REFIND at SemEval-2025 Task 3: Retrieval-Augmented Factuality Hallucination Detection in Large Language Models
This work addresses reliability issues in LLMs for knowledge-intensive tasks like question answering, offering a domain-specific improvement in hallucination detection.
The paper tackles the problem of hallucination detection in large language model outputs by introducing REFIND, a retrieval-augmented framework that uses a novel Context Sensitivity Ratio metric to quantify sensitivity to retrieved evidence, achieving superior IoU scores and robustness across nine languages, including low-resource settings.
Hallucinations in large language model (LLM) outputs severely limit their reliability in knowledge-intensive tasks such as question answering. To address this challenge, we introduce REFIND (Retrieval-augmented Factuality hallucINation Detection), a novel framework that detects hallucinated spans within LLM outputs by directly leveraging retrieved documents. As part of the REFIND, we propose the Context Sensitivity Ratio (CSR), a novel metric that quantifies the sensitivity of LLM outputs to retrieved evidence. This innovative approach enables REFIND to efficiently and accurately detect hallucinations, setting it apart from existing methods. In the evaluation, REFIND demonstrated robustness across nine languages, including low-resource settings, and significantly outperformed baseline models, achieving superior IoU scores in identifying hallucinated spans. This work highlights the effectiveness of quantifying context sensitivity for hallucination detection, thereby paving the way for more reliable and trustworthy LLM applications across diverse languages. Our code is available at https://github.com/oneonlee/REFIND.