A Comparison of Association Rule Mining Methods for XML Data

Abstract

With the increasing use of XML technology for data storage and data exchange, mining XML documents has become a researchable subject. This study makes a comparison of three association rule mining methods, the Apriori, FP-growth and AprioriSP, for mining XML data directly with using TinyXML, which is a XML parser implemented in the C++ language. The AprioriSP is proposed to find only short pattern frequent itemsets instead of all frequent ones in order to improve mining efficiency. Comparison experiments of performances of these three methods are performed on XML documents using different datasets and support levels.