Trustworthy Agent-Based Recommender System in a Mobile P2P Environment

Abstract

Abstract. Current major P2P systems focus on PCs and do not provide services for the mobile environment. Compared to traditional P2P, characteristics of Mobile P2P include unreliable connections, limited bandwidth and constraints of mobile devices. In addition, nomadic users demand applications and services that are context-aware, personalised, secure, and trustworthy. Recommender systems are one of these applications. In this paper, we aim at building a mobile P2P recommender system which dramatically reduces wireless traffic between peers, brings trustworthiness (each peer can choose to rely on opinions of peers whom he trusts), and ffers unobtrusiveness (the target system is mainly autonomous and requires a minimum user intervention). Our solution is based on multi-agent systems and is illustrated on a slow-food restaurant recommender system.