Improve Tree Search For High Utility Itemset Mining On Incremental Database

Abstract

High-utility itemset mining (HUIM) has emerged as a crucial study area in recent decades, as both quantity and profit variables are important when mining high-utility itemsets (HUIs). Ever since then, there have been several similar algorithms proposed. In 2012, Liu and Qu presented the HUI-Miner method (High Utility Itemset Miner) to exploit high utility itemsets by storing all valuable information about an itemset in a new structure called a utility list. The HUI-Miner method is believed to be an effective technique for extracting high utility itemsets. Later, the FHM method used a utility list structure to minimize search space. However, the database is always evolving since fresh data is collected regularly, coal... To leverage HUI on a growing database, an efficient technique is required. The EIHI method presented in 2015 is a very efficient item-mining technique applicable to the growing database, the algorithm used a tree-typed data structure to perform storing and updating whenever a new itemset is founded, this process can cause certain drawbacks in the algorithm, especially when dealing with big data. In this paper, I will focus on the procedure of organizing itemset during the mining of the EIHI algorithm and suggest a better data structure to improve the performance of EIHI algorithm.