---
title: "Post Hoc Agentic Refinement for Improving Precision in"
description: "The authors build a GPT-5 agent that works through MCP tools: a persistent to-do list, regex validators for dates, IDs, URLs… 27 references."
authors: ["Justin Xu", "Alistair Johnson", "Thomas Lin", "David Eyre", "Rodolfo Quispe"]
year: 2026
venue: "BioNLP 2026"
source: "https://aclanthology.org/2026.bionlp-1.11/"
doi: "10.18653/v1/2026.bionlp-1.11"
licence: "CC BY 4.0 (ACL Anthology)"
licence_url: "https://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/xu-2026-post-hoc-agentic-refinement-improving/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/xu-2026-post-hoc-agentic-refinement-improving/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2026](https://md.research.elamur.ai/free/en/papers/2026/page/45/) / Post Hoc Agentic Refinement for Improving Precision in Multilingual Clinical Text De-identification

# Post Hoc Agentic Refinement for Improving Precision in Multilingual Clinical Text De-identification

**Authors:** [Justin Xu](https://md.research.elamur.ai/free/en/authors/justin-xu/), [Alistair Johnson](https://md.research.elamur.ai/free/en/authors/alistair-e-w-johnson/), Thomas Lin, David Eyre and Rodolfo Quispe · **Year:** 2026 · **Venue:** [BioNLP 2026](https://md.research.elamur.ai/free/en/venues/bionlp/) ([BioNLP 2026](https://md.research.elamur.ai/free/en/venues/bionlp/2026/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2026.bionlp-1.11/) · [PDF](https://aclanthology.org/2026.bionlp-1.11.pdf) · [DOI 10.18653/v1/2026.bionlp-1.11](https://doi.org/10.18653/v1/2026.bionlp-1.11) · [OpenAlex](https://openalex.org/W7166872551)

**Citations:** 0 citations (OpenAlex, 2026-10-14)

## Summary

Clinical de-identification tools are usually tuned to catch every possible identifier. The result is heavy over-redaction, which hurts readability and the downstream value of the data. Xu et al. (Oxford and Microsoft, BioNLP 2026) frame the fix as post hoc refinement. Given a high-recall set of PHI/PII spans, a reviewer decides for each span whether to keep it, adjust its boundaries or class, or drop it. It never adds new spans.

The authors build a GPT-5 agent that works through MCP tools:
- a persistent to-do list,
- regex validators for dates, IDs, URLs and emails,
- context retrieval with an adjustable window,
- class-specific guideline "skills" loaded on demand.

They compare it with a single-call GPT-5 judge on three datasets: Spanish clinical notes, simulated German doctor–patient conversations, and English Wikipedia info-boxes. The agent raises precision considerably while losing less recall than the baseline on the medical data, and it improves F1 on all three datasets. It also repairs over 99% of synthetically injected errors. Because it works on top of existing pipelines without retraining, it offers a practical way to control precision.

## Key points

- **Datasets and annotations:** The three evaluation sets are CARMEN-ES (5,614 PHI, from the Spanish part of CARMEN-I), Transcripts-DE (238 PHI) and WikiPII-EN (24,145 PII). In every case, the initial annotations come from few-shot GPT-5 (gpt-5-2025-08-07).
- **Agentic F1 gains:**
  - CARMEN-ES: from 93.65 to 96.65, with precision 97.95 (+7.19) and recall 95.39 (−1.36).
  - Transcripts-DE: from 69.67 to 85.49, with precision +27.67.
  - WikiPII-EN: from 89.41 to 91.75.
- **Baseline comparison:** The single-call GPT-5 refiner reaches lower F1 on all three datasets (94.14 / 75.97 / 91.37). On the medical datasets it also loses more recall than the agent.
- **Synthetic error datasets:** The agent catches 99.01% (CARMEN-ES) and 99.32% (Transcripts-DE) of injected errors, versus 96.69% and 84.26% for the baseline.
- **Component effects and limitations:**
  - The review skills drove most of the precision improvement, and context retrieval mainly reduces token usage.
  - Recall cannot exceed that of the upstream system.
  - The main experiments use only GPT-5; there are preliminary positive results with gpt-oss-120b.
  - No formal ablation study has been done yet.

