Novel measurement for mining effective association rules
Novel measurement for mining effective association rules
复制标题
DOI：
10.1016/j.knosys.2006.05.011
复制
发表时间：
2005-11
期刊：
影响因子：
--
通讯作者：
Jinmao Wei;Wei-Guo Yi;Ming-Yang Wang
中科院分区：
文献类型：
--
作者：
Jinmao Wei;Wei-Guo Yi;Ming-Yang Wang
本文分析了挖掘关联规则时支持置信框架的方法。为了避免标准的限制，我们提出了一种新的匹配方法作为置信度的替代。我们详细分析了所提出的测量的特性。实验结果表明，与支持置信框架生成的规则相比，改进方法生成的规则的前件和后件之间具有更高的相关性。此外，改进的方法减少了冗余规则的生成。
In this paper, we analyze the method of support-confidence framework when mining association rules. In order to avoid the limitation in the criterion, we propose a new method of match as the substitution of confidence. We analyze in detail the property of the proposed measurement. Experimental results show that there is higher correlation between the antecedent and the consequent of the rules produced by the improved method compared with the rules produced by the support-confidence framework. Furthermore, the improved method decreases the generation of redundancy rules.