Classification of COVID-19 and political misinformation on social media

This is the code base for "Classification of COVID-19 and political misinformation on social media", the final project of CS7650 Natural Language at Georgia Tech.

Folder structure

CS7650_final.ipynb: contains code for fine-tuning BERT.

KNN.ipynb: contains code for kNN.

data/: contains the preprocessed train/valid/test splits of the four misinformation datasets we used in our project.

Datasets

We used four misinformation datasets in our project. The data sources are listed as below:

• LIAR: https://www.cs.ucsb.edu/~william/data/liar_dataset.zip
• Political News: https://github.com/rpitrust/fakenewsdata
• Rumor: https://github.com/MickeysClubhouse/COVID-19-rumor-dataset
• CONSTRAINT: https://constraint-shared-task-2021.github.io/

We have proccessed the raw data provided in the datasets and organized them into data/.