Clustering XML Document Based On Path Similarities Using Structure Only

Abstract

We propose a methodology for clustering XML documents on the basis of their structural similarities. This research combines the methods of common XPath and K-means clustering that improve the efficiency for those XML documents with many different structures. The common XPath is used for searching similarities between huge numbers of XML documents’ paths. K-means clustering algorithm is essentially used to accurate clusters. In order to cluster the documents’ paths we indicate the steps by step methods. The first step includes frequent structure mining for searching similarities between the huge amounts of XML documents’ structures by using the F-P growth method. The second step builds dimensional feature vector matrix by using extracted paths. Based on the set of common path vectors collected, we compute the structure similarity between the XML documents. And the last step utilizes the K-means clustering algorithm is used to create accurate clusters which are based on the idea of using path based clustering, which groups the documents according to their common XPaths, i.e. their frequent structures. The quality of clustering can be measured on the dissimilarity of document structures. Also, experimental evaluation performed on both synthetic and real data shows the effectiveness of our approach.