Research on algorithm of user query frequent itemsets mining

An algorithm of mining database user query profiles of transaction level is presented. The algorithm changes the computing method of the support and confidence in association rules mining by adding query structure and attribute relations to the computation. Since there is no causal relationship in the access of attributes in queries, the method is more appropriate to describe user query behaviors than itemsets used by association rules. The method can be used in database intrusion detection system to prevent database from illegal intrusions effectively. Realization, performance and application of the algorithm are discussed in the paper.