Link Prediction And Network Reconstruction Of Social Networks

Finding out the relationship among people is the most important issue in the research field of social network. With the development of Internet, online social network gets rapidly growth, so that people can communicate more easily and conveniently. This makes the study of social network become more complex. In the thesis we use the theory of complex network to study the social network. On one hand we adopt the method of Link Prediction to help us find out the relationship of people. On the other hand, in order to solve the problems appearing in the stage of social network extraction, the concept of Network Reconstruction is proposed.The thesis studies the connecting rule between nodes of social network on the point view of Link Prediction. Firstly we compare the results of several Link Prediction methods used in the social network. Considering the disadvantage of Random Walk in the stage of selecting the next hoop (neighbor of previous node), a Neighbor based Random Walk (RWN) is proposed. The RWN considers all the neighbors’likelihood of the initial node, and it can amend the original likelihood between the initial node and the destination node. On one hand, we have proved that RWN can get good predicting results, low time complexity and high algorithm stability. When used in the friends’ recommendation of social network, RWN can help us find the potential friends. On the other hand, in order to get the social network from Chinese literature we introduce a network detection method based on Chinese literature and we also improved the method to acquire the direct interpersonal relationship. Then we ues RWN to reconstruct the network that acquired from the Chinese literature. Taking the literature about history of the Three Kingdoms for example, we analyze the relationship network which was reconstructed and find out that the relationsips are clearer and more realistic.The RWN can help us understand the evolution mechanism of the social network and the driving force of making connection among people. We can use RWN to recommend friends for people. So network reconstruction provides an important guidance of reconstructing our social network.