Unofficial implementation of (multi-modal) misinformation papers. Please note there may be some changes in the code for the use of the models in real cases. If you are working on research papers, please refer to the official implementations for fair comparison.
mvae.py: Multimodal Generative Models for Scalable Weakly-Supervised Learning- EANN: EANN: Event Adversarial Neural Networks for Multi-Modal Fake News Detection
spotfake.py: SpotFake: A Multi-modal Framework for Fake News Detectionbtic.py: Supervised Contrastive Learning for Multimodal Unreliable News Detection in COVID-19 Pandemic. No contrastive loss due to the lack of timestamp label
Continously update the COVID-19 misinformation datasets below.
| Dataset | Total Data | Misinformation Type | Misinformation Data | Non-Misinformation Type | Non-Misinformation Data |
|---|---|---|---|---|---|
| ESOC COVID-19 Misinformation Dataset | 5636 | 12 | 5376 | 0 | 0 |
| COVID-19 Rumor Dataset | 5279 | 1 | 3580 | 1 | 1699 |
| CONSTRAINT-AAAI21 | 10700 | 1 | 5100 | 1 | 5600 |
| CHECKED: Chinese COVID-19 fake news dataset | |||||
| COVID-19 FAKE NEWS INFODEMIC RESEARCH DATASET | |||||
| COVID-Related Misinformation Videos | |||||
| COVID-19 Healthcare Misinformation Dataset |
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent