed pq-grams will be avoided.Experimental results show that the method in this paper outperforms existing algorithms in efficiency without accuracy loss.