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

## Institutions

- [Microsoft](https://md.research.elamur.ai/free/en/companies/microsoft/) — Company, Redmond, United States · 378 works · 161,630 citations
- [University of Oxford](https://md.research.elamur.ai/free/en/universities/university-of-oxford/) — University, Oxford, United Kingdom · 122 works · 160,277 citations

📚 Shelves (6): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 18,614 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 17,965 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 17,450 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 17,296 works · [🧬 Biochemistry, Genetics and Molecular Biology](https://md.research.elamur.ai/free/en/shelves/biochemistry-genetics-and-molecular-biology/) · 263 works · [🧬 Molecular Biology](https://md.research.elamur.ai/free/en/shelves/biochemistry-genetics-and-molecular-biology/molecular-biology/) · 260 works

🧭 Topics (1): [🌍 Multilingual & cross-lingual NLP](https://md.research.elamur.ai/free/en/topics/multilingual-nlp/) · 882 works

## Cite

Cite as: Justin Xu, Alistair Johnson, Thomas Lin, David Eyre and Rodolfo Quispe. 2026. Post Hoc Agentic Refinement for Improving Precision in Multilingual Clinical Text De-identification. BioNLP 2026.

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{xu2026post,
  title = {{Post Hoc Agentic Refinement for Improving Precision in Multilingual Clinical Text De-identification}},
  author = {Justin Xu, Alistair Johnson, Thomas Lin, David Eyre and Rodolfo Quispe},
  booktitle = {BioNLP 2026},
  year = {2026},
  doi = {10.18653/v1/2026.bionlp-1.11},
  url = {https://doi.org/10.18653/v1/2026.bionlp-1.11}
}
```

- [← previous Position: What Are We Measuring? Rethinking Evaluation in Natural Language Generation](https://md.research.elamur.ai/free/en/papers/zaghouani-2026-position-what-are-we-measuring/)
- [next → pQuant: Towards Effective Low-Bit Language Models via Decoupled Linear Quantization-Aware Training](https://md.research.elamur.ai/free/en/papers/zhang-2026-pquant-effective-low-bit-language/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/45/)

## References

27 references, 7 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Kiana Aghakasiri, Noopur Zambare, JoAnn Thai, Carrie Ye, Mayur Mehta, J. Ross Mitchell, and Mohamed Abdalla. 2025. Not what the doctor ordered: Surveying LLM-based de-identification and quantifying clinical information loss. Preprint, arxiv:2509.14464 [cs]. [find](https://scholar.google.com/scholar?q=Kiana+Aghakasiri%2C+Noopur+Zambare%2C+JoAnn+Thai%2C+Carrie+Ye%2C+Mayur+Mehta%2C+J%2E+Ross+Mitchell%2C+and+Mohamed+Abdalla%2E+2025%2E+Not+what+the+doctor+ordered%3A+Surveying+LLM%2Dbased+de%2Didentification+and+quantifying+cl)
2. Emily Alsentzer, John R. Murphy, Willie Boag, Wei-Hung Weng, Di Jin, Tristan Naumann, and Matthew B. A. McDermott. 2019. [Publicly available clinical BERT embeddings](https://md.research.elamur.ai/free/en/papers/alsentzer-2019-publicly-available-clinical-bert-embeddings/). Preprint, arxiv:1904.03323 [cs].
3. Hong-Jie Dai, Tatheer Hussain Mir, Ching-Tai Chen, Chien-Chang Chen, Hao-Ping Yang, Chung-Hong Lee, Yi-Yun Chou, Yu-Chin Teng, Shalini Gupta, Omkar Panchal, Divyabharathy Ramesh Nadar, Wei-Hsiang Liao, Yu-Chuan Lin, Zi-Rui Zhao, Richard Tzong-Han Tsai, Yung-Chun Chang, and Jitendra Jonnagaddala. 202… — [arXiv:2504.03964](https://arxiv.org/abs/2504.03964) [find](https://scholar.google.com/scholar?q=Hong%2DJie+Dai%2C+Tatheer+Hussain+Mir%2C+Ching%2DTai+Chen%2C+Chien%2DChang+Chen%2C+Hao%2DPing+Yang%2C+Chung%2DHong+Lee%2C+Yi%2DYun+Chou%2C+Yu%2DChin+Teng%2C+Shalini+Gupta%2C+Omkar+Panchal%2C+Divyabharathy+Ramesh+Nadar%2C+Wei%2DHsiang+Liao)
