A New Algorithm of Mining High Utility Sequential Pattern in Streaming Data

Abstract

High utility sequential pattern (HUSP) mining has emerged as a novel topic in data mining, its computational complexity increases compared to frequent sequences mining and high utility itemsets mining. A number of algorithms have been proposed to solve such problem, but they mainly focus on mining HUSP in static databases and do not take streaming data into account, where unbounded data come continuously and often at a high speed. The efficiency of mining algorithms is still the main research topic in this field. In view of this, this paper proposes an efficient HUSP mining algorithm named HUSP-UT (utility on Tail Tree) based on tree structure over data stream. Substantial experiments on real datasets show that HUSP-UT identifies high utility sequences efficiently. Comparing with the state-of-the-art algorithm HUSP-Stream (HUSP mining over data streams) in our experiments, the proposed HUSP-UT outperformed its counterpart significantly, especially for time efficiency, which was up to 1 order of magnitude faster on some datasets.