Link Prediction in Social Networks Using the Extraction of Graph Topological Features

The social relationships in social network analysis are considered using the terms node and link. The nodes are the individual actors in the networks and the links are the relationships between these actors. Regarding these links and their communications, the problem of link prediction is one of the most important challenges in social network analysis. Link prediction is one of the most popular applications in the online social network services helping users find new friends with similar interests. The present study tends to provide a hybrid criterion for comparing similarity among users with respect to the characteristics of the graph topology structure. The proposed method consists of two steps. In the first step, the similarity of users is calculated with different factors according to the similarities extracted from the features and the internetwork local circles. In the second step, the similarity of users is calculated using a hybrid similarity criterion. Different probabilistic factors have been used in the proposed similarity criterion, showing the effect of each feature on the final similarity criterion. The probability of any feature is optimized using a hill-climbing algorithm. The actual dataset from Twitter social network has been used in order to evaluate the performance of the proposed method. The results of the above tests indicate the higher performance of the proposed method compared with other similar methods.