一种改进的最大频繁项目集挖掘算法

An Improved Maximal Frequent Itemsets Mining Algorithm

本文提出了一种基于布尔矩阵FP-array的最大频繁项目集挖掘的并行算法。该算法利用基于前缀的划分方法将事务数据集划分为较小的子空间,并将具有完全包含关系的项目集分配到同一处理机,然后各处理机站点Si分别进行局部最大频繁项目集的挖掘,再将挖掘结果传送到主站点S,最后得到全局最大频繁项目集。

An improved parallel algorithm for mining maximal frequent itemsets based on FP-array is proposed in this paper. The algorithm divides the transaction data sets into smaller one based on the prefix, and distributes the itemsets which have the complete include relationship to the same site. Then each site Si mines local maximal frequent itemsets reapectively, and delivers the mining results to the main site S. Finally, the global maximal frequent itemsets are obtained.