在线社会网络的动态社区发现及演化

Dynamic Community in Online Social Networks

在线社会网络研究中,动态隐含社区或群组结构的发现及演化探测是一个十分关键的核心问题,它对于在中观(Mesoscopic)视图观察在线社会网络隐结构特征、预测演化趋势、掌控网络势态、发现网络异常群体事件等具有重要意义.文中首先分析了动态社区发现和社区演化研究的关系,给出动态社区研究中关键挑战问题;然后根据问题背景的不同,从"同构社会网络的动态社区研究"和"异构社会网络的动态社区研究"两个方面进行国内外相关研究现状的阐述和分析,其中,在"同构社会网络的动态社区研究"中,根据评价方法的差异和关注问题的不同将当前相关研究分为基于时空独立评价、时空集成评价、统一评价和增量式算法4大类进行综述,同时对动态社区发现的重要应用——异常群体发现的研究进行介绍;最后对在线社会网络动态社区领域的难点和发展趋势进行分析和展望.

It is an important issue in online social networks to detect hidden communities and track their evolution process, which will help understanding the latent topology, predicting its evolution trend, discovering abnormal events and controlling the network. We firstly give the explanation of the relationship between community detection research and community evolution research, and put forward their main challenges. Then we introduce the related research from two different angles, one is dynamic community in homogenous social networks and the other is that in heterogeneous social networks. To clearly state the first area, we introduce the related work by dividing them into 4 classes on the evaluation mechanism, temporal-spatial independent evaluation based, temporal-spatial integrated evaluation based, unified evaluation based and incremental algorithms. An important application is also reviewed that is detection abnormal swarm events. At last some future research topics are given.