一种改进的频繁模式树生长算法  

An improved frequent pattern tree growth algorithm

作　　者：连志春[1] 伊凤新[1] 

机构地区：[1]朝阳师范高等专科学校电子计算机系,辽宁朝阳122000

出　　处：《应用科技》2008年第6期47-51,共5页Applied Science and Technology

基　　金：国家863高技术研究发展计划基金资助项目(2004AA115090)

摘　　要：针对垂直数据分布数据库FP-tree生长基本算法中存在的链接点表空间随问题规模线性增加的问题提出一种改进算法.采用定长的链接点表进行分段扫描,在空间需求恒定的前提下构造FP-tree.证明了改进算法与基本算法构造的FP-tree是同构的.实验与分析结果表明,当应用于同一数据集时,改进算法所需空间恒定.In frequent pattern tree growth algorithm for databases with vertically distributed data, the memory space required by the link-point list increases linearly with increasing size of the data set, leading to a huge memory consumption when mining large database. To solve this problem, a new algorithm with a fixed-length link-point list and a section-by-section scanning strategy is presented, in which the FP-tree is erected under the precondition of constant memory space. It is proved that the FP-trees built by this algorithm and the original algorithm are isomorphic. Experimental and analytical results showed that, compared with the variable memory space of original algorithm, the memory space demanded by the new algorithm is constant.

关 键 词：数据挖掘 最大频繁项目集 关联规则 频繁模式树 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]