Critique-Based Mobile Recommender Systems

Recommender systems provide targeted product suggestions for users either overwhelmed by the large number of alternative options offered nowadays in eCommerce websites, or not having enough knowledge to autonomously select the most suited product. Recommender systems are particularly useful for mobile users; here decisions must be normally taken in a short time and the effort required for interacting with the system must be limited as much as possible. In this article, we propose an approach to the generation of mobile recommendations based on the interactive elicitation of user needs and wants through critiques. In this approach the system does not mandatory require the user to explicitly communicate her preferences at the beginning of the interaction; but rather involves her in a dialogue where the system proposes candidate products, and the user feedbacks her critiques about the recommended products. These critiques are then interpreted and incorporated in the user’s preferences model managed by the system. This results, step by step, in a better understanding of the user’s preferences and needs and in a better ranking of products. The proposed approach has been implemented in a mobile travel recommender system which aims at supporting on-the-move travelers in the selection of travel related services (restaurant). In this article we present the results of the empirical evaluation of this system.