QReason cuts redundant reasoning to speed up passage reranking
QReason: Query-Focused Decoupled Chain-of-Thought for Efficient Passage Reranking
Information Retrieval
Summary
When computers look for the best answers in lots of text passages, they often waste time repeating the same thought process over and over. The authors created QReason, which separates understanding the question from checking each passage. This way, the computer figures out the main question once, then quickly ranks many passages without repeating itself. Their tests show QReason works as well as methods that do more repeated thinking but much faster and more efficiently.
What this means in practice
- •For search engine engineers: Use QReason to speed up relevance ranking of text snippets while maintaining high accuracy, reducing computational cost in large-scale search systems.
- •For digital assistant developers: Integrate QReason to improve quick, relevant passage selection for user queries without incurring delays from repeated reasoning steps.
Authors
Yang Zhang, Wenhan Liu, Qiannan Zhu, Mingming Li, Yuanfei Huang
Abstract
Passage reranking plays a crucial role in information retrieval by refining the ordering of candidate passages to better reflect relevance. Existing listwise LLM rerankers with Chain-of-Thought (CoT) reasoning can handle complex queries effectively, but they suffer from substantial redundancy and high latency due to sliding-window strategies, which repeatedly generate highly similar CoTs. To address this, we propose QReason, a decoupled framework that separates query-focused reasoning from window-specific passage relevance assessment. Specifically, QReason introduces a dedicated rewriter that generates a ranking-oriented reasoning query once, capturing the query's core intent while avoiding redundant reasoning, and then reuses it across all windows with a non-reasoning reranker. The rewriter is trained via a two-stage process that first uses supervised fine-tuning with relevant-passage guidance through semantic evidence to produce deeply grounded, query-focused CoTs. It then applies reinforcement learning to align CoT generation with both the inference-time setting and the reranking objective, optimizing listwise metrics and passage-level discrimination to produce reusable reasoning chains for reranking. Experiments on the BRIGHT benchmark demonstrate that QReason significantly reduces redundant reasoning, achieves ranking performance comparable to or better than strong reasoning-based rerankers, and outperforms existing query rewriting models.