Abstract:

An improved Apriori algorithm for association rule mining called Apriori-BR was proposed,which was based on bit operation and reverse index.Specifically,the reverse index from frequent itemsets to transactions was constructed firstly by scanning twice of database,and the reverse index was grouped by the length of transactions.Then in the mining process,bit operation was adopted to accelerate subset detection together with the dynamical elimination of invalid low-dimensional transactions.The numerical results show that the Apriori-BR proposed can substantially improve mining efficiency when compared with the conventional Apriori algorithm and the improved ones in the literature.

Key words: Apriori algorithm, bit operation, reverse index