Beyond Relevance: On the Relationship Between Retrieval and RAG Information Coverage
On the map: Retrieval-augmented generation → RAG pipeline design and evaluation
Abstract (arXiv)
Retrieval-augmented generation (RAG) systems combine document retrieval with a generative model to address complex information seeking tasks like report generation. While the relationship between retrieval quality and generation effectiveness seems intuitive, it has not been systematically studied. We investigate whether upstream retrieval metrics can serve as reliable early indicators of the final generated response's information coverage. Through experiments across two text RAG benchmarks (TREC NeuCLIR 2024 and TREC RAG 2024) and one multimodal benchmark (WikiVideo), we analyze 15 text retrieval stacks and 10 multimodal retrieval stacks across four RAG pipelines and multiple evaluation frameworks (Auto-ARGUE and MiRAGE). Our findings demonstrate strong correlations between coverage-based retrieval metrics and nugget coverage in generated responses at both topic and system levels. This relationship holds most strongly when retrieval objectives align with generation goals, though more complex iterative RAG pipelines can partially decouple generation quality from retrieval effectiveness. These findings provide empirical support for using retrieval metrics as proxies for RAG performance.
Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Placed in RAG pipeline design and evaluation, one of 15 directions in Retrieval-augmented generation. Builds on 39 of its 69 references that are in the map. First posted on arXiv 9 Mar 2026.
Builds on
39 of 69 references.
Closest ancestors · 7
Same direction, newest first
- 2026RoutIR: Fast Serving of Retrieval Pipelines for Retrieval-Augmented Generation15 Jan
- 2025Auto-ARGUE: LLM-Based Report Generation Evaluation30 Sep
- 2025IterKey: Iterative Keyword Generation with LLMs for Enhanced Retrieval Augmented Generation13 May
- 2025Support Evaluation for the TREC 2024 RAG Track: Comparing Human versus LLM Judges21 Apr
- 2025GINGER: Grounded Information Nugget-Based Generation of Responses23 Mar
- 2024Initial Nugget Evaluation Results for the TREC 2024 RAG Track with the AutoNuggetizer Framework14 Nov
- 2024Ragnar"ok: A Reusable RAG Framework and Baselines for TREC 2024 Retrieval-Augmented Generation Track24 Jun
Other references · 32
- 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,456
- 2023Lost in the Middle: How Language Models Use Long Contextscited by 3,383
- 2023Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflectioncited by 1,800
- 2025Qwen3 Embedding: Advancing Text Embedding and Reranking Through Foundation Modelscited by 1,175
- 2023LanguageBind: Extending Video-Language Pretraining to N-modality by Language-based Semantic Alignmentcited by 366
- 2025Why Language Models Hallucinatecited by 205
- 2024xRAG: Extreme Context Compression for Retrieval-augmented Generation with One Tokencited by 123
- 2025Context Length Alone Hurts LLM Performance Despite Perfect Retrievalcited by 118
- 2024Non-Determinism of "Deterministic" LLM Settingscited by 84
- 2024SPLADE-v3: New baselines for SPLADEcited by 69
- 2025Rank1: Test-Time Compute for Reranking in Information Retrievalcited by 68
- 2024IM-RAG: Multi-Round Retrieval-Augmented Generation Through Learning Inner Monologuescited by 40
20 more
- 2025Tevatron 2.0: Unified Document Retrieval Toolkit across Scale, Language, and Modalitycited by 30
- 2025Knowing You Don't Know: Learning When to Continue Search in Multi-round RAG through Self-Practicingcited by 24
- 2024On the Evaluation of Machine-Generated Reportscited by 23
- 2024Translate-Distill: Learning Cross-Language Dense Retrieval by Translation and Distillationcited by 22
- 2025Video-ColBERT: Contextualized Late Interaction for Text-to-Video Retrievalcited by 20
- 2025LLMs Provide Unstable Answers to Legal Questionscited by 19
- 2024MultiVENT 2.0: A Massive Multilingual Benchmark for Event-Centric Video Retrievalcited by 17
- 2025Beyond Factual Accuracy: Evaluating Coverage of Diverse Factual Information in Long-form Text Generationcited by 15
- 2025WikiVideo: Article Generation from Multiple Videoscited by 13
- 2025KiRAG: Knowledge-Driven Iterative Retriever for Enhancing Retrieval-Augmented Generationcited by 12
- 2021Streamlining Evaluation with ir-measurescited by 10
- 2025Milco: Learned Sparse Retrieval Across Languages via a Multilingual Connectorcited by 9
- 2020QBSUM: a Large-Scale Query-Based Document Summarization Dataset from Real-world Applicationscited by 8
- 2025Conversational Gold: Evaluating Personalized Conversational Search System using Gold Nuggetscited by 8
- 2025MMMORRF: Multimodal Multilingual Modularized Reciprocal Rank Fusioncited by 8
- 2017Benchmark for Complex Answer Retrievalcited by 6
- 2025HLTCOE at LiveRAG: GPT-Researcher using ColBERT retrievalcited by 6
- 2025HLTCOE at TREC 2024 NeuCLIR Trackcited by 5
- 2024An Exam-based Evaluation Approach Beyond Traditional Relevance Judgmentscited by 4
- 2026Incorporating Q&A Nuggets into Retrieval-Augmented Generationcited by 2
Built on by
Citing papers per year: 2026: 1.
1 paper.
Similar, not cited
Close in wording, no citation either way
- 2025RAGTrace: Understanding and Refining Retrieval-Generation Dynamics in Retrieval-Augmented Generation
- 2025Enhancing Retrieval-Augmented Generation: A Study of Best Practices
- 2025RAGSmith: A Framework for Finding the Optimal Composition of Retrieval-Augmented Generation Methods Across Datasets
- 2026Beyond Relevance: Utility-Centric Retrieval in the LLM Era
- 2024Evaluation of Retrieval-Augmented Generation: A Survey
- 2025Retrieval Augmented Generation Evaluation in the Era of Large Language Models: A Comprehensive Survey
- 2024RAGChecker: A Fine-grained Framework for Diagnosing Retrieval-Augmented Generation
- 2025Controlled Retrieval-augmented Context Evaluation for Long-form RAG