A layered overlap and parallelization community discovery method

A layered overlap and parallelization community discovery method In view of the fact that social networking is a large-scale graph, the hierarchical overlapping community detection algorithm based onlocal modularity is parallelized, Using the GraphLab framework based on BSP model, and for each update function in the hierarchical overlapping community discovery algorithm based on local modularity,the Gather, Apply and Scatter execution models of the update function are designed. Experimental results show that the parallel algorithm of hierarchical overlapping community discovery under the framework of GraphLab can improve the efficiency of community discovery and has a wide application prospect in large-scale graphs such as social networks. Comparing the original single-machine algorithmwith the parallel algorithm, it is found that the efficiency of community-level overlap structure discovery is improved without affecting the quality of community discovery. 本发明公开了种层次重叠并行化社区发现方法，针对社交网络是个大规模的图的特点，对基于局部模块度的层次重叠社区