CP-Rec: Contextual Prompting for Conversational Recommender Systems

The conversational recommender system (CRS) aims to provide high-quality recommendations through interactive dialogues. However, previous CRS models have no effective mechanisms for task planning and topic elaboration, thus they hardly maintain coherence in multi-task recommendation Inspired by recent advances prompt-based learning, we propose a novel contextual prompting framework dialogue management, which optimizes prompts based on context, topics, user profiles. Specifically, develop controller sequentially plan the subtasks, prompt search module construct context-aware prompts. We further adopt external knowledge enrich profiles make knowledge-aware recommendations. Incorporating these techniques, with prompting, namely CP-Rec. Experimental results demonstrate that it achieves state-of-the-art accuracy generates more coherent informative conversations.