An Adaptive Method for Discovering Maximal Frequent Itemsets to Large Databases

A novel adaptive method included two phases for discovering maximal frequent itemsets is proposed. A flexible hybrid search method is given, which exploits key advantages of both the top-down strategy and the bottomup strategy. Information gathered in the bottom-up can be used to prune in the other top-down direction. Some efficient decomposition and pruning strategies are implied, which can reduce the original search space rapidly in the iterations. The compressed bitmap technique is employed in the counting of itemsets support. According to the big space requirement for the saving of intact bitmap, each bit vector is partitioned into some blocks, and hence every bit block is encoded as a shorter symbol. Therefore the original bitmap is impacted efficiently. Experimental and analytical results are presented in the end.