Fighting against Fake News on Newly-Emerging Crisis: A Case Study of COVID-19

• Migyeong Yang, Chaewon Park, Jiwon Kang, Daeun Lee, Daejin Choi, Jinyoung Han, Fighting against Fake News on Newly-Emerging Crisis: A Case Study of COVID-19, ACM Web Conference 2024 (WWW '24 Companion).

Dataset

This folder includes the whole dataset that we collected.

Components

• COVID-19 Fake News Claim
  • FakeCovid
  • CoAID
• Non-COVID-19 Fake News Claim
  • Crawled from Politifact and Snopes
• Topic Keyword
  • Extracted through BERT-NER and KeyBERT
• Propagation Data
  • Crawled from YouTube with GoogleAPI

File Explanation

1. [train_ex1.pickle] Data used for training in Experiment 1, containing non-COVID-19 news claims.
2. [train_ex2.pickle] Data used for training in Experiment 2, containing non-COVID-19 news claims and COVID-19 news claims published before April 1, 2020.
3. [test.pickle] Data used for testing in Experiment 1 and Experiment 2, containing COVID-19 news claims published after April 1, 2020.

Data

This folder contains only the data necessary for model training.

File Explanation

1. [train_ex1.pkl] A version extracted from './dataset/train_ex1.pickle', including news claims (title), ner_onehot, titles and descriptions of two YouTube videos, and labels.
2. [train_ex2.pkl] A version extracted from './dataset/train_ex2.pickle', including news claims (title), ner_onehot, titles and descriptions of two YouTube videos, and labels.
3. [test.pkl] A version extracted from './dataset/test.pickle', including news claims (title), ner_onehot, titles and descriptions of two YouTube videos, and labels.
4. [train_ex1.npy / train_ex2.npy / test.npy] These are numpy-formatted adjacency matrices required for training and testing.

Model

This folder consists of Python scripts for the model.

Get Started

Python 3.7 & Pytorch 1.10.1

pip install -r requirements.txt

Run

For Exp. 1 (Without COVID-19 Data)

python3 ./model/main-ex1.py

For Exp. 2 (With a few COVID-19 Data)

python3 ./model/main-ex2.py