Closed High Utility Pattern Mining over Data Stream Based on Projection in the Window

Abstract

A fast and effective algorithm EFIM_Closed_DS was proposed to mine closed and high utility itemsets in the data stream environment. The algorithm is based on the projection technology in the window, and the database projection technology and transaction merging method are used in each window to effectively reduce the cost of database scanning. Using efficient pruning techniques and effective closed itemset detection methods can prune a large number of low-utility itemsets and non-closed itemsets. A large number of experimental results show that the proposed algorithm is more effective in memory and time than previous state-of-the-art algorithm CHUI_DS.

Keywords

• pattern mining
• data stream
• closed pattern
• high utility pattern mining
• projection in the window