A Survey of High-utility Itemsets Mining

Abstract

Data mining is of significance for finding useful information in massive data. Frequent itemsets mining (FIM ) and high-utility itemsets mining(HUIM) are extremely common and wide application in research and real life. For one thing, HUIM algorithm focuses on utility, which is more practical. It can be used to find high profit goods, items with user’s preference, etc. For another, the difference between utility and frequency determines that HUIM and FIM algorithms are different. In order to introduce HUIM algorithms in the round, this paper showed typical HUIM algorithms for static data and stream data separately in section 2 and section 3. Meanwhile, section 2 partitioned algorithms based on candidates generation and threshold. Section 3 showed algorithms in terms of window model which is necessary to stream data mining. Lastly, this paper made a conclusion of referred HUIM algorithms and proposed some research prospects for this work.