A Hybrid Method for XML Clustering by Structure and Content

Abstract

An effective XML cluster method called neighbor center clustering algorithm (NCC) is presented in this paper, whose similarity is obtained through both structural and content information contained in XML files. Structural similarity is firstly measured by frequency-path model and its similarity calculation algorithm with position and frequency weight by longest common subsequence is introduced. In order to improve the performance and precision, the frequency-path model is further extended by considering the structure and content information simultaneously. Experiments show that the NCC embed with hybrid similarity calculation method can obtain high purity and F-measure value and is effective and applicable for clustering XML with both homogenous and heterogeneous structures.