摘要：

针对关联规则数量巨大并且存在极大冗余的问题，提出无冗余告警关联规则产生算法。从规则本身的关联性和等效性上定义规则的冗余性，通过构造频繁闭项集邻接图发现无冗余告警关联规则，用户可以通过发现的规则推导出其余所有冗余规则，并且得到用户所需的规则形式。该方法不仅能够减少关联规则数量，而且不会带来规则丢失。仿真结果表明，在相同的数据集、最小支持度门限和最小置信度门限条件下，无冗余关联规则数量和产生时间都小于冗余关联规则数量和产生时间，支持度门限越小，差距越显著。

关键词: 无冗余规则, 频繁闭项集, 邻接图, 最小生成项集

Abstract:

Non-redundant association rules mining algorithm is proposed to deal with the problem of huge rules’ number and redundancy. Redundancy is defined by relationships and equivalences among rules. Adjacent graph of frequent closed itemsets is constructed to find non-redundant rules. The user can deduce all rules by found rules and can get required rules’ form easily. The method can reduce rules’ number with no message lose. Results show that with same datasets, minimum support threshold and minimum confidence threshold, the number and generation time of non-redundant rules are less than those of redundant rules remarkably. When minimum support threshold becomes lower, the difference becomes more evident.

Key words: non-redundant rules, frequent closed itemsets, adjacent graph, minimal generator itemsets

中图分类号: 

• TP311.13