The Study Of Link Prediction Algorithms Based On Similarity In Social Networks

In recent years,with the widespread popularity of online social networking sites,social network analysis has become a hot topic in academic research.Link prediction is one of the tasks of social network analysis,it has a very important theoretical and practical significance.For the structural features of social networks,this topic mainly study on link prediction issue of social network analysis.On the basis of in-depth study of social network related theories and knowledge,according to research status on link prediction in social network,this paper mainly study on link prediction in social network from the following two aspects:1.For the existing node similarity calculation methods based on local information only considered the individual characteristics of the common neighbors and have not taken full advantage of the network structural information,through in-depth analysis of the relationship between common neighbor nodes' adjacent nodes and the predicted nodes,this paper proposed a method to calculate the link strength of the neighbor node together with the two predicted nodes.Based on the node link strength,this paper proposed a new node similarity calculation method to achieve predict links.2.For the existing link prediction algorithms based on similarity only considered the node structure of the network and have not used the link as research subject to study the link prediction problem,by introducing the concept of link similarity,this paper analyzed the relationship between the unknown link and the known neighbor links,and proposed the definition of link credibility as the link predictive index to predict links.The two algorithms presented in this paper respectively used the link and the node as the research subjects to predict links.Wherein the link prediction algorithm based on the node link strength is the improved node similarity algorithm,It takes full account of the local information of the network,and better distinguishes the importance of each neighbor to link forming;by introducing the concept of link similarity,the link prediction algorithm based on the link credibility predicts links by studying links.Experimental compared with the current mainstream link prediction algorithms on three kinds of real data sets,the results show that the two algorithms presented in this paper are effectiveness and feasibility to predict links.