Graph Retrieval-Augmented LLM for Conversational Recommendation Systems

2025-03-09

AI Summary

Conversational Recommendation Systems (CRS) face three key challenges: sparse user preferences, weak semantic understanding, and insufficient domain knowledge. To address these, we propose G-CRS—a training-free, graph-augmented retrieval-augmented generation framework. First, we construct a product-attribute knowledge graph and employ Graph Neural Networks (GNNs) coupled with Personalized PageRank (PPR) to perform multi-hop preference reasoning over the graph. Second, we integrate structured graph-based context into large language models (LLMs) via graph-aware retrieval-augmented generation (RAG), enabling accurate recommendations without supervision or fine-tuning. Our contributions include: (i) the first training-free graph-enhanced RAG paradigm for CRS; (ii) the first application of PPR to discover contextual relevance in conversational history; and (iii) synergistic GNN–LLM reasoning without any parameter adaptation. On two public benchmarks, G-CRS achieves up to 12.7% absolute improvement in recommendation accuracy over state-of-the-art methods, while eliminating hallucination and domain adaptation overhead.

Technology Category

• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Knowledge Representation and Reasoning: Preferences
• Machine Learning: Learning Preferences or Rankings

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Conversational Recommender Systems (CRSs) have emerged as a transformative paradigm for offering personalized recommendations through natural language dialogue. However, they face challenges with knowledge sparsity, as users often provide brief, incomplete preference statements. While recent methods have integrated external knowledge sources to mitigate this, they still struggle with semantic understanding and complex preference reasoning. Recent Large Language Models (LLMs) demonstrate promising capabilities in natural language understanding and reasoning, showing significant potential for CRSs. Nevertheless, due to the lack of domain knowledge, existing LLM-based CRSs either produce hallucinated recommendations or demand expensive domain-specific training, which largely limits their applicability. In this work, we present G-CRS (Graph Retrieval-Augmented Large Language Model for Conversational Recommender Systems), a novel training-free framework that combines graph retrieval-augmented generation and in-context learning to enhance LLMs' recommendation capabilities. Specifically, G-CRS employs a two-stage retrieve-and-recommend architecture, where a GNN-based graph reasoner first identifies candidate items, followed by Personalized PageRank exploration to jointly discover potential items and similar user interactions. These retrieved contexts are then transformed into structured prompts for LLM reasoning, enabling contextually grounded recommendations without task-specific training. Extensive experiments on two public datasets show that G-CRS achieves superior recommendation performance compared to existing methods without requiring task-specific training.

Problem

• Addresses knowledge sparsity in conversational recommendation systems.
• Enhances semantic understanding and complex preference reasoning.
• Eliminates need for expensive domain-specific training in LLM-based CRSs.

Innovation

• Graph retrieval-augmented generation enhances LLMs.
• Two-stage architecture: GNN and Personalized PageRank.
• Structured prompts enable training-free recommendations.