基于共邻矩阵的复杂网络社区结构划分方法

Partition methods based on common neighbors matrix for the community structure in complex networks

提出了一种基于共邻矩阵和增益函数的划分算法来发现复杂网络中的社区结构.共邻矩阵中元素的含义为结点对之间拥有相同邻居的数目.以增益函数作为网络社区结构划分的目标函数,进一步推导出基于增益矩阵和增量矩阵的特征值和特征向量的社区结构划分方法.最后把这种算法应用于三个常用的实际网络数据中,并和Newman基于模块度矩阵的谱算法结果做了比较,以验证该算法的可行性和有效性.

Based on common neighbors matrix and gain function we propose a method of analyzing the community structure in complex networks.The elements in common neighbors matrix means the number of common neighbors between nodes.With the gain function as the objective function of analyzing the community structure,we derive a partition method based on the eigenvalues and eigenvectors of gain matrix and increment matrix.Further more,we apply this method to three common real network data and compare the computational results with modularity-based analysis methods proposed by Newman. Computational results demonstrate that the proposed method is feasible and effective.