A Unified Framework for Utility Based Measures for Mining Itemsets

Abstract

A pattern is of utility to a person if its use by that per- son contributes to reaching a goal. Utility based measures use the utilities of the patterns to re∞ect the user's goals. In this paper, we flrst review utility based measures for itemset mining. Then, we present a unifled framework for incorpo- rating several utility based measures into the data mining process by deflning a unifled utility function. Next, within this framework, we summary the mathematical properties of utility based measures that will allow the time and space costs of the itemset mining algorithm to be reduced.