期刊名称：International Journal of Computer Science and Information Technologies
电子版ISSN：0975-9646
出版年度：2014
卷号：5
期号：2
页码：1393-1397
出版社：TechScience Publications
摘要：In association rule mining, frequency of an itemset alone does not assure its interestingness because it does not contain information on subjectively defined utility such as profit in rupees or some other variety of utility. This leads to a fruitful deviation in the frequent itemset mining called utility based data mining. Mining high utility itemsets upgrades the standard frequent itemset mining framework. Fast Utility Frequent Mining (FUFM) is a popular algorithm to find utility frequent itemsets (UFIS) based on the extended support measure. But, when applying FUFM on transactional databases with highly fluctuated transaction utility values, it generates less number of UFIS. This is due to the effect of transactions with low utility values. In this paper, an algorithm Fast Value-Added Utility Frequent Mining (FVAUFM) is proposed to find UFIS. FVAUFM works on the efficient and reduced database got by removing the transactions below a threshold value from the given transactional database. The proposed algorithm FVAUFM uses a novel measure called average-utility measure to find UFIS. Experiments show that FVAUFM generates more number of UFIS than FUFM and also saves a significant amount of memory space and running time
关键词：Utility frequent itemsets; Association rule mining;Apriori algorithm
