Graph-based learning on sparse data for recommendation systems in social networks

Abstract

Graph-based supervised learning is a useful tool to model data supported by powerful algebraic techniques. This paper proposes a novel approach using Graph-based supervised learning to handle the problem of building Recommendation Systems in Social Networks. Specifically, the main contribution of this paper is to propose two ways to construct Recomendations Systems based on Colaborative Filtering. The first way consists in building a feature matrix from the Web of Trust of each user. The second way builds a feature matrix based on similiraties among users according to the criterion of the system without taking into consideration the Web of Trust of the target user.