Consistent continuous trust-based recommendation systems

Abstract

The goal of a trust-based recommendation system is to generate personalized recommendations from known opinions and trust relationships. Prior work introduced the axiomatic approach to trust-based recommendation systems, but has been extremely limited by considering binary systems, while allowing these systems to be inconsistent. In this work we introduce an axiomatic approach to deal with consistent continuous trust-based recommendation systems. We introduce the model, discuss some basic axioms, and provide a characterization of a class of systems satisfying a set of basic axioms. In addition, as it turns out, relaxing some of the axioms leads to additional interesting systems, which we examine. © 2009 Springer-Verlag Berlin Heidelberg.