A Systems-Level Analysis of Sensitivity, Robustness, and Stability in Retrieval-Augmented Generation

• URL: http://arxiv.org/abs/2606.28337v1
• Date: Fri, 29 May 2026 17:24:40 GMT
• Title: A Systems-Level Analysis of Sensitivity, Robustness, and Stability in Retrieval-Augmented Generation
• Abstract summary: Retrieval-Augmented Generation (RAG) systems are often evaluated using final answer accuracy.<n>This paper presents a controlled empirical study of RAG sensitivity, robustness, and stability across 56 experimental runs.
• Score: 0.0
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Retrieval-Augmented Generation (RAG) systems are often evaluated using final answer accuracy, even though their failures can originate from preprocessing, retrieval, context packing, or generation. This paper presents a controlled empirical study of RAG sensitivity, robustness, and stability across 56 experimental runs. We evaluate how chunk size, retrieval depth (top k), embedding-based reranking, probabilistic retrieval noise, and repeated seeded runs affect retrieval, context packing, and generation behavior. Using a fixed 500-question QA subset mapped to 20,958 unique corpus contexts, we analyze both final answer metrics and intermediate failure modes. Across these experiments, retrieval-oriented metrics improved under broader retrieval settings, while downstream exact-match and F1 scores often behaved non-monotonically. We also observe preprocessing-induced answer loss under smaller chunk sizes, progressive degradation under retrieval corruption, and higher observed variance in broader retrieval regimes. These findings suggest that RAG evaluation should include sensitivity, robustness, stability, and multi-stage failure analysis rather than relying only on final answer accuracy.

Related papers

• Testing Retrieval-Augmented Generation Systems with Chunk Coverage [7.72636722205447]
  RAG-based systems are increasingly deployed in high-stakes settings where correct behaviour depends on the retrieval component.<n>Chunk Coverage (CC) is an oracle-independent test adequacy criterion for testing the retrieval component of RAG systems.<n>We show how CC can be used to guide test selection and generation by prioritising queries that expand coverage of previously unexercised retrieval regions.
  arXiv  Detail & Related papers  (2026-07-20T16:53:05Z)
• When Poison Fails After Retrieval: Revisiting Corpus Poisoning under Chunking and Reranking Pipelines [13.701523553327116]
  Retrieval-Augmented Generation (RAG) systems are vulnerable to corpus poisoning attacks that manipulate downstream model outputs through malicious knowledge injection.<n>Existing studies mainly evaluate poisoning under simplified retrieval settings, overlooking practical RAG pipelines involving document chunking, dense retrieval, reranking, and grounded generation.<n>We propose Chunk-aware and Rerank-Consistent Poisoning (CRCP), a poisoning framework that jointly optimize retrieval relevance, reranker consistency, and chunk-boundary robustness.
  arXiv  Detail & Related papers  (2026-06-09T04:45:28Z)
• FRESCO: Benchmarking and Optimizing Re-rankers for Evolving Semantic Conflict in Retrieval-Augmented Generation [73.22935457705057]
  We introduce FRESCO, a benchmark for evaluating re-rankers in temporally dynamic contexts.<n>By pairing recency-seeking queries with historical Wikipedia revisions, FRESCO tests whether re-rankers can prioritize factually recent evidence while maintaining semantic relevance.<n>Our evaluation reveals a consistent failure mode across existing re-rankers: a strong bias toward older, semantically rich documents, even when they are factually obsolete.
  arXiv  Detail & Related papers  (2026-04-14T17:04:25Z)
• Variational Rectification Inference for Learning with Noisy Labels [74.85528327499662]
  We propose variational rectification inference (VRI) to formulate the adaptive rectification for loss functions.<n>VRI is constructed as a hierarchical Bayes by treating the rectifying vector as a latent variable.<n>By introducing the variational term in VRI, the conditional posterior is estimated accurately and avoids collapsing to a Dirac delta function.
  arXiv  Detail & Related papers  (2026-03-18T01:25:08Z)
• Beyond Relevance: On the Relationship Between Retrieval and RAG Information Coverage [89.58253972744531]
  Retrieval-augmented generation (RAG) systems combine document retrieval with a generative model to address complex information seeking tasks like report generation.<n>We investigate whether upstream retrieval metrics can serve as reliable early indicators of the final generated response's information coverage.<n>Our findings demonstrate strong correlations between coverage-based retrieval metrics and nugget coverage in generated responses at both topic and system levels.
  arXiv  Detail & Related papers  (2026-03-09T18:20:20Z)
• When Iterative RAG Beats Ideal Evidence: A Diagnostic Study in Scientific Multi-hop Question Answering [0.2796197251957245]
  We study whether synchronized iterative retrieval and reasoning can surpass an idealized static upper bound (Gold Context) RAG.<n>We benchmark eleven state-of-the-art LLMs under three regimes: (i) No Context, measuring reliance on parametric memory; (ii) Gold Context, where all evidence is supplied at once; and (iii) Iterative RAG, a training-free oracle controller that alternates retrieval, hypothesis refinement, and evidence-aware stopping.<n>Across models, Iterative RAG consistently outperforms Gold Context, with gains up to 25.6 percentage points, especially for non-reasoning
  arXiv  Detail & Related papers  (2026-01-27T17:35:05Z)
• Investigating the Robustness of Retrieval-Augmented Generation at the Query Level [4.3028340012580975]
  Retrieval-augmented generation (RAG) has been proposed as a solution that dynamically incorporates external knowledge during inference.<n>Despite its promise, RAG systems face practical challenges-most notably, a strong dependence on the quality of the input query for accurate retrieval.
  arXiv  Detail & Related papers  (2025-07-09T15:39:17Z)
• RARE: Retrieval-Aware Robustness Evaluation for Retrieval-Augmented Generation Systems [33.389969814185214]
  Retrieval-Augmented Generation (RAG) enhances recency and factuality in answers.<n>Existing evaluations rarely test how well RAG systems cope with real-world noise, conflicting between internal and external retrieved contexts, or fast-changing facts.<n>We introduce Retrieval-Aware Robustness Evaluation (RARE), a unified framework and large-scale benchmark that jointly stress-test query and document perturbations over dynamic, time-sensitive corpora.
  arXiv  Detail & Related papers  (2025-06-01T02:42:36Z)
• Chain-of-Retrieval Augmented Generation [91.02950964802454]
  This paper introduces an approach for training o1-like RAG models that retrieve and reason over relevant information step by step before generating the final answer.<n>Our proposed method, CoRAG, allows the model to dynamically reformulate the query based on the evolving state.
  arXiv  Detail & Related papers  (2025-01-24T09:12:52Z)
• Controlling Risk of Retrieval-augmented Generation: A Counterfactual Prompting Framework [77.45983464131977]
  We focus on how likely it is that a RAG model's prediction is incorrect, resulting in uncontrollable risks in real-world applications.<n>Our research identifies two critical latent factors affecting RAG's confidence in its predictions.<n>We develop a counterfactual prompting framework that induces the models to alter these factors and analyzes the effect on their answers.
  arXiv  Detail & Related papers  (2024-09-24T14:52:14Z)
• RAGGED: Towards Informed Design of Scalable and Stable RAG Systems [51.171355532527365]
  Retrieval-augmented generation (RAG) enhances language models by integrating external knowledge.<n>RAGGED is a framework for systematically evaluating RAG systems.
  arXiv  Detail & Related papers  (2024-03-14T02:26:31Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.