一种基于TFP树的频繁项集改进挖掘算法  

A improved FP-growth mining algorithm based on TFP-tree

作　　者：董平[1] 胥杰[1] 苏力萍[1] 

机构地区：[1]北京科技大学信息工程学院,北京100083

出　　处：《微计算机信息》2007年第33期139-140,97,共3页Control & Automation

基　　金：教育部基金资助项目(03023)

摘　　要：FP-growth算法是一种被证明有效的频繁模式挖掘算法。但是由于在挖掘频繁模式时需要递归地生成大量的条件FP-树,其时空效率较低,本文针对这一问题,首先构造一种改进的TFP-树结构,然后在构造的TFP-tree基础上引入被约束子树提出一种基于TFP树的频繁项集的改进挖掘算法,并对该算法进行性能分析,结果证明该算法在运行速度得到很大提高。FP-growth algorithm is a frequent pattern mining algorithm which has been proved to be efficiency .But this algorithm must generate a huge number of conditional FP-trees recursively in process of mining , so the efficiency of FP-growth remains unsatisfactory. In this paper, an improved TFP-tree frame was built firstly ,then introducing constrained sub tree in the base of TFP- tree. Finally ,an improved frequent pattern mining algorithm based on const rained TFPtree is proposed ,at the same time a per- formance analysis was carried out .The result proved that this algorithm＇s run speed was improved a lot .

关 键 词：数据挖掘 临时频繁模式树 被约束子树 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]