A Dynamic Binary Group Approach for Fast Mining Frequent Closed Itemsets

摘要

Frequent closed itemsets play an important role in pruning redundant rules fast. A lot of algorithms for mining FCI by vertical data formats have been developed. Previous methods often consume more memory for storage Bit-Vectors and the time for computing the intersection among Bit-Vectors. In this paper, it uses Dynamic Binary Group for compress the database and generates DBG-lattice for mining FCI. Advantages of this method are fast computing the support and the intersection of two DBGs. Experimental results show the efficient of this method in both the mining time and memory usage.