ClueAnchor: Clue-Anchored Knowledge Reasoning Exploration and Optimization for Retrieval-Augmented Generation

TL;DR

ClueAnchor tackles factuality gaps in retrieval-augmented generation by grounding reasoning on key clues extracted from retrieved documents. It couples Knowledge Reasoning Exploration (three path types: Internal, External, and Clue-Anchored) with Knowledge Reasoning Optimization that uses reward-based preference signals to select and refine the most effective reasoning path. By explicitly predicting and validating clues and training via Direct Preference Optimization, ClueAnchor achieves improved faithfulness and robustness, particularly under noisy retrieval. Empirical results across diverse QA benchmarks show consistent wins over strong baselines and robust performance under varying knowledge conditions, with evidence that clue-grounded reasoning enhances both grounding and interpretability.

Abstract

Retrieval-Augmented Generation (RAG) augments Large Language Models (LLMs) with external knowledge to improve factuality. However, existing RAG systems frequently underutilize the retrieved documents, failing to extract and integrate the key clues needed to support faithful and interpretable reasoning, especially in cases where relevant evidence is implicit, scattered, or obscured by noise. To address this issue, we propose ClueAnchor, a novel framework for enhancing RAG via clue-anchored reasoning exploration and optimization. ClueAnchor extracts key clues from retrieved content and generates multiple reasoning paths based on different knowledge configurations, optimizing the model by selecting the most appropriate reasoning path for the given context through reward-based preference optimization. Experiments show that ClueAnchor significantly outperforms prior RAG baselines in the completeness and robustness of reasoning. Further analysis confirms its strong resilience to noisy or partially relevant retrieved content, as well as its capability to identify supporting evidence even in the absence of explicit clue supervision during inference. All codes are available at https://github.com/thunlp/ClueAnchor.

Figures (8)

• Figure 1: Illustration of ClueAnchor. Standard RAG frequently misled by irrelevant content, while ClueAnchor identifies key clues from retrieved documents and uses them to guide faithful reasoning and answer generation.
• Figure 2: Overview of the ClueAnchor Framework. The Knowledge Reasoning Exploration (KRE) module generates diverse reasoning paths conditioned on different evidence scenarios. The Knowledge Reasoning Optimization (KRO) module selects and refines the most effective responses using reward-based preference signals.
• Figure 3: Effectiveness of Knowledge Reasoning Optimization in ClueAnchor. Results are shown on Llama-3.1-8B-Instruct.
• Figure 4: Performance of Different RAG Methods under Noisy Substitution Scenario.
• Figure 5: Performance of Different RAG Methods under Noisy Injection Scenario.