多周期经常性关联规则的发现

1 引言近年来,现实生活中数据量在高速增长,而在数据中发现有效知识的技术却相对匮乏,因此数据挖掘这一领域成为大家注目的焦点。对事务数据库进行分析的一个十分重要内容是关联规则的发现。此问题被Rakesh Agrawal等首先提出,尔后得到了广泛的研究,如文考虑了发现关联规则的效率问题,文考虑了增量发现问题,文考虑了在时间序数据库中发现周期性模式等等。但以上的工作都是将数据库看成是一个整体,没有考虑时间段的问题。最近,B.Oz-den等研究了发现“完全的”周期性关联规则的问

Discovery of All Frequent Cyclic Association Rules

The problem of founding all frequent cyclic association rules,which are present in most but not the entire series of time units that have periodic time intervals, has been considered in this paper.Discovering these rules and their periodicity may reveal interesting information that can be used for pre-dictions and making decisions. When the range of cycle lengths had been specified,the frequent associa-tion rules could been found by methods already existed,while this paper proposed an algorithm to re-duce the unnecessary amount of work performed during the data mining process. Furthermore, we showed the effectiveness of this method through experiments.