Predicting future link in social network

Abstract

A social network can have many different types of links or margins between nodes. Those, for example, could be social contacts, major contacts, or calls. Link Prediction is the problem of predicting edges that may not be present in a given or present time, but that have not yet been discovered and may occur in the near future. We are developing predictive linking methods based on step-by-step analysis of network nodes. Consider a network of collaborative writing among scientists, e.g. The two scientists closest to the network will have similar colleagues, so they may be working together soon. Our goal is to make this accurate idea more accurate and to understand what steps to take to approach the network that lead to the most accurate prediction of a link. Link prediction algorithms can be divided into three categories: Node Neighborhood Mode, Mode-based Mode, and Meta Mode. The node location method is based on network location features, which focuses mainly on the node structure (i.e., based on the number of common friends shared by two users). Local-based measures are: general neighborhood, Jaccard Coefficient, Adamic/Adar, and preferred attachments. P