一种新的重叠社区发现算法*

基于节点的中心度和节点对社区的适应度,提出了一种新的重叠社区发现算法。该算法以中心度很大的节点作为初始社区,然后访问社区的邻居节点,把对社区适应度最大的节点加入到社区。如果节点对多个社区都具有很大的适应度,则这些节点归属于多个社区。考虑到社区之间的重叠性,将社区相似度很大的社区合并为一个社区。将该算法应用到Zachary空手道俱乐部网络和海豚社会网络中,实验表明该算法能够很好地划分出网络中的重叠社区。

New algorithm for detecting overlapping communities

Based on the node of centrality degree and the node’s fitness to community,this paper proposed a new algorithm to detect overlapping communities.This algorithm took the nodes with a great centrality degree as the initial communities,then visited community’s neighbor nodes,and joined the node with a biggest fitness to the community.If the node had the very big fitness to many communities,then it belonged to several communities.Considering the overlapping property of community,it combined communities with great similarities into one community.It applied this algorithm into the Zachary’s Karate Club Network and the Dolphins Social Network,and the experiment indicates this algorithm can divide overlapping communities efficiently in the network.