Self-Supervised Bot Play for Transcript-Free Conversational Recommendation with Justification

Anonymous

Abstract: Conversational recommender systems offer a way for users to engage in multi-turn conversations to find items they enjoy. Dialog agents for conversational recommendation rely on expensive human dialog transcripts, limiting their usage to domains where such data exists. We develop an alternative, two-part framework for training multi-turn conversational recommenders that accommodate a common paradigm of conversation: experts provide and justify suggestions, while users can critique and respond. We can thus adapt conversational recommendation to a wider range of domains where crowd-sourced ground truth dialogs are not available. First, we train a recommender system to jointly suggest items and justify its reasoning via subjective aspects. We then fine-tune this model to incorporate iterative user feedback via self-supervised bot-play. Experiments on three real-world datasets demonstrate that our system can be applied to different recommendation models across diverse domains to achieve state-of-the-art performance in multi-turn recommendation. Human studies show that systems trained with our framework provide more useful, helpful, and knowledgeable suggestions in warm- and cold-start settings.

Paper Type: long