A new social recommender system based on link prediction across heterogeneous networks

Abstract

Nowadays, the wide use of the Internet around the world allows people to socialize and connect together. This results of the explosion of the Web 2.0 giving rise to a growing demand for Social Recommendation Systems. Social recommendation systems are introduced to rescue users from searching and choosing by predicting users’ preferences. In this paper, we will focus on recommendation via link prediction across heterogeneous social network. The main objective is to recommend items by predicting the missing or unobserved interactions between actors within a social network while pinpointing different types of objects and links. Probabilistic relational models will be used for prediction of new interactions in a citation network.