Cluster-based Collaborative Filtering for Link Prediction in Social Networks

In this paper, we proposed a method for link prediction based on collaborative filtering algorithm. Social networks are usually sparse, and the challenge is that collaborative filtering does not work properly in sparse data. In contrast, social networks are usually sparse. So, one of the main challenges is to integrate these two nonhomogeneous subjects. To suggest a solution, we first use Infomap algorithm in Mapequation framework to cluster the network. Then, some manipulations on the clusters are done in order to propose a better collaborative filtering algorithm based on the process of the predicted links between clusters. As a result, we achieve an algorithm with higher accuracy and lower time complexity in comparison with neighbor based link prediction methods