基于prefixspan 的序列模式挖掘改善算法.pdfVIP

第35 卷 第23 期 计 算 机 工 程 2009 年12 月 ol Computer Engineering December 2009 V .35 No.23 ·软件技术与数据库· 文章编号：1000—3428(2009)23—0056—03 文献标识码：A 中图分类号：TP311 基于PrefixSpan 的序列模式挖掘改进算法 1,2 1 汪林林 ，范 军 (1. 重庆邮电大学计算机科学与技术学院，重庆400065 ；2. 重庆工学院，重庆400050) 摘 要：针对序列模式挖掘算法PrefixSpan 在挖掘过程中需要构造大量投影数据库的不足，提出IPMSP 算法，在递归挖掘过程中，通过检 查序列数据库关于前缀的前缀，避免对同一频繁前缀模式构造重复投影数据库，同时舍弃对非频繁项的存储并在投影序列数小于最小支持 度时停止扫描投影数据库，从而提高PrefixSpan 算法的时空性能。实验结果证明，IPMSP 算法在时间和空间性能上优于PrefixSpan 算法。 关键词：序列模式；PrefixSpan 算法；投影数据库 Improved Algorithm for Sequential Pattern Mining Based on PrefixSpan WANG Lin-lin1,2, FAN Jun1 (1. College of Computer Science Technology, Chongqing University of Post Tele., Chongqing 400065; 2. Chongqing Institute of Technology, Chongqing 400050) 【Abstract 】Aiming at the PrefixSpan algorithm produce huge amount of project databases in mining sequence patterns, this paper proposes an Improved PrefixSpan algorithm for Mining Sequential Patterns(IPMSP) alaorithm. By avoid produce duplicated project databases with the same prefix pattern through checking the prefix with regard to prefix of the sequence database and abnegating the non-frequent items and project databases which sequential number is lower than minimum support in the recursive mining process, the performance of PrefixSpan is well improved. Experiment results shows that the time and space performance of IPMSP algorithm are better than that of PrefixSpan. 【Key words 】sequential p