基于兴趣度的关联规则挖掘

关联规则挖掘是数据挖掘领域中的重要研究内容之一。然而,传统的基于支持度-可信度框架的挖掘方法可能会产生大量不相关、甚至是误导的关联规则。针对现有关联规则挖掘的评价标准存在的问题,提出在评价标准中增加兴趣度,并给出了兴趣度的定义和基于兴趣度的关联规则挖掘算法。利用兴趣度将关联规则分为正关联规则和负关联规则,从而可以用算法挖掘带有负项的关联规则。实验结果分析表明,在传统挖掘方法的基础上引入兴趣度,可以有效地减少正关联规则的规模,产生有意义的负关联规则。

Mining Association Rules Based on Interest Measure

Mining of association rules is an important research topic among the various data mining problems.However the common approaches based on support-confidence framework maybe get a great number of redundant and wrong association rules.In order to solve the problems,an interest measure is defined and added to the mining algorithm for association rules.According to the value of interest measure,association rules are classified into positive and negative association rules.The new algorithm can find out the negative-item-contained rules.The experimental result shows that introducing interest measure based on common approach to association rules mining can reduce the scale of positive association rules,and mine a lot of interesting negative association rules.