文章基本信息

标题：An Efficient Top-Down Mining of Negative Association Rules
本地全文：下载
作者：Noriko Ide ; Koji Iwanuma ; Yoshitaka Yamamoto 等
期刊名称：人工知能学会論文誌
印刷版ISSN：1346-0714
电子版ISSN：1346-8030
出版年度：2014
卷号：29
期号：4
页码：406-415
DOI：10.1527/tjsai.29.406
出版社：The Japanese Society for Artificial Intelligence
摘要：Negative association rules represent some relationships between presence and absence of itemsets. In general, the number of negative association rules is enormously huge even if compared with that of positive association rules. Therefore, an efficient mining method is quite important. In this paper, we propose a novel top-down mining method for negative association rules in the forms of X ⇒￢ Y and ￢ X ⇒ Y . The proposed method search a suffix tree over frequent itemsets in a top-down manner, and efficiently extract all of valid negative rules of these two types, step by step. The suffix tree plays very important roles for effectively pruning a lot of redundant searches such as the one producing non-minimal valid negative rules. We also show some good results of experiments for evaluating our proposed method. The proposed method is given for simple negative rule mining based on the support and confidence measures, therefore is definitely the most fundamental and important framework, into which additional measures can be easily introduced if necessary.
关键词：negative association rule ; mining ; top-down search ; suffix tree