mlsanityData to 7 Oct 2026

How Do Document Parsers Break? Auditing Structural Vulnerability in Document Intelligence

arXiv 2605.19309 ↗First posted 19 May 2026cs.CLcited by 0

On the map: Retrieval-augmented generation → Multimodal RAG

Abstract (arXiv)

Document Layout Analysis (DLA) pipelines provide structured page representations for retrieval-augmented generation, long-document question answering, and related applications. Yet their robustness evaluation remains largely area-centric. We identify this Footprint Bias and propose ProSA, a lightweight output-level auditing framework that decouples controlled probing, policy-driven targeting, and structure-aware diagnosis. ProSA combines Block-level Structural Loss Rate (B-SLR), granularity-aware exposure descriptors, and pathway attribution to analyze where structural identity is lost, at what exposure granularity failures emerge, and how failures propagate. Across MinerU and PP-StructureV3 on 1,000 pages, affected area weakly tracks perturbation-induced OCR instability (R2=0.384/0.110), whereas B-SLR aligns much more closely with it (R2=0.727/0.916). Exposure descriptors further separate occlusion- and topology-dominant pathways, while matched-footprint structural probes cause much larger downstream QA/retrieval drops than area-matched erasure. These results shift DLA robustness evaluation from footprint-based measurement toward structure-aware vulnerability auditing.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Multimodal RAG, one of 15 directions in Retrieval-augmented generation. Builds on 22 of its 29 references that are in the map. First posted on arXiv 19 May 2026.

Builds on

22 of 29 references.

Closest ancestors · 1

Same direction, newest first

  1. 2025SCAN: Semantic Document Layout Analysis for Textual and Visual Retrieval-Augmented Generation20 May

Other references · 21

  1. 2017Improved Regularization of Convolutional Neural Networks with Cutoutcited by 1,623
  2. 2018ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustnesscited by 1,430
  3. 2018On the Spectral Bias of Neural Networkscited by 1,321
  4. 2017Random Erasing Data Augmentationcited by 1,235
  5. 2017Adversarial Patchcited by 514
  6. 2019PubLayNet: largest dataset ever for document layout analysiscited by 349
  7. 2019Benchmarking Robustness in Object Detection: Autonomous Driving when Winter is Comingcited by 332
  8. 2019High Frequency Component Helps Explain the Generalization of Convolutional Neural Networkscited by 310
  9. 2019A Fourier Perspective on Model Robustness in Computer Visioncited by 288
  10. 2024MinerU: An Open-Source Solution for Precise Document Content Extractioncited by 255
  11. 2022DocLayNet: A Large Human-Annotated Dataset for Document-Layout Analysiscited by 146
  12. 2024DocLayout-YOLO: Enhancing Document Layout Analysis through Diverse Synthetic Data and Global-to-Local Adaptive Perceptioncited by 80
9 more
  1. 2023DocMath-Eval: Evaluating Math Reasoning Capabilities of LLMs in Understanding Long and Specialized Documentscited by 45
  2. 2018On the Structural Sensitivity of Deep Convolutional Networks to the Directions of Fourier Basis Functionscited by 31
  3. 2024RoDLA: Benchmarking the Robustness of Document Layout Analysis Modelscited by 12
  4. 2026Real5-OmniDocBench: A Full-Scale Physical Reconstruction Benchmark for Robust Document Parsing in the Wildcited by 11
  5. 2020CLEval: Character-Level Evaluation for Text Detection and Recognition Taskscited by 10
  6. 2025DocPTBench: Benchmarking End-to-End Photographed Document Parsing and Translationcited by 8
  7. 2019TedEval: A Fair Evaluation Metric for Scene Text Detectorscited by 3
  8. 2024Feature Averaging: An Implicit Bias of Gradient Descent Leading to Non-Robustness in Neural Networkscited by 3
  9. 2025Benchmarking Adversarial Patch Selection and Locationcited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026RealDocBench: A Benchmark for Field-Level QA and Layout Understanding on Real-World Regulated Documents
  2. 2025LED Benchmark: Diagnosing Structural Layout Errors for Document Layout Analysis
  3. 2025Equipping Retrieval-Augmented Large Language Models with Document Structure Awareness
  4. 2026Does Document Structure Help Dense Retrieval? A Placebo-Controlled Ablation of Four Mechanisms Across Two Corpora
  5. 2026Locating Failure in Multi-Page Visually Rich Document Understanding: An Empirical Attribution
  6. 2026LED: A Benchmark for Evaluating Layout Error Detection in Document Analysis
  7. 2026Structured Layout Priors for Robust Out-of-Distribution Visual Document Understanding
  8. 2026SPIRE: Structure-Preserving Interpretable Retrieval of Evidence