Astute RAG: Overcoming Imperfect Retrieval Augmentation and Knowledge Conflicts for Large Language Models

2024-10-09

arXiv.org

AI Summary

This work addresses two core challenges in retrieval-augmented generation (RAG): (1) retrieved content often contains irrelevant, misleading, or adversarial information; and (2) conflicts frequently arise between the large language model’s (LLM’s) internal knowledge and externally retrieved knowledge. To tackle these issues, we propose a robust RAG framework featuring a novel source-aware iterative fusion mechanism that jointly integrates internal and external knowledge. Specifically, the framework explicitly elicits the LLM’s intrinsic knowledge, dynamically weights internal and retrieved evidence based on source credibility, and guides reliability-aware answer generation. It is compatible with closed-source models including Gemini and Claude. Extensive experiments demonstrate that, even under worst-case retrieval conditions, our method matches or surpasses pure-LLM baselines—and significantly outperforms existing robust RAG approaches—in both answer accuracy and consistency, thereby enhancing output trustworthiness and stability.

Technology Category

• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Natural Language Processing: Safety and Robustness
• Machine Learning: Adversarial Learning & Robustness

Application Category

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

Abstract

Retrieval augmented generation (RAG), while effectively integrating external knowledge to address the inherent limitations of large language models (LLMs), can be hindered by imperfect retrieval that contain irrelevant, misleading, or even malicious information. Previous studies have rarely connected the behavior of RAG through joint analysis, particularly regarding error propagation coming from imperfect retrieval and potential conflicts between LLMs' internal knowledge and external sources. Through comprehensive and controlled analyses under realistic conditions, we find that imperfect retrieval augmentation is inevitable, common, and harmful. We identify the knowledge conflicts between LLM-internal and external knowledge from retrieval as a bottleneck to overcome imperfect retrieval in the post-retrieval stage of RAG. To address this, we propose Astute RAG, a novel RAG approach designed to be resilient to imperfect retrieval augmentation. It adaptively elicits essential information from LLMs' internal knowledge, iteratively consolidates internal and external knowledge with source-awareness, and finalizes the answer according to information reliability. Our experiments with Gemini and Claude demonstrate the superior performance of Astute RAG compared to previous robustness-enhanced RAG approaches. Specifically, Astute RAG is the only RAG method that achieves performance comparable to or even surpassing conventional use of LLMs under the worst-case scenario. Further analysis reveals the effectiveness of Astute RAG in resolving knowledge conflicts, thereby improving the trustworthiness of RAG.

Problem

• Overcoming imperfect retrieval in RAG for LLMs
• Resolving knowledge conflicts between internal and external sources
• Enhancing RAG robustness against misleading information

Innovation

• Adaptively elicits internal knowledge from LLMs
• Iteratively consolidates internal and external knowledge
• Finalizes answer based on information reliability