大规模复杂网络下重叠社区的识别

The Identification of Overlapping Communities in Large-Scale Complex Networks

随着网络规模的不断扩大,经典的复杂网络重叠社识别算法已不能高效处理现有的大规模网络图数据.本文在GraphLab并行计算模型上提出了基于重要节点扩展的重叠社区识别算法DOCVN(Detecting the Overlapping Community algorithm based on Vital Node Expanding in GraphLab).算法选取网络中PageRank值大的节点作为重要节点,计算其他节点归属于重要节点的节点归属度,并以重要节点为中心形成核心社区及扩展社区,最后根据重要节点间的连接紧密度合并核心社区及扩展社区,并计算出每个节点在所属社区里的节点重要度,实现了大规模网络的重叠社区识别.实验表明该算法与PD(Propinquity Dynamics)等现有并行算法相比更能有效地识别大规模网络的重叠社区结构.

With the unceasing expanding of network scale, many classic detection algorithms of overlapping communities cannot work efficiently in large-scale complex network. Detecting the overlapping community algorithm based on vital node expanding in parallel framework GraphLab （DOCVN） is introduced to identify the overlapping communities. In this algorithm, nodes with high PageRank value are regarded as vital nodes, and then the affiliation degree of other nodes to these vital nodes are computed. After that, kernel communities and expanding communities are identified respectively. Finally, the kernel communities and expanding communities are combined into some overlapping communities by judging whether they connect tightly. And the importance weight of each node in its community is also computed. Experimental results show that the algorithm is more effective than the existing parallel algorithms like PD （Propinquity Dynamics ） to identify large-scale overlapping communities.