Mining Frequent User Query Patterns from XML Query Streams

An XML query stream is a massive and unbounded sequence of queries that are continuously generated at a fast speed from users over the Internet. Compared with traditional approaches of mining frequent user query patterns in static XML query databases, pattern mining in XML query streams is more challenging since several extra requirements need to be satisfied. In this paper, a mining algorithm is proposed to discover frequent user query patterns over an XML query stream. Unlike most of existing algorithms, the proposed algorithm works based on a novel encoding scheme. Through the scheme, only the leaf nodes of XML query trees are considered in the system and result in higher mining performance. The performance of the proposed algorithm is tested and analyzed through a series of experiments. These experiment results show that the XSM outperforms other algorithms in its execution time.