Quin

An easy to use framework for large-scale fact-checking and question answering. [Demo]

Usage

The project was tested with Python 3.7. For the setup and execution:

1. Download the model weights and extract them into the models/weights folder:

• NLI model and
• Dense Encoder M (multitask for QA and Fact-Checking) or
• Dense Encoder FC (single-task for Fact-Checking) or
• Dense Encoder QA (single-task for Question Answering)

2. Install the required packages:

   pip3 install -r requirements.txt

3. Index a list of documents:

   python quin.py --index example_docs.jsonl

4. Serve a Flask API:

   python quin.py --port 1234

Datasets

• Factual-NLI Dataset
• COVID-19 news dataset

References

@inproceedings{samarinas2021improving,
  title={Improving Evidence Retrieval for Automated Explainable Fact-Checking},
  author={Samarinas, Chris and Hsu, Wynne and Lee, Mong Li},
  booktitle={Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies: Demonstrations},
  pages={84--91},
  year={2021}
}

@inproceedings{samarinas2020latent,
  title={Latent Retrieval for Large-Scale Fact-Checking and Question Answering with NLI training},
  author={Samarinas, Chris and Hsu, Wynne and Lee, Mong Li},
  booktitle={2020 IEEE 32nd International Conference on Tools with Artificial Intelligence (ICTAI)},
  pages={941--948},
  year={2020},
  organization={IEEE}
}

License

Quin is licensed under MIT License, and the Factual-NLI dataset under Attribution 4.0 International (CC BY 4.0) license.