Item-Driven Recommendation: A Recommendation System for Other Items Based on the Users' Choice

Abstract

It is important for interactive agents in commercial facilities to build a relationship with users to facilitate their purchases. However, interactions that interfere with the users’ normal purchasing behavior may sometimes cause users to feel burdened or uncomfortable, which in turn may decrease the effectiveness of the recommendation. In this study, we propose "Item-Driven Recommendation" in which the item selected by a user recommends other items by voice accompanied by movement. In general, a user selects items due to trust, interest, or concern for them. We believe that the user’s emotion toward the selected item is directly related to the trust relationship with the agent, which increases the acceptability of the recommendation. The results of our preliminary experiments showed that the percentage of users who selected the recommended item because of the recommendation was the highest in the proposed system, suggesting that a relationship may have been established between them.