MSCRS: Multi-modal Semantic Graph Prompt Learning Framework for Conversational Recommender Systems
Yibiao Wei, Jie Zou, Weikang Guo, Guoqing Wang, Xing Xu, Yang Yang
TL;DR
MSCRS addresses the challenge of sparse conversational contexts in CRSs by introducing a multi-modal semantic graph prompt learning framework. It builds three modality-specific graphs (collaborative, textual, image) and fuses them into a unified representation that guides LLM-based recommendation and conversation tasks via prompt learning. The approach combines LightGCN-based graph embeddings, contrastive fusion, and pre-training to predict entities, yielding state-of-the-art results on ReDial and INSPIRED for both item recommendation and response generation. The findings demonstrate that leveraging multi-modal item semantics with graph-structural cues and prompt-based LLMs can significantly improve personalization accuracy and the naturalness of generated conversations, with broad implications for multimodal conversational AI systems.
Abstract
Conversational Recommender Systems (CRSs) aim to provide personalized recommendations by interacting with users through conversations. Most existing studies of CRS focus on extracting user preferences from conversational contexts. However, due to the short and sparse nature of conversational contexts, it is difficult to fully capture user preferences by conversational contexts only. We argue that multi-modal semantic information can enrich user preference expressions from diverse dimensions (e.g., a user preference for a certain movie may stem from its magnificent visual effects and compelling storyline). In this paper, we propose a multi-modal semantic graph prompt learning framework for CRS, named MSCRS. First, we extract textual and image features of items mentioned in the conversational contexts. Second, we capture higher-order semantic associations within different semantic modalities (collaborative, textual, and image) by constructing modality-specific graph structures. Finally, we propose an innovative integration of multi-modal semantic graphs with prompt learning, harnessing the power of large language models to comprehensively explore high-dimensional semantic relationships. Experimental results demonstrate that our proposed method significantly improves accuracy in item recommendation, as well as generates more natural and contextually relevant content in response generation.