基于局部模块度的社区层次结构发现方法

Hierarchical Community Structure Detection Based on Local Modularity

为了发现复杂网络中社区之间的层次关系,提出了一种基于局部模块度的社区层次结构发现方法。文章方法克服了多分辨率方法无法给出整个网络的层次划分以及无法直接定位造成社区层次变化的分辨率等方面不足,选取网络中的大度数节点基于R公式进行社区层次结构探测,根据局部模块度值变化过程中产生的极大值和极小值定义了社区层次区分度来判断是否到达层次边界。并对网络进行裁剪,从不同的大度数节点出发来发现网络中的全部层次结构。在经典数据集和人工生成网络上进行了实验,并与现有算法进行比较,实验结果证明章算法的有效性。

A method based on the local modularity is proposed to detect the hierarchical community structure in complex network. The method overcomes the short of multi-scale method which can not detect all the hierarchical community structure and can not directly get the parameter which causes the hierarchical structure changing. By selecting the great degree node to detect hierarchical commu- nity structure based on the R formula, the hierarchical division is defined by the max and rain local modularity to judge whether it arrives at the boundary of hierarchical community. The method cuts out the network and selects the different great degree nodes to detect all the hierarchical community structure. The proposed method is tested on both common network and synthetic network, and compares with the typical methods. Experimental results verify and confirm the validity of the proposed method.