【作者】 陈德华； 解维； 李悦；

【Author】 Chen Dehua,Xie Wei,and Li Yue (School of Computer Science & Technology,Donghua University,Shanghai 201620)

【机构】 东华大学计算机科学与技术学院；

【摘要】 图聚类作为一种重要的图数据挖掘技术,在许多实际应用中已得到广泛的关注.针对图数据日益普及、规模不断增大的现实,提出了一种高效的分布式并行图聚类算法,即MR-LSH算法.该算法是在MapReduce框架的基础上,通过对传统的位置敏感哈希(locality sensitive hashing,LSH)算法的并行化改造,使其可在分布式的集群环境中实现对大规模图数据的高效聚类.通过在真实数据集上的实验,表明该算法的可行性与有效性.更多还原

【Abstract】 As an important graph data mining technology,graph clustering has been received a lot of concerns in many real applications.These days have seen the fact that graph data is becoming popular and scale.Thus this paper presents an efficient parallel graph clustering algorithm,namely the MRLSH algorithm,in a large distributed environment.The MR-LSH algorithm is presented by reforming the traditional Locality Sensitive Hashing algorithm into a parallel and distributed algorithm using MapReduce framework,which can achieve efficient clustering on large-scale graph data in a large machine cluster environment.Experiments on real datasels showed that the algorithm is feasible and effective.更多还原