AFMRL: Attribute-Enhanced Fine-Grained Multi-Modal Representation Learning in E-commerce

arXiv cs.CL / 4/23/2026

📰 NewsModels & Research

共有:

Key Points

The paper introduces AFMRL, an approach to improve fine-grained multimodal understanding for e-commerce tasks like identical product retrieval.
AFMRL formulates fine-grained product comprehension as an attribute generation problem using multimodal LLMs, extracting key attributes from both product images and text.
It uses a two-stage training strategy: Attribute-Guided Contrastive Learning (AGCL) to focus learning on hard samples while reducing noisy false negatives, and Retrieval-aware Attribute Reinforcement (RAR) to refine attribute generation using retrieval performance as a reward signal.
Experiments on large-scale e-commerce datasets show AFMRL achieves state-of-the-art results across multiple downstream retrieval tasks, supporting the idea of leveraging generative models for fine-grained representation learning.

Abstract

Multimodal representation is crucial for E-commerce tasks such as identical product retrieval. Large representation models (e.g., VLM2Vec) demonstrate strong multimodal understanding capabilities, yet they struggle with fine-grained semantic comprehension, which is essential for distinguishing highly similar items. To address this, we propose Attribute-Enhanced Fine-Grained Multi-Modal Representation Learning (AFMRL), which defines product fine-grained understanding as an attribute generation task. It leverages the generative power of Multimodal Large Language Models (MLLMs) to extract key attributes from product images and text, and enhances representation learning through a two-stage training framework: 1) Attribute-Guided Contrastive Learning (AGCL), where the key attributes generated by the MLLM are used in the image-text contrastive learning training process to identify hard samples and filter out noisy false negatives. 2) Retrieval-aware Attribute Reinforcement (RAR), where the improved retrieval performance of the representation model post-attribute integration serves as a reward signal to enhance MLLM's attribute generation during multimodal fine-tuning. Extensive experiments on large-scale E-commerce datasets demonstrate that our method achieves state-of-the-art performance on multiple downstream retrieval tasks, validating the effectiveness of harnessing generative models to advance fine-grained representation learning.