Research On Social Network Link Prediction Method Based On Node Similarity

With the rapid development of science and technology,social network analysis has become a new research focus.As one of the tasks of social network analysis,link prediction predicts the existence of relations between two nodes without links according to existing network structure and node information.It can be applied to a variety of social network research,and plays an increasingly important role.Therefore,link prediction has strong theoretical and practical value.The existing similarity-based link prediction methods utilize the similarity between nodes to predict whether there is a link between them.However,most algorithms only consider the number of neighbors in the network and do not take into account the relations among neighbor nodes,resulting in inaccurate prediction of the nodes in the network.Based on the existing algorithms,this paper deeply analyzes the calculation of node similarity.According to the structural characteristics of social networks,the different contributions of each common neighbor to link prediction are distinguished,and the concept of node cohesion is proposed.Based on the node cohesion,a link prediction algorithm considering node cohesion based on local path(NCLP)is proposed.Both the influence of the relations between neighbor nodes on the prediction results and the influence of the path factors and the end points on the similarity are studied.The experimental results show that the proposed algorithm can improve the accuracy of link prediction.In recent years,a lot of research on the real networks reflects that weighted networks can describe the real network system more comprehensively and profoundly.Therefore,this paper also studies the link prediction methods in weighted networks,and extends the NCLP algorithm to weighted networks.Experimental verification shows that the weighted methods can achieve better prediction effect than the unweighted link prediction methods.