基于HMT和哈希树的Apriori并行算法研究

摘要

为了进一步提高基于HMT和哈希树的Apriori算法的性能,提出了一种基于独立内存并行环境的并行化方案,充分利用空闲的计算资源来提高关联规则数据挖掘的效率.将原始数据集平均分配到并行环境中的各个子计算节点中,在各个子计算节点中并行地进行关联规则支持度计数,并从各个子计算节点中收集合并支持度计数的结果,得到目标频繁项集,进而实现Apriori算法的并行化.实验结果表明,该并行化方案可以很好地提高原算法的效率.%In order to improve the performance of Apriori algorithm based on HMT and Hash trees, a parallel method based on independent memory parallel environment is introduced. The vacant computing resources are fully used to enhance the efficiency of the association rule mining. The initial data is equally distribute to each child calculation nodes in parallel environment, then the support of association rule is parallel counted in each child calculation nodes, finally the results of support counting are col-lected and merged, consequently the target frequent itemset obtained. The experimental results show that the efficiency of origi-nal algorithm is commendably improved by this parallel method.