Research on Personalized Financial Product Recommendation by Integrating Large Language Models and Graph Neural Networks

2025-06-06

AI Summary

To address the challenges of modeling user implicit preferences and capturing complex product relationships in financial recommendation—leading to suboptimal accuracy and interpretability—this paper proposes an LLM-GNN fusion framework built upon a heterogeneous graph. Our method innovatively designs a joint text-semantic and graph-structural message-passing mechanism to enable cross-modal feature alignment, and introduces a multi-source encoder that end-to-end integrates user reviews, social connections, and item attributes. Semantic representations are extracted via LLaMA or BERT, while adaptive heterogeneous graph neural networks model intricate relational patterns. Evaluated on both public and real-world financial datasets, our approach achieves average improvements of 12.7% in accuracy, recall, and NDCG over unimodal baselines. It delivers both high predictive performance and inherent interpretability, and is designed for industrial-scale deployment.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

With the rapid growth of fintech, personalized financial product recommendations have become increasingly important. Traditional methods like collaborative filtering or content-based models often fail to capture users' latent preferences and complex relationships. We propose a hybrid framework integrating large language models (LLMs) and graph neural networks (GNNs). A pre-trained LLM encodes text data (e.g., user reviews) into rich feature vectors, while a heterogeneous user-product graph models interactions and social ties. Through a tailored message-passing mechanism, text and graph information are fused within the GNN to jointly optimize embeddings. Experiments on public and real-world financial datasets show our model outperforms standalone LLM or GNN in accuracy, recall, and NDCG, with strong interpretability. This work offers new insights for personalized financial recommendations and cross-modal fusion in broader recommendation tasks.

Problem

• Enhancing personalized financial product recommendation accuracy
• Capturing latent user preferences and complex relationships
• Integrating LLMs and GNNs for cross-modal feature fusion

Innovation

• Integrates LLMs and GNNs for recommendations
• Fuses text and graph data via GNNs
• Tailored message-passing optimizes hybrid embeddings