基于节点吸引力的社区发现算法  

The Community Discovery Algorithm Based on Nodes Attraction

作　　者：常富蓉 兀松贤[2] 

机构地区：[1]喀什师范学院信息工程技术系,新疆喀什844008 [2]喀什师范学院数学系,新疆喀什844008

出　　处：《喀什师范学院学报》2015年第3期51-54,共4页Journal of Kashgar Teachers College

摘　　要：提出了基于节点吸引力的分裂社区发现算法,算法认为网络中节点之间是存在一定吸引力的,不同相连节点之间的吸引力大小差异很大.如果相连两个节点的度数越相近,那么这两个节点相互吸引力越小;否则认为度数较大的节点对度数较小的节点的吸引力大.在本算法中,通过计算网络中节点之间的吸引力,删除节点间吸引力最小的连边,重复计算吸引力,并删除相应的边,实现网络的分裂,通过计算网络分裂后社区的中心性,实现社区的发掘.实验证明该算法可以有效划分网络社区.This paper presents a community discovery algorithm based on nodes attraction. This algorithm consid-ers that there is some certain attraction between nodes in the network, there exist big differences between the at-tractive sizes when connected to different nodes：The closer the degrees of connected to nodes is, the smaller of the attraction, otherwise the bigger the degree of the node is, the stronger the attraction it becomes as opposed to the smaller degree node. By calculating the attraction between network nodes, the algorithm adopted in this paper deletes the side which is the least attraction between two nodes, and then repeats calculating the attraction of con-nected to nodes, and delete the side, to realize the division of the network, through the calculation in the network community centricity, at least to realize discovery of communities. Experiments prove that the algorithm can effec-tively divide different network communities.

关 键 词：分裂算法 社区发现 吸引力 中心性 

分 类 号：TP393.02[自动化与计算机技术—计算机应用技术]