AirRAG: Autonomous Strategic Planning and Reasoning Steer Retrieval Augmented Generation

TL;DR

AirRAG introduces an autonomous strategic planning framework for retrieval-augmented generation, expanding reasoning capabilities with Monte Carlo Tree Search to improve performance on complex tasks.

Contribution

It presents a novel integration of autonomous strategic planning with reasoning actions in RAG, utilizing MCTS to broaden solution space and enhance reasoning in LLMs.

Findings

Significant performance improvements on complex QA datasets.

Effective expansion of reasoning space via MCTS.

Enhanced flexibility and integration with other models.

Abstract

Leveraging the autonomous decision-making capabilities of large language models (LLMs) has demonstrated superior performance in reasoning tasks. However, despite the success of iterative or agentic retrieval-augmented generation (RAG) techniques, these methods are often constrained to a single solution space when confronted with complex problems. In this paper, we propose a novel thinking pattern in RAG that integrates autonomous strategic planning with efficient reasoning actions, significantly activating intrinsic reasoning capabilities and expanding the solution space of specific tasks via Monte Carlo Tree Search (MCTS), which we refer to as AirRAG. Specifically, our approach designs five fundamental reasoning actions, which are expanded to a broad tree-based reasoning space using MCTS. The approach also incorporates self-consistency verification to explore potential reasoning paths and inference scaling law. Additionally, computationally optimal strategies are employed to allocate more inference resources to key actions, thereby enhancing overall performance. Experimental results demonstrate the effectiveness of AirRAG, showing significant performance gains on complex question-answering datasets. Furthermore, AirRAG is flexible and lightweight, making it easy to integrate with other advanced technologies and models.