BTIC

This repository configured for paper: Supervised Contrastive Learning for Multimodal Unreliable News Detection in COVID-19 Pandemic

Wenjia Zhang, Lin Gui, Yulan He

In this work, we propose a BERT-based multimodal unreliable news detection framework, which captures both textual and visual information from unreliable articles utilising the contrastive learning strategy. The contrastive learner interacts with the unreliable news classifier to push similar credible news (or similar unreliable news) closer while moving news articles with similar content but opposite credibility labels away from each other in the multimodal embedding space. Experimental results on a COVID-19 related dataset, ReCOVery, show that our model outperforms a number of competitive baseline in unreliable news detection.

Figure 1: The framework of our proposed approach.

Figure 2: The heat map of attention weights from the interaction layer of the transformer between text and image (resized by bilinear interpolation). In text, words highlighted in blue have higher attentions weights with the ‘unreliable’ category, while those in red are more closely related to the ‘reliable’ category.

Requirements

pandas
numpy
os
PIL
torch
space
sklearn
matplotlib
seaborn

Citation

If you think our work is useful, please cite as:

@inproceedings{zhang2021supervised,
  title={Supervised Contrastive Learning for Multimodal Unreliable News Detection in COVID-19 Pandemic},
  author={Zhang, Wenjia and Gui, Lin and He, Yulan},
  booktitle={30th ACM International Conference on Information and Knowledge Management (CIKM)},
  year={2021}
}

Thank you.

Contact

If you have any question on the paper,or the code, please contact wenjia.zhang@warwick.ac.uk