E-MMKGR: A Unified Multimodal Knowledge Graph Framework for E-commerce Applications

2026-02-24

AI Summary

Existing cross-modal recommendation systems are limited by fixed modality sets and task-specific objectives, hindering their scalability and generalization to multi-task scenarios. To address this, this work proposes E-MMKG, an extensible and unified multimodal knowledge graph framework tailored for e-commerce. By integrating multi-source product information through graph neural networks and knowledge graph refinement techniques, E-MMKG constructs a structured semantic graph to learn a shared, unified product representation across tasks. This approach effectively decouples modality dependencies from task-specific requirements. Extensive experiments on real-world Amazon datasets demonstrate significant performance gains: the proposed method achieves up to a 10.18% improvement in Recall@10 for recommendation tasks and enhances product search effectiveness by up to 21.72% compared to vector-based retrieval baselines.

Technology Category

• Machine Learning: Multimodal Learning
• Data Mining & Knowledge Management: Mining of Visual, Multimedia & Multimodal Data
• Computer Vision: Multi-modal Vision

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Multimodal recommender systems (MMRSs) enhance collaborative filtering by leveraging item-side modalities, but their reliance on a fixed set of modalities and task-specific objectives limits both modality extensibility and task generalization. We propose E-MMKGR, a framework that constructs an e-commerce-specific Multimodal Knowledge Graph E-MMKG and learns unified item representations through GNN-based propagation and KG-oriented optimization. These representations provide a shared semantic foundation applicable to diverse tasks. Experiments on real-world Amazon datasets show improvements of up to 10.18% in Recall@10 for recommendation and up to 21.72% over vector-based retrieval for product search, demonstrating the effectiveness and extensibility of our approach.

Problem

• Multimodal recommender systems
• modality extensibility
• task generalization
• e-commerce
• knowledge graph

Innovation

• Multimodal Knowledge Graph
• Graph Neural Network
• Unified Item Representation
• E-commerce Recommendation
• Task Generalization