Link Prediction on Social Attribute Network Using Lévy Flight Firefly Optimization

The problem of link prediction largely depends on the topological information. Social attribute network model is employed where the nodes represent both the social nodes and also the attribute nodes. The edge represents the interaction between the social nodes, the interaction between the social node and the attribute nodes but not the interaction between the attribute nodes. In this paper, firefly optimization algorithm using Levy search is employed to predict links. The proposed algorithm accuracy is measured in terms of AUC and precision and compared with similar methods in literature. From the experimental results, it is evident that Levy walk outperforms over other existing algorithms. From the results, we also infer that exploiting the Levy search firefly algorithm taking shorter jumps has improved the accuracy of the link prediction algorithm over the methods that take longer jumps based on random walks.