Dual-Expert Collaborative Network for Fake News Detection with External Knowledge Integration


Wenxin Luo (Beijing Jiaotong University), Zhu Teng (Beijing Jiaotong Univeristy), Wei Zhang (Beijing Jiaotong University), Baopeng Zhang (Beijing Jiaotong Univeristy)
The 35th British Machine Vision Conference
PDFPosterVideo (Right click to download)Supplementary

Abstract

The pervasive dissemination of misinformation has imposed a profound negative impact on societal structures, making multi-modal fake news detection an urgent research priority. Traditional approaches mainly focus on evaluating the authenticity and consistency of the original news modalities, often neglecting the contextual understanding necessary to avoid cognitive biases in complex scenarios. In this paper, we propose a Dual-Expert Collaborative Network (DECNet) for fake news detection, a novel approach that integrates original news content with external knowledge. Specifically, we leverage multi-scale images to extract visual features and design a Content-Driven Expert (CDE) that performs hierarchical fusion with textual data, enhancing semantic alignment and uncovering deeper insights of news content. Simultaneously, inspired by the reasoning capabilities of Large Vision-Language Models (LVLMs), we propose a Knowledge-Augmentation Expert (KAE) that leverages generated explanations alongside the inherent multi-modal information of the news to capture subtle cues indicative of fake news. To facilitate interaction between these two branches, a gating network is designed for seamless collaboration. Extensive experiments on two widely-employed datasets demonstrate that our method outperforms current state-of-the-art approaches.

Citation

@inproceedings{Luo_2025_BMVC,
author    = {Wenxin Luo and Zhu Teng and Wei Zhang and Baopeng Zhang},
title     = {Dual-Expert Collaborative Network for Fake News Detection with External Knowledge Integration},
booktitle = {36th British Machine Vision Conference 2025, {BMVC} 2025, Sheffield, UK, November 24-27, 2025},
publisher = {BMVA},
year      = {2025},
url       = {https://bmva-archive.org.uk/bmvc/2025/assets/papers/Paper_228/paper.pdf}
}


Copyright © 2025 The British Machine Vision Association and Society for Pattern Recognition
The British Machine Vision Conference is organised by The British Machine Vision Association and Society for Pattern Recognition. The Association is a Company limited by guarantee, No.2543446, and a non-profit-making body, registered in England and Wales as Charity No.1002307 (Registered Office: Dept. of Computer Science, Durham University, South Road, Durham, DH1 3LE, UK).

Imprint | Data Protection