一种无候选项的闭合序列模式挖掘算法  

A CLOSED SEQUENTIAL PATTERN MINING ALGORITHM WITHOUT CANDIDATE TERMS

作　　者：杨斐[1] 张万桢[2] 陆垂伟[1] 

机构地区：[1]湖北理工学院计算机学院,湖北黄石435003 [2]桂林电子科技大学,广西桂林541004

出　　处：《计算机应用与软件》2016年第3期279-283,共5页Computer Applications and Software

基　　金：湖北省自然科学基金项目(2013CFB039);湖北省教育厅重点科学研究项目(D20144403);湖北省教育厅科学研究项目(B2013064);湖北理工学院优秀青年科技创新团队项目(13xtz10)

摘　　要：算法Clo Span在挖掘闭合序列模式时分两阶段进行,首先产生候选的闭合序列模式,然后在此基础上挖掘闭合序列模式。针对Clo Span算法中大量候选模式影响挖掘效率的问题,提出改进的算法ss Clo Span。该算法在序列模式增长时,利用支持度和末节点哈希表剪枝非闭合模式,同时利用频繁项头表进行闭合性检测。实验结果表明,对于不含项集项的序列,当存在较长频繁序列时,挖掘效率得到了有效的提高。When mining the closed sequential patterns,the algorithm Clo Span does it in two phases. First,it will generate the candidate closed sequential patterns,and then based on this it mines the closed sequential patterns. Aiming at the problem of numerous candidate patterns impacting the mining efficiency in Clo Span algorithm,we proposed an improved algorithm called ss Clo Span. This algorithm prunes the non-closed patterns using support degree and last node hash table in the course of sequential pattern growth,meanwhile it uses frequent item header table for closure checking. Experimental results showed that for the sequences without itemsets and with longer frequent sequence,the mining efficiency was well improved.

关 键 词：闭合序列模式 支持数剪枝 末节点哈希表 频繁项头表 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]