Survey and Analysis of Temporal Link Prediction in Online Social Networks

Online Social Network has tremendously captured the attention of users in recent years. Some social networking sites are explicitly designed for social interaction, while some of them are application based providing content sharing along with social communication. Link prediction is a new interdisciplinary research direction in which, existing links are analyzed and future links are predicted among millions of users of social network. Traditional link prediction methods had focused on the use of graph metrics to determine, where new links are likely to arise. A small amount of work has been done on analyzing long-term graph trends. This paper does the survey and analysis of temporal evolution of link prediction. It has been found out that the earlier graph generation models were unrealistic in their prediction and can be complemented through the use of temporal metrics, resulting in a highly accurate link prediction. In addition, the paper is concluded with the proposed framework which exploits temporal feature along with local similarity attributes.