The Great Nugget Recall: Automating Fact Extraction and RAG Evaluation with Large Language Models

2025-04-21

AI Summary

To address the reliance on manual annotation and poor scalability in evaluating Retrieval-Augmented Generation (RAG) systems, this paper proposes AutoNuggetizer—a novel framework that achieves, for the first time, a fully automated, closed-loop implementation of the classic TREC nugget-based evaluation paradigm. AutoNuggetizer leverages large language models (LLMs) to precisely extract atomic factual units (“nuggets”) from reference answers and automatically assess their coverage in RAG outputs via semantic matching. A key innovation is its component-wise calibration strategy, ensuring high alignment between each module’s outputs and human annotations. Evaluated on the TREC 2024 RAG Track, AutoNuggetizer achieves strong operational-level agreement with human judgments (Spearman ρ > 0.9) for nugget-level scoring, and its standalone nugget assignment module outperforms baseline methods. This work establishes a new, efficient, reliable, and reproducible paradigm for automated RAG evaluation.

Technology Category

• Natural Language Processing: Interpretability, Analysis, and Evaluation of NLP Models
• Machine Learning: Calibration & Uncertainty Quantification
• Knowledge Representation and Reasoning: Automated Reasoning and Theorem Proving

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

Large Language Models (LLMs) have significantly enhanced the capabilities of information access systems, especially with retrieval-augmented generation (RAG). Nevertheless, the evaluation of RAG systems remains a barrier to continued progress, a challenge we tackle in this work by proposing an automatic evaluation framework that is validated against human annotations. We believe that the nugget evaluation methodology provides a solid foundation for evaluating RAG systems. This approach, originally developed for the TREC Question Answering (QA) Track in 2003, evaluates systems based on atomic facts that should be present in good answers. Our efforts focus on"refactoring"this methodology, where we describe the AutoNuggetizer framework that specifically applies LLMs to both automatically create nuggets and automatically assign nuggets to system answers. In the context of the TREC 2024 RAG Track, we calibrate a fully automatic approach against strategies where nuggets are created manually or semi-manually by human assessors and then assigned manually to system answers. Based on results from a community-wide evaluation, we observe strong agreement at the run level between scores derived from fully automatic nugget evaluation and human-based variants. The agreement is stronger when individual framework components such as nugget assignment are automated independently. This suggests that our evaluation framework provides tradeoffs between effort and quality that can be used to guide the development of future RAG systems. However, further research is necessary to refine our approach, particularly in establishing robust per-topic agreement to diagnose system failures effectively.

Problem

• Automating fact extraction for RAG evaluation using LLMs
• Validating automatic evaluation against human annotations
• Improving RAG system development with tradeoffs between effort and quality

Innovation

• Automated RAG evaluation using LLMs
• AutoNuggetizer framework for fact extraction
• Calibrated automatic vs human nugget scoring