Comparative Analysis of Different Algorithms in Link Prediction on Social Networks

Abstract

In today’s world, the social-based online network systems are rapidly expanding, which results in large accumulation of data. When we try to examine this enormous amount of data that has been collected in these systems, we face new difficulties. The forecasting of user social relationships is one of the intensively researched topics. Link prediction locates broken links or forecasts the probability of new links. The problem of Link prediction is an extremely well-researched issue and has uses across a wide range of fields, some of which can frequently be observed in recommendation algorithms, like new connections (suggested friends/followers) on social networks or associated merchandise on online store. A lot of algorithms have already been presented in context for solving the link prediction problem. We selected three real-world social networks and seven most widely used link prediction algorithms. To the best of our knowledge, we conduct a survey of these Link Prediction approaches that are currently in use and compare them in this work.