基于动态项集计数的加权频繁项集算法  

Weighted Frequent Itemset Algorithm Based on Dynamic Itemset Counting

作　　者：秦丽君[1,2] 罗雄飞[1] 

机构地区：[1]中国科学院软件研究所,北京100190 [2]中国科学院研究生院,北京100190

出　　处：《计算机工程》2012年第3期31-33,共3页Computer Engineering

基　　金：国家"863"计划基金资助项目(2007AA040702)

摘　　要：基于Apriori的加权频繁项集挖掘算法存在扫描数据集次数多的问题。为此,提出一种基于动态项集计数的加权频繁项集算法。该算法采用权值键树的数据结构和动态项集计数的方法,满足向下闭合特性,并且动态生成候选频繁项集,从而减少扫描数据集的次数。实验结果证明,该算法生成的加权频繁项集具有较高的效率和时间性能。The existing weighted fiequent itemset mining algorithms which are based on Apriori require multiple dataset scans. This paper proposes a weighted frequent itemset algorithm weighted frequent itemset mining based on dynamic itemset counting which uses the structure of weighted trie tree and the method of dynamic itemset counting. This algorithm satisfies the downward closure property and dynamically generates candidate frequent itemsets, thereby reduces the number of scanning datasets and improves the performance. Experimental results show that the proposed algorithm not only generates the weighted frequent itemsets, but also has high efficiency and time performance.

关 键 词：数据挖掘 加权频繁项集挖掘 动态项集计数 加权支持度 权值键树 向下闭合特性 最大权值 

分 类 号：TP311.1[自动化与计算机技术—计算机软件与理论]