Learning to Reason for Hallucination Span Detection
This work addresses the need for more precise hallucination detection in real-world applications, though it is incremental as it builds on existing reasoning and reinforcement learning methods.
The paper tackles the problem of detecting hallucinated spans in large language model outputs by proposing RL4HS, a reinforcement learning framework with span-level rewards, which outperforms pretrained reasoning models and supervised fine-tuning on the RAGTruth benchmark across summarization, question answering, and data-to-text tasks.
Large language models (LLMs) often generate hallucinations -- unsupported content that undermines reliability. While most prior works frame hallucination detection as a binary task, many real-world applications require identifying hallucinated spans, which is a multi-step decision making process. This naturally raises the question of whether explicit reasoning can help the complex task of detecting hallucination spans. To answer this question, we first evaluate pretrained models with and without Chain-of-Thought (CoT) reasoning, and show that CoT reasoning has the potential to generate at least one correct answer when sampled multiple times. Motivated by this, we propose RL4HS, a reinforcement learning framework that incentivizes reasoning with a span-level reward function. RL4HS builds on Group Relative Policy Optimization and introduces Class-Aware Policy Optimization to mitigate reward imbalance issue. Experiments on the RAGTruth benchmark (summarization, question answering, data-to-text) show that RL4HS surpasses pretrained reasoning models and supervised fine-tuning, demonstrating the necessity of reinforcement learning with span-level rewards for detecting hallucination spans.