Code repository for the paper titled:
Empowering the Fact-checkers! Automatic Identification of Claim Spans on Twitter. Megha Sundriyal, Atharva Kulkarni, Vaibhav Pulastya, Md Shad Akhtar, Tanmoy Chakraborty
Accepted at the 2022 Conference on Empirical Methods in Natural Language Processing (EMNLP'22), Abu Dhabi, December 7–11, 2022.
Directions to implement the code:
- Create and activate a virtual environment.
- Install all the dependencies from the requirements.txt file: pip install -r requirements.txt
- Add all datasets in the dataset folder (train, test, and validation). The dataset folder already contains the claim descriptions embeddings from RoBERTa.
- Run python daberta.py to run the code: python daberta.py
- The model for each epoch will get saved in the models folder.