Fast mining of association rules in large-scale problems

In this paper we study scalability problem of apriori-like algorithms that are used in mining association rules. We show how apriori suffers form performance deterioration for large-scale problems and propose an alternative data structures and operations that can be used to apply the apriori-trick optimization method in large-scale problems. In the proposed method, the database is transformed into a more efficient structure that is used along with the intersection operation, to find the frequent itemsets in the database. The performance evaluation shows that, with a minor increase in the storage requirement, the proposed technique outperforms significantly the existing algorithms especially in large-scale problems.