Shabaz–Urvashi Link Prediction (SULP): A Novel Approach to Predict Future Friends in a Social Network

Abstract

With the growth of social networks, the problem of linking the isolated or missing nodes appears. Thus, link prediction comes into existence to resolve this problem. Link prediction may be defined as an approach to predict an optimistic relationship that may exist or is likely to exist between nodes. Predicting the prospect link formed in future between nodes either in a dense or sparse network, the number of techniques exist intending to establish a link based on a certain similarity between the nodes. After conducting in-depth research on almost every link prediction technique, we reach the conclusion that every technique evaluates the probability score to predict future links. This research work discusses almost every previous technique and puts forward a comparatively similar technique for link prediction. The proposed technique is named Shabaz–Urvashi Link Prediction (SULP), which is based on a formula derived from an empirical theory after making a node matrix and altering the position of the neighbouring nodes, which states, ‘A node is predicted to establish a friendship if it has a maximum degree in its common neighbouring row and a minimum degree in its common neighbouring column’. SULP is tested using established datasets and compared with other link prediction techniques on the statistical measures such as Area Under Receiver Operating characteristic Curve (AUROC), precision and recall. SULP performs better as compared to other link prediction techniques on most of the testing datasets.