基于滑动窗口的数据流频繁闭项集挖掘  

Data Stream Frequent Closed Itemsets Mining Based on Sliding Window

作　　者：李俊[1] 杨天奇[1] 

机构地区：[1]暨南大学信息科学与技术学院,广州510632

出　　处：《计算机工程》2009年第13期37-39,共3页Computer Engineering

基　　金：广东省科技计划基金资助项目(2007B031300001)

摘　　要：针对数据流的特点,根据Moment算法提出一种基于频繁闭项集挖掘的增量式维护算法。该算法通过滑动窗口增量更新数据流中的事务,采取一种高效的项的位序列表示方法降低窗口滑动的时间和空间复杂度,应用压缩的模式树进行频繁闭项集检查,以确保挖掘结果的准确性。实验证明了该方法的有效性。Aiming at the features of data streams, this paper presents an incremental maintaining algorithm based on frequent closed itemsets mining according to Moment algorithm. It incrementally updates the transactions in the data stream via a sliding window, and uses an effective bit-sequence representation of items to reduce time and memory of the sliding window. A compressed pattern tree is used in frequent closed itemsets test to guarantee veracity. Experimental results show that this algorithm is effective.

关 键 词：滑动窗口 数据流 频繁闭项集 Moment算法 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]