Matrix Based Community Evolution Events Detection in Online Social Networks

In recent years, the evolution of online social networks has become an important research topic in online social network analysis. An important approach to this problem is to detect community evolution events so as to understand the evolution of the whole network. Considering the huge amount of data in large social networks, an efficient and scalable community evolution detection algorithm is necessary. In this paper, we focus on community evolution events detection in dynamic social networks. First, we divide the Facebook and DBLP data set into a series of snapshots and apply the Louvain algorithm to find the communities in each snapshot. Then, we propose a light weight evolution events detection algorithm to find the community evolution patterns between adjacent snapshots, which statistically show the evolution trend of the entire network. Simulation results show that our algorithm can effectively detect the community evolution events in online social networks.