RAG-Fusion Based Information Retrieval for Fact-Checking

Yuki Momii, Tetsuya Takiguchi, Yasuo Ariki


Abstract
Fact-checking involves searching for relevant evidence and determining whether the given claim contains any misinformation. In this paper, we propose a fact verification system based on RAG-Fusion. We use GPT-4o to generate questions from the claim, which helps improve the accuracy of evidence retrieval.Additionally, we adopt GPT-4o for the final judgment module and refine the prompts to enhance the detection accuracy, particularly when the claim contains misinformation. Experiment showed that our system achieved an AVeriTeC Score of 0.3865 on the AVeriTeC test data, significantly surpassing the baseline score of 0.11.
Anthology ID:
2024.fever-1.4
Volume:
Proceedings of the Seventh Fact Extraction and VERification Workshop (FEVER)
Month:
November
Year:
2024
Address:
Miami, Florida, USA
Editors:
Michael Schlichtkrull, Yulong Chen, Chenxi Whitehouse, Zhenyun Deng, Mubashara Akhtar, Rami Aly, Zhijiang Guo, Christos Christodoulopoulos, Oana Cocarascu, Arpit Mittal, James Thorne, Andreas Vlachos
Venue:
FEVER
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
47–54
Language:
URL:
https://aclanthology.org/2024.fever-1.4
DOI:
Bibkey:
Cite (ACL):
Yuki Momii, Tetsuya Takiguchi, and Yasuo Ariki. 2024. RAG-Fusion Based Information Retrieval for Fact-Checking. In Proceedings of the Seventh Fact Extraction and VERification Workshop (FEVER), pages 47–54, Miami, Florida, USA. Association for Computational Linguistics.
Cite (Informal):
RAG-Fusion Based Information Retrieval for Fact-Checking (Momii et al., FEVER 2024)
Copy Citation:
PDF:
https://aclanthology.org/2024.fever-1.4.pdf