基于 MapReduce 的 H-mine 算法

摘要

频繁模式挖掘是一种非常有效地从数据中获取知识的方法，但是随着大数据时代的来临，现有算法及其计算环境的运算速度、内外存容量面临严峻挑战。针对以上问题，紧密结合 MapReduce 模型提供的高效分布式编程和运行框架，在深入分析 H-mine 频繁模式挖掘算法的基础上，通过对 H-mine 算法频繁模式挖掘过程的并行化改进，提出了一种新颖的基于 MapReduce 模型的 H-mine 算法（简称 MRH-mine）。MRH-mine 算法实现了对 H-mine 算法在分布式运行环境下的改造，实验表明该算法在面对数据大规模增长的情况下具有良好的性能和扩展性。%Frequent pattern mining is a very effective knowledge acquisition method from the data,but with the advent of the era of big data,the existing algorithm and its computing environments computation speed and storage capacity is facing severe challenges.Fortunately,MapReduce model provides an efficient framework for distributed programming and operation frame-work.Based on depth analysis of the H-mine algorithms and parallelizing H-mine algorithms,this paper proposed a novel Ma-pReduce-based H-mine algorithm (it call the parallel algorithm MRH-mine).MRH-mine algorithm achieve the transformation of H-mine algorithm in the distributed operation environment,experimental results show that in the face of massive data growth,MRH-mine algorithm have a good performance and scalability.