点击次数：

论文类型：期刊论文

发表时间：2012-10-01

发表刊物：Journal of Information and Computational Science

收录刊物：EI、Scopus

卷号：9

期号：11

页面范围：3189-3199

ISSN号：15487741

摘要：Utility itemsets mining is an extension of frequent itemsets mining, considering both the quantities of items in the transactions and the profits of the items. In traditional utility itemsets mining, the utility of an itemset is the summation of the utilities of the itemset in all the transactions containing this itemset while ignoring the length of the itemset and will increase along with an increase in its length. To eliminate the effect of the length of an itemset, average utility measurement was proposed, and average utility of an itemset was defined as the total utility of the itemset divided by its length. In this article, we propose a pattern-wise algorithm to mine high average utility itemsets. Experimental results show that the proposed algorithm outperforms the existing HAUP-mine algorithm. ? 2012 Binary Information Press.