基于种子节点选择的重叠社区发现算法*

识别网络社区对于了解社会网络的结构和功能具有重要意义。由于网络中某些节点可能属于多个社区,因此重叠社区的研究已经吸引了人们越来越多的关注。本文针对目前从局部社区扩展成全局社区时有关算法的种子节点选择不合理的情形,提出了一种基于种子节点选择的重叠社区发现算法。本算法首先根据影响力函数找出局部影响力最大的节点,由这些节点构成的种子集合较好的分布在整个网络中,然后以这些种子点构造初始社区,根据设定的吸引度函数选择性添加节点来进行社区扩展。实验结果表明,该算法在真实网络上进行测试时能够有效的挖掘网络中的重叠社区。

Overlapping community detection algorithm based on selection of the seed nodes

Identification of communities is significant in understanding the structures and functions of the social network. Because of some nodes in the network may belong to more than one community, the study of overlapping communities has attracted more and more attention recently. This article in view of the unreasonable selection in seed algorithm from the local community expanding into a global community at present, proposed an overlapping community detection algorithm based on selection of the seed nodes. The algorithm used the influence function to find out the strongest nodes in local node influence, which structuring the seeds distributing throughout the network, and then utilized the seeds to construct the initial community, selectively adding nodes to expand the community according to the set attraction function. The experimental results show that the algorithm tested in a real network can effectively dig out Overlapping community in the network.