Link Prediction Based On Similarity In Social Network

Social network covers various fields of human reality. In recent years, in order to study the interaction law, the issue of the social network link prediction received widespread attention in the academic fields. Link prediction can help to understand the dynamic evolution mechanism of social network structure, and has great value to promote the development of social network research. Weighted network and symbolic networks are more complex social networks than general networks. However, there is less research about the prediction link of these two types of networks, and there are limitations based on the similarity of method, the predictive ability is relatively low. Based on the above problems, this paper makes a further research on the link prediction problem according to the structure characteristics of the network.First, in the weighted network, in order to show the importance of weight in the network, the weight is integrated into the similarity measure in the form of node strength. In order to make full use of the network topology and take into account the execution efficiency of the prediction algorithm, the advantages of global similarity and local similarity are complementary, and the similarity calculation method of multi path transmission is proposed.Secondly, in the symbolic network, this paper solves link prediction combining the balance theory and the similarity method. In order to get rid of the limitations of the traditional theory of social balance, this paper puts forward the balance ring structure concept on the basis of traditional social balance theory, and puts forward multi loop transfer similar prediction algorithm. By using this algorithm, two kinds of problems are explored, namely, the boundary value of existing edge and the link of the non existence.Finally, experiments are carried out on weighted networks and symbolic networks, and the correctness and effectiveness of the proposed algorithms are verified by experiments.