Unifying Recommender Systems and Conversational User Interfaces
Alain Dominique Starke, Minha Lee · 2022
This paper considers unifying research on conversational user interfaces and recommender systems. Studies on conversational user interfaces (CUIs) typically examine how conversations can be facilitated (i.e., optimizing the means). Recommender systems research (RecSys) aims to retrieve and present recommendations in a user’s session (i.e., optimizing the ends). Though these aims are overlapping across both areas, they can be better examined together to target the means and ends of what people can achieve with technology as conversational recommender systems (CRSs). We discuss the intersection of conversational user interfaces, recommender systems, and conversational recommender systems. We argue how conversations and recommendations can be designed holistically, in which recommendations can also be a means to foster engaging conversational interaction, while conversations as ends can better sustain curated, long-term recommendations.