Mining Association Rules: A Continuous Incremental Updating Technique

Abstract

A continuous incremental updating technique is proposed for efficient maintenance of the mining association rules when new transaction data are added to a transaction database. FP-growth algorithm can mine the complete set of frequent patterns by pattern fragment growth. To efficient maintenance of the mining association rules, we improve the FP-growth algorithm in three aspects: 1) an optimization technique for reducing the database size during the update process is discussed, and 2) the construction algorithm of a transaction tree T-tree, and 3) the candidate pattern pools are proposed based-on the structure of T-tree. Then, a continuous incremental updating algorithm, or CIU algorithm for short, is proposed. Our performance study shows that the continuous incremental updating technique is efficient and scalable for mining both long and short frequent patterns.