An event detection method for social networks based on link prediction.

In the field of social network analysis, network evolution and event detection are the main current challenges. To meet them, current research work proposed many different models based on different network evolutions, and then evaluated the simulation results with different similarity indexes. However, the previous work usually had three problems: (1) Each method is only designed for a particular network; (2) There are many network statistics, so the evaluation on the performances of network models lacks an unified platform; (3) Without considering the temporal information, it is hard to track the network evolution and detect events. This paper proposes an event detection method for social networks based on link prediction, which can also evaluate the volatility of different networks. The main contributions are as follows. (1) The volatility of the network evolution can be effectively reflected on the similarity index corresponding to a good link prediction performance. (2) Inspired by link prediction, a similarity computing algorithm ( SimC ) is proposed to compute the similarity of networks. (3) Based on the output of SimC , the volatility of network evolution is evaluated and an event detection algorithm ( EventD ) is proposed. The experimental results show that the proposed method can effectively solve the problem of tracking network evolutions and detecting events. [ABSTRACT FROM AUTHOR]