QACheck

Data and Codes for "QACHECK: A Demonstration System for Question-Guided Multi-Hop Fact-Checking" (EMNLP 2023, System Demonstrations).

System Overview

We introduce the Question-guided Multi-hop Fact-Checking (QACheck) system, which provides an explainable fact-checking process by asking and answering a series of relevant questions.

• Claim Verifier $\mathcal{D}$: determine the sufficiency of the existing context to validate the claim, i.e., $\mathcal{D}(c, C) \rightarrow {\text{True}, \text{False}}$.

• Question Generator $\mathcal{Q}$: generate the next question that is necessary for verifying the claim, i.e., $Q(c, C) \rightarrow q$.

• Question-Answering Model $\mathcal{A}$: answer the question and provide the supported evidence, i.e., $\mathcal{A}(q) \rightarrow a, e$.

• Validator $\mathcal{V}$: validate the usefulness of the newly-generated (Q, A) pair based on the existing context and the claim, i.e., $\mathcal{V}(c, {q, a}, C) \rightarrow {\text{True}, \text{False}}$.

• Reasoner $\mathcal{R}$: utilize the relevant context to justify the veracity of the claim and outputs the final label, i.e., $\mathcal{R}(c, C) \rightarrow {\text{Supported}, \text{Refuted}}$.

Demo System

Clone the github to your local machine and install the required packages.

pip install flask
pip install openai
pip install backoff

Run the demo system.

python run-demo.py \
    --model_name <gpt-4 or gpt-3.5-turbo> \
    --API_KEY <Your OpenAI API key> \

Reference

Please cite the paper in the following format if you use this dataset during your research.

@inproceedings{PanQACheck23,
  author       = {Liangming Pan, Xinyuan Lu, Min-Yen Kan, Preslav Nakov},
  title        = {QACHECK: A Demonstration System for Question-Guided Multi-Hop Fact-Checking},
  booktitle    = {Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing System Demonstrations Track (EMNLP 2023 Demo Track)},
  address      = {Singapore},
  year         = {2023},
  month        = {Dec}
}

Q&A

If you encounter any problem, please either directly contact the Liangming Pan or leave an issue in the github repo.