复杂网络社区的分形聚类检测方法

Fractal clustering method for uncovering community of complex network

提出了两阶段盒子覆盖法,并且以两阶段盒子覆盖法作为节点聚类方法,提出了分形聚类社区检测算法FCUC。FCUC算法将分形聚类过程映射到树型结构,通过对树型结构进行分割得到复杂网络的社区结构。在人造网络和现实网络上对FCUC算法进行了测试,实验结果表明:FCUC算法可以有效地检测出社区结构。

Box-counting algorithm is an important approach to inspect the fractal property of complex networks.However,it neglects the weight of edges.To address this problem,a novel two-step boxcounting algorithm is proposed,which is a clustering method.The proposed algorithm is applied to Fractal Cluster for Uncovering Community（FCUC）.In FCUC,the nodes are clustered by two-step covering-box algorithm,and the fractal tree progress is mapped to a tree structure.Community structure can be uncovered through cutting the fractal tree according to evaluation function.Extensive tests on artificial networks and real networks give excellent results.