Retro*: Optimizing LLMs for Reasoning-Intensive Document Retrieval

2025-09-29

AI Summary

Addressing the challenge of retrieving implicitly relevant documents for complex tasks—where task-document semantic relationships are indirect and require fine-grained reasoning—this paper proposes Retro*. Its core contributions are threefold: (1) a rubric-based, interpretable relevance assessment mechanism that grounds relevance judgments in explicit evaluation criteria; (2) a multi-reasoning-path fusion strategy leveraging test-time trajectory ensembling to enhance retrieval robustness; and (3) a dual-composite reward reinforcement learning algorithm that jointly optimizes both reasoning processes and final retrieval outcomes. Evaluated on the BRIGHT benchmark, Retro* substantially outperforms state-of-the-art information retrieval (IR) and retrieval-augmented generation (RAG) methods, achieving new SOTA performance. It simultaneously delivers high accuracy, strong interpretability, computational efficiency, and favorable scalability.

Technology Category

• Search and Optimization: Learning to Search
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Computer Vision: Image and Video Retrieval

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

With the growing popularity of LLM agents and RAG, it has become increasingly important to retrieve documents that are essential for solving a task, even when their connection to the task is indirect or implicit. Addressing this problem requires fine-grained reasoning to accurately assess the relevance between the task and each candidate document. This capability, however, poses a significant challenge for existing IR techniques. Despite recent progress in reasoning-enhanced IR, existing approaches still face significant challenges in applicability, scalability, and efficiency. In this work, we propose Retro*, a novel approach for reasoning-intensive document retrieval. Our method introduces a rubric-based relevance scoring mechanism, enabling the model to reason about the relationship between a task and a document based on explicitly defined criteria, whereby producing a fine-grained, interpretable relevance score. Retro* also supports test-time scaling by combining multiple reasoning trajectories via score integration, which produces more reliable relevance estimates. To optimize Retro*'s reasoning capabilities, we introduce a novel reinforcement learning algorithm tailored for its relevance scoring mechanism, which employs two composite rewards to fully exploit the trajectories of each training sample. Our experiments show that Retro* outperforms existing document retrieval methods with notable advantages, leading to state-of-the-art performance on the BRIGHT benchmark.

Problem

• Optimizing reasoning-intensive document retrieval for LLM agents
• Addressing scalability and efficiency challenges in relevance assessment
• Improving fine-grained interpretable scoring for indirect document relationships

Innovation

• Rubric-based scoring for fine-grained relevance reasoning
• Test-time scaling via multi-trajectory score integration
• Reinforcement learning with composite rewards for optimization