基于网络嵌入与局部合力的复杂网络社区划分算法

社区划分可以揭示复杂网络中的内在结构和行为动态特点,是当前的研究热点。文中提出了一种基于网络嵌入和局部合力的社区划分算法。该算法将网络的拓扑空间转化成欧氏空间,把网络节点转换成向量表示的数据点,首先基于重力模型和网络拓扑结构,提出局部合力和局部合力余弦中心性指标(Local Resultant Force Cosine Centrality,LFC),通过节点的LFC和节点间的距离来确定各个初始小社区的中心节点,然后将网络中其他的非中心节点划入与其最近的中心节点所在的初始小社区内,最后通过优化模块度的方法来合并初始小社区并找到最优的网络社区结构。在6个现实世界网络和可调参数人工网络上与6种知名社区划分方法进行比较,比较结果表明了新算法良好的社区划分的性能。

Community Detection Algorithm in Complex Network Based on Network Embedding and Local Resultant Force

Community detection can reveal the inherent structure dynamic behavior in complex networks and it is the current research hotspot.In this paper,we propose a community detection algorithm based on network embedding and local resultant force.The network topological space is transformed into euclidean space,and network nodes are converted into vector data points.First,based on the gravity model and the network topology,a local resultant force and a local resultant force cosine centrality index(LFC)are proposed.The center node of each initial small community is determined by the LFC of the node and the distance between nodes.Then the rest of the non-central nodes are classified to the nearest central node to form the initial small community.Finally,communities are merged by optimizing the modularity to find the optimal network community structure.Compared with 6 well-known community detection algorithms on 6 real-world networks and artificial networks with adjustable parameters,the new proposed algorithm shows good performance in the community detection.