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。

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

Abstract: The existing weighted frequent 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.

Key words: data mining, weighted frequent itemset mining, dynamic itemset counting, weighted support degree, weighted trie tree, downward closure property, maximum weight

中图分类号: 

• TP311.1