An efficient strategy for mining high utility itemsets

Abstract

Methods for mining high utility itemsets from databases have been discussed widely in recent years. They mine itemsets having high utility from databases. Pruning candidates based on transaction-weighted utilisation value is a good method at all. In this paper, we develop a tree structure called WIT-tree, and use it in the proposed TWU-mining algorithm, an algorithm for improving the mining time and the search space. Using Diffset for fast computing transaction-weighted utilisation values and saving memory will be discussed. Experimental results show that proposed algorithms are more efficient. © 2011 Inderscience Enterprises Ltd.