Abstract: While mining generalizing association rules with negative items，the search space becomes more significant because of the introduction of negative items，and the number of negative association rules also becomes greater.But many of them are uninteresting，even wrong.Therefore a concept named maximum support is introduced，which is used to bind the frequent items mined，and exclude meaningless association rules.It can improve the efficiency at the same time.Respective minimum support thresholds are used for positive and negative items，making mining more flexible.Experiments indicate that the improvement is valid.

Key words: association rules, negative association rules, negative items, maximum support

摘要： 在含负项目的一般化关联规则的挖掘中，由于负项目的引入使得频繁项集的搜索空间变得更加巨大，而同时挖掘出的关联规则数量也随之增大，但其中很多规则对用户来说是不感兴趣的，而且可能包含一些冗余和错误的规则。因此提出了最大支持度的概念，用来约束频繁项集的挖掘，排除没有意义的关联规则同时也提高了挖掘的效率。在挖掘中对正负项目分别采用不同的最小支持度，使得挖掘更加灵活。并通过实验证明改进是行之有效的。

关键词: 关联规则, 负关联规则, 负项目, 最大支持度