Mining Non-Redundant Substitution Rules Between Sets of Items in Large Databases

Yi-Chun Chen ； Guan-Ling Lee

《Journal of Information Science and Engineering》 31卷2期 (2015/03) Pp. 659-674

https://doi.org/10.6688/JISE.2015.31.2.16

並列摘要

The mining of association rules has been studied for years, but few studies have considered the mining of substitution rules though such rules also provide valuable knowledge of market prediction. In prior work, several efficient mining algorithms were proposed to explore substitution rules. However, all the methods may produce redundant substitution rules. Therefore, this paper discusses the problem of mining non-redundant substitution rules and proposes the NRM (non-redundant substitution rules mining) algorithm as a solution. A substitution rule is said to be non-redundant if the provided information is not covered by other rules. To make the mining process more efficient, the property of frequent closed patterns is utilized and three lemmas are proposed to prune redundant substitution rules. Our experimental results show that the performance of NRM algorithm is superior to that of naïve apriori-based approaches

並列關鍵字

non-redundant substitution rule ； negative association rule ； correlation analysis ； data mining ； frequent closed pattern