基于拓扑特征的二分网络社团结构划分

摘要

After analyzing the topological properties of community in bipartite networks, a new community algorithm based on the idea of core and outer layers is proposed for resolving the problems that some community algorithms on bipartite networks can only classify one type of nodes or depend on additional parameters. This algorithm completely depends on the original network topology and it allows overlapping between communities. Experimental results show that this algorithm can exactly detect the number of communities and community structure of two types of nodes in realistic networks without additional parameters.%目前社团结构划分算法只能划分1类节点并且依赖于额外参数.为此,在分析二分网络社团拓扑特征的基础上,利用社团核与外层的思想,提出一种新的社团结构划分算法.该算法完全依赖于原始网络本身的拓扑结构,并且允许社团间重叠.实验结果表明,该算法无需任何额外参数,即可比较准确地识别实际网络的社团个数,同时划分2类节点的社团结构.