MCFEND
所属分类:数值算法/人工智能
开发工具:Python
文件大小:0KB
下载次数:0
上传日期:2024-02-07 06:29:50
上 传 者:
sh-1993
说明: “MCFEND:一个用于中文假新闻检测的多源基准数据集”的PyTorch官方实现
(Official PyTorch Implementation of "MCFEND: A Multi-source Benchmark Dataset for Chinese Fake News Detection")
文件列表:
modal fusion-based model/
multi-modal model/
tree-based model/
uni-modal model/
LICENSE
# MCFEND
The repository "MCFEND: A Multi-source Benchmark Dataset for Chinese Fake News Detection" includes the dataset and code.
- **Dataset**: The MCFEND dataset is the first multi-source Chinese fake news detection dataset, which comprises multi-modal content and social context of 23,974 Chinese news pieces collected from 14 fact-checking agencies.
- **Code**: The implementation of the baselines and evaluations is based on PyTorch.
## Dataset.
The dataset is available at: [Google Drive](https://drive.google.com/drive/folders/1tflhQTkMT_gTTwEw3ESfKS7Sr5w__5u5?usp=sharing)
## Code
### Evaluation Settings
In this work, we conducted *cross-source* and *multi-source* evaluations.
*Cross-source*: In this setting, baseline systems were exclusively trained using news sourced from Weibo and their corresponding social context in the train set of the MCFEND dataset.
*Multi-source*: In this setting, baseline systems were trained with the complete train set of the MCFEND dataset, including news from all sources covered in our dataset and their corresponding social contexts.
### Structure
The codes for four types of baseline systems are organized in the following structure:
```
- ./uni-modal model/
- ./multi-modal model/
- ./tree-based model/
- ./modal fusion-based model/
```
## Citation
If you utilize this repository or dataset, please consider citing our paper presented at WWW 2024.
```
@INPROCEEDINGS{mcfend,
title={MCFEND: A Multi-source Benchmark Dataset for Chinese Fake News Detection},
author={Li, Yupeng and He, Haorui and Bai, Jin and Wen, Dacheng},
booktitle={Proc.~of WWW (to apear)},
year={2024},
}
```
## Acknowledgement
The implementation of this repository is based on the following repos:
- [BERT and RoBERTa](https://github.com/649453932/Bert-Chinese-Text-Classification-Pytorch)
- [CLIP](https://huggingface.co/docs/transformers/model_doc/clip)
- [BERT-EMO](https://github.com/RMSnow/WWW2021)
- [CAFE](https://github.com/cyxanna/CAFE)
- [dEFEND](https://github.com/cuilimeng/dEFEND-web)
- [Tree-RvNN and Tree-Transformer](https://github.com/majingCUHK/Rumor_RvNN)
近期下载者:
相关文件:
收藏者: