Dynamic elicitation of user preferences in a chat-based group recommender system

Abstract

Group recommender systems (GRSs) support users in a group when facing decision making tasks. Many GRSs simply perform a one shot aggregation of the individual preferences and then compute recommendations. In this paper, we propose a different approach; it allows users to iteratively express and revise their preferences during the decision making process and supports such a conversational process with a chat-based interface. The implemented system observes users'interactions with the discussed options and offers appropri- ate support as well as recommendations to facilitate group members to come up with an agreement and a final decision. We have evaluated the effectiveness of our approach by conducting a controlled live user study. The empirical results show that the mobile application that implements the proposed approach achieves usability as well as high recommendation quality and choice satisfaction.