关联规则评价指标的研究

The Research of Evaluation Index in Association Rules

关联规则挖掘是数据挖掘中重要的研究课题,对于如何从海量的信息中挖掘出有效的、可信的、可理解的、感兴趣的关联规则来帮助人们进行分析与决策,已经成为迫切需要解决的内容。现有的关联规则评价标准除了不能很好地满足用户需求外,还存在着含义和分类的不清晰性。本文在综合分析了现有的评价指标的基础上,提出了关联规则评价指标体系结构,明确了各评价指标的含义,并从系统论的角度将评价指标划分为基本评价指标、定量评价指标和定性评价指标三类,以帮助人们在应用评价时参考与使用。

Association rule is a capital research orientation in data mining. How to mine effective,believable,intelligible and interested association rules from vast information for helping people analyze and make decision, which has become important question. At present, the existing evaluation standard not only dissatisfies people very well, but also has unclear meanings and classification. On the basis of analyzing the existing evaluation index synthetically, this paper puts forward the architecture of evaluation index, makes clear the meanings of evaluation index and classifies the evaluation index to elementary evaluation index, quantitative evaluation index and qualitative evaluation index, which helps people refer and apply in evaluation of association rules.