Tell me What you Like: Introducing Natural Language Preference Elicitation Strategies in a Virtual Assistant for the Movie Domain

Abstract Preference elicitation is a crucial step for every recommendation algorithm. Traditional interaction strategies for eliciting users’ interests and needs range from button-based interfaces, where users have to select what they like among a set of fixed alternatives, to more recent conversational interfaces, where users have to reply to some questions formulated by the algorithm about their preferences. However, none of these strategies either mimics the dynamics of real-world open-ended interactions, or allows users to express their needs with an adequate level of expressiveness and control. In this paper, we present a strategy that allow users to express their preferences and needs through natural language statements. In particular, our natural language preference elicitation pipeline allows users to express preferences on objective movie features (e.g., actors, directors, etc.) that are extracted from a structured knowledge base, as well as on subjective features that are collected by mining user-written movie reviews. To validate our claims, we carried out a user study in the movie domain (N = 249). The main finding of our experiment is that users tend to express their preferences by using objective features, whose usage largely overcomes that of subjective features, that are more complicated to be expressed. However, the combination of objective and subjective features often leads to better recommendations, at the cost of a slightly longer conversation. We have also identified the main challenges that arise when users talk to the virtual assistant by using subjective features, and this paves the way for future developments of our methodology.