A Parallel Algorithm of Frequent Itemsets Mining Based on Bit Matrix

Mining frequent item sets is an important issue in association rules community. This paper proposes a parallel algorithm for mining frequent item sets based on bit matrix. The algorithm reduces the memory space and the I/O overhead, for it scans database only once and builds a compressed bit matrix. It combines both top-down approach and bottom-up approach to improve the efficiency of pruning, and uses dynamic scheduling parallel multi-threaded of OpenMP to mine frequent item sets. The experiments show that this algorithm has higher computing efficiency than Apriori algorithm.