静态全符号网络的社群结构

The theory of community structure in fully signed networks is proposed to overcome the difficulty of traditional community structure in analyzing the individual and relation features in the process of collective actions. A community structure in fully signed networks is a kind of network structure that consists of sub-groups and internal nodes within a sub-group having the same attribute and are connected by dense positive edges while the nodes within different sub-groups having different attributes and are connected by sparse negative edges. First, a definition of the community structure in fully signed networks (consisting of both node and edge signs) is proposed by comprehensively considering the attributes of nodes, edges and network structures. Then, a new modularity is proposed, based on the theory of modularity given by Newman and the new modularity considers three factors of nodes, edges and structures. A genetic algorithm is then designed to detect the community structure by optimizing the new proposed modularity. Experiments based on different network attributes and a comparison with the traditional modularity show that the proposed modularity effectively identifies the differences of these attributes and solves the problem that the traditional community-structure-detecting method cannot recognize features of nodes and edges.