Link prediction of community in Microblog based on Exponential Random Graph Model

Link prediction is to predict the possible links between nodes, which recommend the most possible new ties or missing links according to the information of nodes and edges existed in the network. Microblog is a new-style social network. People connect each other in Microblog with their own interest and form the community in Microblog. For the community, link prediction is to recommend users to each other, which is important for people interaction and information spreading. Traditional methods of link prediction are appropriate for all kinds of networks, but ignore the social attributes of people in the community. In this paper, the features of social network are fully considered. A sociological model: Exponential Random Graph Model (ERGM) has been introduced in link prediction for Microblog. Synthesizing the data of user attributes and network topology, a link prediction model has been established. It is different from other methods, this model considers the community as a global network, and all nodes and edges contribute to the prediction of links. The result shows that this model has a significant effect for link prediction of community in Microblog.