Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Towards Agentic RAG with Deep Reasoning: A Survey of RAG-Reasoning Systems in LLMs

Yangning Li, Weizhi Zhang, Yuyao Yang, Wei-Chieh Huang, Yaozu Wu, Junyu Luo, Yuanchen Bei, Henry Peng Zou, Xiao Luo, Yusheng Zhao, Chunkit Chan, Yankai Chen, Zhongfen Deng, Yinghui Li, Hai-Tao Zheng, Dongyuan Li, Renhe Jiang, Ming Zhang, Yangqiu Song, Philip S. Yu

TL;DR

This survey analyzes three evolutionary trajectories in aligning retrieval with reasoning in LLMs: Reasoning-Enhanced RAG, which injects reasoning into RAG stages; RAG-Enhanced Reasoning, which grounds long Chain-of-Thought with external knowledge; and Synergized RAG-Reasoning, where iterative, agentic loops jointly optimize search and inference. It provides a unified taxonomy across retrieval optimization, integration, and generation, then extends to external knowledge sources, in-context cues, and agent orchestration in deep research systems. Key contributions include a comprehensive framework, an extensive benchmark panorama, and open challenges for efficiency, multimodality, trust, and human collaboration. The work highlights practical implications for building robust, knowledge-grounded, and transparent LLM systems capable of complex, knowledge-intensive reasoning.

Abstract

Retrieval-Augmented Generation (RAG) lifts the factuality of Large Language Models (LLMs) by injecting external knowledge, yet it falls short on problems that demand multi-step inference; conversely, purely reasoning-oriented approaches often hallucinate or mis-ground facts. This survey synthesizes both strands under a unified reasoning-retrieval perspective. We first map how advanced reasoning optimizes each stage of RAG (Reasoning-Enhanced RAG). Then, we show how retrieved knowledge of different type supply missing premises and expand context for complex inference (RAG-Enhanced Reasoning). Finally, we spotlight emerging Synergized RAG-Reasoning frameworks, where (agentic) LLMs iteratively interleave search and reasoning to achieve state-of-the-art performance across knowledge-intensive benchmarks. We categorize methods, datasets, and open challenges, and outline research avenues toward deeper RAG-Reasoning systems that are more effective, multimodally-adaptive, trustworthy, and human-centric. The collection is available at https://github.com/DavidZWZ/Awesome-RAG-Reasoning.

Towards Agentic RAG with Deep Reasoning: A Survey of RAG-Reasoning Systems in LLMs

TL;DR

This survey analyzes three evolutionary trajectories in aligning retrieval with reasoning in LLMs: Reasoning-Enhanced RAG, which injects reasoning into RAG stages; RAG-Enhanced Reasoning, which grounds long Chain-of-Thought with external knowledge; and Synergized RAG-Reasoning, where iterative, agentic loops jointly optimize search and inference. It provides a unified taxonomy across retrieval optimization, integration, and generation, then extends to external knowledge sources, in-context cues, and agent orchestration in deep research systems. Key contributions include a comprehensive framework, an extensive benchmark panorama, and open challenges for efficiency, multimodality, trust, and human collaboration. The work highlights practical implications for building robust, knowledge-grounded, and transparent LLM systems capable of complex, knowledge-intensive reasoning.

Abstract

Retrieval-Augmented Generation (RAG) lifts the factuality of Large Language Models (LLMs) by injecting external knowledge, yet it falls short on problems that demand multi-step inference; conversely, purely reasoning-oriented approaches often hallucinate or mis-ground facts. This survey synthesizes both strands under a unified reasoning-retrieval perspective. We first map how advanced reasoning optimizes each stage of RAG (Reasoning-Enhanced RAG). Then, we show how retrieved knowledge of different type supply missing premises and expand context for complex inference (RAG-Enhanced Reasoning). Finally, we spotlight emerging Synergized RAG-Reasoning frameworks, where (agentic) LLMs iteratively interleave search and reasoning to achieve state-of-the-art performance across knowledge-intensive benchmarks. We categorize methods, datasets, and open challenges, and outline research avenues toward deeper RAG-Reasoning systems that are more effective, multimodally-adaptive, trustworthy, and human-centric. The collection is available at https://github.com/DavidZWZ/Awesome-RAG-Reasoning.

Paper Structure

This paper contains 35 sections, 2 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Background and Preliminary
  3. Reasoning-Enhanced RAG
  4. Retrieval Optimization
  5. Reasoning-Aware Query Reformulation
  6. Retrieval Strategy and Planning
  7. Retrieval Model Enhancement
  8. Integration Enhancement
  9. Relevance Assessment & Filtering
  10. Information Synthesis & Fusion
  11. Generation Enhancement
  12. Context-Aware Synthesis Strategies
  13. Grounded Generation Control
  14. RAG-Enhanced Reasoning
  15. External Knowledge Retrieval
  16. ...and 20 more sections

Figures (2)

  • Figure 1: Overview of the RAG-Reasoning System. The Reasoning-Enhanced RAG methods and RAG-Enhanced Reasoning methods represent one-way enhancements. In contrast, the Synergized RAG-Reasoning System performs reasoning and retrieval iteratively, enabling mutual enhancements.
  • Figure 2: Taxonomy of Recent Advances in RAG-Reasoning System.