@inproceedings{27de517320524e6f9c022548a8d1c572,
title = "Progressive Fusion of LLM-Augmented Multi-Modal Data for Enhanced Product Recommendations",
abstract = "Recommender systems in E-commerce often face challenges such as data sparsity and cold-start problems, which hinder their ability to provide accurate and personalized recommendations. To address these issues, we propose Progressive Fusion Muti-modal Graph Convolution Networks (PFMGCN), a novel framework tailored for product recommendations. In PFMGCN, we first employ a Large Language Model (LLM) to generate more product attribute data for enriching contextual information. A knowledge graph (KG) is then constructed with augmented nodes, and their embeddings are extracted via pre-trained models. A cross-attention module is used to dynamically fuse textual and visual features, capturing fine-grained product details. Progressive graph convolutional networks integrated with multi-modal cross-attention are then applied to extract deep features and deliver accurate recommendations. Extensive experiments on real-world E-commerce datasets demonstrate that our approach significantly outperforms state-of-the-art methods, achieving superior accuracy in product recommendation tasks.",
keywords = "cross-attention, knowledge graph, large language model, multi-modality, recommender system",
author = "Xinhao Li and Lin Gong and Xin Liu and Xi Xiang",
note = "Publisher Copyright: {\textcopyright}2025 IEEE.; 6th IEEE International Seminar on Artificial Intelligence, Networking and Information Technology, AINIT 2025 ; Conference date: 11-04-2025 Through 13-04-2025",
year = "2025",
doi = "10.1109/AINIT65432.2025.11035374",
language = "English",
series = "2025 IEEE 6th International Seminar on Artificial Intelligence, Networking and Information Technology, AINIT 2025",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "2047--2050",
booktitle = "2025 IEEE 6th International Seminar on Artificial Intelligence, Networking and Information Technology, AINIT 2025",
address = "United States",
}