Adaptive Retrieval for Reasoning-Intensive Retrieval

2026-01-08

arXiv.org

AI Summary

This work addresses a critical limitation in existing reasoning-intensive retrieval methods: their inability to recall “bridge” documents that, while not directly relevant to the initial query, are essential for multi-hop reasoning, as well as their susceptibility to error propagation from flawed reasoning plans. To overcome these challenges, we propose REPAIR, a novel framework that reformulates the reasoning plan into dense feedback signals and introduces a selective adaptive retrieval mechanism during reranking. This mechanism dynamically refines the reasoning path and precisely retrieves crucial supporting documents. REPAIR establishes an end-to-end reasoning-intensive retrieval system that significantly outperforms current baselines on complex question answering tasks, achieving a 5.6 percentage point improvement in performance.

Technology Category

• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Knowledge Representation and Reasoning: Reasoning with Beliefs
• Search and Optimization: Metareasoning and Metaheuristics

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Graph Algorithms and Modeling for the Web: Querying, indexing, and retrieval in Web-related graphs

Abstract

We study leveraging adaptive retrieval to ensure sufficient"bridge"documents are retrieved for reasoning-intensive retrieval. Bridge documents are those that contribute to the reasoning process yet are not directly relevant to the initial query. While existing reasoning-based reranker pipelines attempt to surface these documents in ranking, they suffer from bounded recall. Naive solution with adaptive retrieval into these pipelines often leads to planning error propagation. To address this, we propose REPAIR, a framework that bridges this gap by repurposing reasoning plans as dense feedback signals for adaptive retrieval. Our key distinction is enabling mid-course correction during reranking through selective adaptive retrieval, retrieving documents that support the pivotal plan. Experimental results on reasoning-intensive retrieval and complex QA tasks demonstrate that our method outperforms existing baselines by 5.6%pt.

Problem

• reasoning-intensive retrieval
• bridge documents
• adaptive retrieval
• recall limitation
• planning error propagation

Innovation

• adaptive retrieval
• reasoning-intensive retrieval
• bridge documents
• reasoning plans
• mid-course correction