现代计算机（普及版）
現代計算機（普及版）
현대계산궤（보급판）
MODERN COMPUTER
2014年
10期
43-46
,共4页
K-means聚类算法%Hadoop%并行化
K-means聚類算法%Hadoop%併行化
K-means취류산법%Hadoop%병행화
K-means Clustering%Hadoop%Parallelization
对于具有海量信息的个性化推荐问题，K-means聚类算法的传统实现方式已不能快速准确地满足要求。基于目前最为流行的开源云计算平台Hadoop及分布式计算框架MapReduce，实现K-means聚类算法的并行化。给出该算法的具体实现，实验表明能够较好地解决时间瓶颈问题。
對于具有海量信息的箇性化推薦問題，K-means聚類算法的傳統實現方式已不能快速準確地滿足要求。基于目前最為流行的開源雲計算平檯Hadoop及分佈式計算框架MapReduce，實現K-means聚類算法的併行化。給齣該算法的具體實現，實驗錶明能夠較好地解決時間瓶頸問題。
대우구유해량신식적개성화추천문제，K-means취류산법적전통실현방식이불능쾌속준학지만족요구。기우목전최위류행적개원운계산평태Hadoop급분포식계산광가MapReduce，실현K-means취류산법적병행화。급출해산법적구체실현，실험표명능구교호지해결시간병경문제。
In terms of the personalized recommendation problem of mass information, the traditional implementation of K-means clustering algorithm can not meet the requirements. Based on the most popular open source cloud computing platform Hadoop and distributed computing framework, realizes the parallelization of K-means clustering algorithm. Finally gives a concrete realization of this algorithm and the ex-periment shows that the problem of time bottleneck can be well solved.
