Multi-layered ontology-based user profiles and semantic social networks for recommender systems

Abstract

This paper describes a strategy that automatically clusters ontology-based user profiles taking into account their common interests for domain con-cepts. The obtained semantic clusters are used to identify similarities among indi-viduals at multiple semantic preference layers, and to define emergent, layered so-cial networks that can be applied in collaborative and recommender systems. As an applicative development of our method, we have experimented with building a personalized information retrieval model that provides ranked item lists based on the existing concept clusters and multi-layered user networks.