mlsanityData to 7 Oct 2026

Grounded Revision vs. Prior Injection: Probing Retrieval-Augmented Patent Claim Amendment

arXiv 2609.36550 ↗First posted 29 Sep 2026cs.CLcited by 0

On the map: Retrieval-augmented generation → Legal and financial RAG also fits: RAG robustness and adaptive retrieval

Abstract (arXiv)

Retrieval-augmented generation is widely used in professional writing, yet whether retrieval grounds revision or merely injects templates is rarely tested where "correct" has a definable meaning. Patent claim amendment supplies that signal: the examiner names the attacked limitation and cites prior art, providing per-case ground truth. We release three artifacts: (i) a corpus of 7,385 USPTO prosecution cases with XML-aligned pre/post claims, rejection, and cited prior art; (ii) a seven-probe battery comparing random and structural-match retrieval as two policies under a fixed prompt scaffold; (iii) a deterministic five-channel metric (C1-C3 and C5 in main, C4 supplementary) requiring no LLM evaluation. Across 9,600 pre-registered calls on four frontier LLMs (Claude Sonnet 4, Claude Haiku 4.5, GPT-5.4, GPT-4o-mini), no tested model exhibits detectable classical prior-injection behavior; retrieval effects are small and direction-inconsistent between random and structural retrieval, and the null is unchanged under a dense (semantic) retriever, across retrieval depths k in {1,3,5,10}, and under a paraphrase-sensitive grounding metric. Revision locality reveals a model-specific difference that the template channel misses. The four-cell taxonomy, which we treat as exploratory, leaves the prior-injector cell unoccupied.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Legal and financial RAG, one of 15 directions in Retrieval-augmented generation. Builds on 25 of its 35 references that are in the map. First posted on arXiv 29 Sep 2026.

Builds on

25 of 35 references.

References · 25

  1. 2020Retrieval-Augmented Generation for Knowledge-Intensive NLP Taskscited by 9,160
  2. 2021Evaluating Large Language Models Trained on Codecited by 8,053
  3. 2023Lost in the Middle: How Language Models Use Long Contextscited by 3,383
  4. 2023SWE-bench: Can Language Models Resolve Real-World GitHub Issues?cited by 3,193
  5. 2020REALM: Retrieval-Augmented Language Model Pre-Trainingcited by 2,126
  6. 2023Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflectioncited by 1,800
  7. 2021BEIR: A Heterogenous Benchmark for Zero-shot Evaluation of Information Retrieval Modelscited by 1,225
  8. 2020Beyond Accuracy: Behavioral Testing of NLP models with CheckListcited by 979
  9. 2022Atlas: Few-shot Learning with Retrieval Augmented Language Modelscited by 940
  10. 2024The Power of Noise: Redefining Retrieval for RAG Systemscited by 275
  11. 2019BIGPATENT: A Large-Scale Dataset for Abstractive and Coherent Summarizationcited by 155
  12. 2026Towards End-to-End Automation of AI Researchcited by 150
13 more
  1. 2024Enhancing Noise Robustness of Retrieval-Augmented Language Models with Adaptive Adversarial Trainingcited by 88
  2. 2021Preregistering NLP Researchcited by 24
  3. 2024InstructPatentGPT: Training patent language models to follow instructions with human feedbackcited by 17
  4. 2024AutoPatent: A Multi-Agent Framework for Automatic Patent Generationcited by 16
  5. 2025PatentScore: Multi-dimensional Evaluation of LLM-Generated Patent Claimscited by 10
  6. 2026PaperOrchestra: A Multi-Agent Framework for Automated AI Research Paper Writingcited by 10
  7. 2025Towards Better Evaluation for Generated Patent Claimscited by 8
  8. 2025PANORAMA: A Dataset and Benchmarks Capturing Decision Trails and Rationales in Patent Examinationcited by 5
  9. 2024Patent-CR: A Dataset for Patent Claim Revisioncited by 4
  10. 2026PILOT-Bench: A Benchmark for Legal Reasoning in the Patent Domain with IRAC-Aligned Classification Taskscited by 4
  11. 2025PEDANTIC: A Dataset for the Automatic Examination of Definiteness in Patent Claimscited by 3
  12. 2025PATENTWRITER: A Benchmarking Study for Patent Drafting with LLMscited by 2
  13. 2025ToC: Tree-of-Claims Search with Multi-Agent Language Modelscited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Stale-Document Poisoning: When Outdated Retrieval Overrides Correct Model Answers
  2. 2026Beyond Probabilistic Similarity: Structural, Temporal, and Causal Limitations of Retrieval-Augmented Generation in the Legal Domain
  3. 2026How Robust Is Multimodal Claim Verification to LLM Rewriting?
  4. 2026Deceptive Grounding: Entity Attribution Failure in Clinical Retrieval-Augmented Generation
  5. 2026Diagnosing Evidence Utilization in Long-Context and Retrieval-Augmented Language Models under Matched Evidence Conditions
  6. 2024ClashEval: Quantifying the tug-of-war between an LLM's internal prior and external evidence
  7. 2026CUE-R: Beyond the Final Answer in Retrieval-Augmented Generation
  8. 2026Evaluating RAG Reliability under Clean, Misleading, and Mixed Retrieval