Mining Incrementally Closed Itemsets with a New Intermediate Structure

Abstract

The problem of closed frequent itemset discovery is a fundamental issue of data mining, having applications in numerous domains. Until now, the general technic for incremental mining is using an intermediate structure in order to update the structure whenever there is a variation in the data. As for incremental mining closed itemsets, the intermediate structure used is a concept lattice. The concept lattice promotes the efficiency of the search process, but it is costly to adjust the lattice when there is an addition or removal, as well as it is difficult in developing parallelization strategy. This article proposes incremental algorithms to search all closed itemsets with a new intermediate structure which is a linear list. To the best of our knowledge, this is the first algorithm for incremental mining closed itemsets using a linear list as an intermediate structure proposed so far. When comparing experimental results between using intermediate structure concept lattice and linear list initially show that the greater number of transactions and the number of closed itemsets obtained in the mining process, the more efficient the use of linear list promotes.