AprioriTid算法的MapReduce并行化实现  

Parallel implementation of AprioriTid algorithm with MapReduce

作　　者：周国军[1] 梁燕红[1] 唐微[1] 

机构地区：[1]玉林师范学院数学与信息科学学院,广西玉林537000

出　　处：《微型机与应用》2015年第24期22-24,27,共4页Microcomputer & Its Applications

基　　金：广西高校科学技术研究立项项目(LX2014300)

摘　　要：为解决AprioriTid算法对大数据执行效率不高的问题,根据Hadoop平台的MapReduce模型,分析了AprioriTid算法的并行化方法,给出了并行化的主要步骤和Map、Reduce函数的描述。与串行的AprioriTid算法相比,并行算法利用了多个节点的计算能力,缩短了从大数据集中挖掘关联规则的时间。对并行算法的性能进行了测试,实验结果表明,并行AprioriTid算法具有较高的执行效率和较好的可扩展性。To solve the problem which execution efficiency is not high when AprioriTid algorithm is applied for big data, according to MapReduce model of Hadoop platform, a parallelization method of AprioriTid algorithm is analyzed, and main steps of the method and description of Map and Reduce functions are given. Compared with serial AprioriTid algorithm, parallel AprioriTid uses computing power of many nodes to mine association rules, which shortens the execution time. The performance of parallel AprioriTid algorithm is tested, the experimental results show that the parallel algorithm has high execution efficiency and good scalability.

关 键 词：APRIORITID算法 MAPREDUCE HADOOP 关联规则 

分 类 号：TP311.1[自动化与计算机技术—计算机软件与理论]