基于权重信息挖掘社会网络中的隐含社团

社团结构是一种普遍存在于各类真实网络中的结构特性.挖掘网络的社团结构对于理解网络的功能与行为有着重要作用.然而,现有的各种社团挖掘算法仅仅基于网络拓扑结构信息,而忽视了蕴涵于真实社会网络边权信息中丰富的语义信息.目前普遍使用的基于模块性最大化的社团挖掘算法倾向于将小社团合并,这使得语义上丰富的小社团容易湮灭于基于拓扑结构信息所挖掘出的大社团中.而挖掘出这些隐含于大社团中的有着丰富语义内涵的小社团对于加深社会网络语义层面的理解有着重要作用.为此,提出一个接近线性复杂度的有权网络社团挖掘算法.通过充分利用权重信息,算法可以将社会网络划分为富含语义信息的粒度较细且相对较小的隐含社团.通过对基于DBLP作者合作网络的实证分析,证实了新算法的有效性和高效性.

Mining Hidden Communities in Social Networks Based on Weight Information

Community structure is a common property across a variety of real networks.Finding the community structure is helpful to understand the function and behavior of the network including social networks.However,most of previous community detection algorithms heavily rely on the topology information of the network without considering the meaningful semantic information implied in the edge weight of a network.Many existing community detection algorithms following the framework of maximizing the modularity tend to merge small communities into a large community,despite of the fact that these small communities may be semantically meaningful.Nevertheless,mining these meaningful small communities hidden in some large community detected by those topology-based community detection algorithms is helpful for US to gain deep understanding about the social network from the semantic aspect.Hence,in this paper,an almost linear algorithm is proposed,which exploits the edge-weight information of social networks,to detect hidden communities that are semantically meaningful from social networks.The authors construct co-author network from DBLP and perform extensive analysis on the network to show the effectiveness and efficiency of the new algorithm.