4. Chong Li, Shaonan Wang, Jiajun Zhang, and Chengqing Zong. 2024. Improving in-context learning of multilingual generative language models with cross-lingual alignment. Preprint, arxiv:2311.08089 [cs]. [find](https://scholar.google.com/scholar?q=Chong+Li%2C+Shaonan+Wang%2C+Jiajun+Zhang%2C+and+Chengqing+Zong%2E+2024%2E+Improving+in%2Dcontext+learning+of+multilingual+generative+language+models+with+cross%2Dlingual+alignment%2E+Preprint%2C+arxiv%3A2311%2E08089+%5Bcs%5D%2E)
5. Xinkui Lin, Yuhui Zhang, Yongxiu Xu, Kun Huang, Hongzhang Mu, Yubin Wang, Gaopeng Gou, Li Qian, Li Peng, Wei Liu, Jian Luan, and Hongbo Xu. 2025. MAKAR: a multi-agent framework based knowledge-augmented reasoning for grounded multimodal named entity recognition. In Proceedings of the 2025 Conference… [find](https://scholar.google.com/scholar?q=Xinkui+Lin%2C+Yuhui+Zhang%2C+Yongxiu+Xu%2C+Kun+Huang%2C+Hongzhang+Mu%2C+Yubin+Wang%2C+Gaopeng+Gou%2C+Li+Qian%2C+Li+Peng%2C+Wei+Liu%2C+Jian+Luan%2C+and+Hongbo+Xu%2E+2025%2E+MAKAR%3A+a+multi%2Dagent+framework+based+knowledge%2Daugment)
6. Meng Lu, Yuzhang Xie, Zhenyu Bi, Shuxiang Cao, and Xuan Wang. 2025. CROSSAGENTIE: Cross-type and cross-task multi-agent LLM collaboration for zero-shot information extraction. In Findings of the Association for Computational Linguistics: ACL 2025, pages 13953–13977. Association for Computational Lin… [find](https://scholar.google.com/scholar?q=Meng+Lu%2C+Yuzhang+Xie%2C+Zhenyu+Bi%2C+Shuxiang+Cao%2C+and+Xuan+Wang%2E+2025%2E+CROSSAGENTIE%3A+Cross%2Dtype+and+cross%2Dtask+multi%2Dagent+LLM+collaboration+for+zero%2Dshot+information+extraction%2E+In+Findings+of+the+Assoc)
7. Zhuojun Ding, Wei Wei, and Chenghao Fan. 2025. Selecting and merging: Towards adaptable and scalable named entity recognition with large language models. Preprint, arxiv:2506.22813 [cs]. [find](https://scholar.google.com/scholar?q=Zhuojun+Ding%2C+Wei+Wei%2C+and+Chenghao+Fan%2E+2025%2E+Selecting+and+merging%3A+Towards+adaptable+and+scalable+named+entity+recognition+with+large+language+models%2E+Preprint%2C+arxiv%3A2506%2E22813+%5Bcs%5D%2E)
8. Leland McInnes, John Healy, and James Melville. 2020. [UMAP: Uniform manifold approximation and projection for dimension reduction](https://md.research.elamur.ai/free/en/papers/mcinnes-2018-umap-uniform-manifold-approximation-projection/). Preprint, arxiv:1802.03426 [stat].
9. Eulalia Farre Maduell, Salvador Lima-Lopez, Santiago Andres Frid, Artur Conesa, Elisa Asensio, Antonio Lopez-Rueda, Helena Arino, Elena Calvo, Maria Jesús Bertran, Maria Angeles Marcos, Montserrat Nofre Maiz, Laura Tañá Velasco, Antonia Marti, Ricardo Farreres, Xavier Pastor, Xavier Bor-rat Frigola,… [find](https://scholar.google.com/scholar?q=Eulalia+Farre+Maduell%2C+Salvador+Lima%2DLopez%2C+Santiago+Andres+Frid%2C+Artur+Conesa%2C+Elisa+Asensio%2C+Antonio+Lopez%2DRueda%2C+Helena+Arino%2C+Elena+Calvo%2C+Maria+Jes%C3%BAs+Bertran%2C+Maria+Angeles+Marcos%2C+Montserrat+Nof)
10. Ishna Neamatullah, Margaret M. Douglass, Li-wei H. Lehman, Andrew Reisner, Mauricio Villarroel, William J. Long, Peter Szolovits, George B. Moody, Roger G. Mark, and Gari D. Clifford. 2008. Automated de-identification of free-text medical records. BMC medical informatics and decision making, 8:32. [find](https://scholar.google.com/scholar?q=Ishna+Neamatullah%2C+Margaret+M%2E+Douglass%2C+Li%2Dwei+H%2E+Lehman%2C+Andrew+Reisner%2C+Mauricio+Villarroel%2C+William+J%2E+Long%2C+Peter+Szolovits%2C+George+B%2E+Moody%2C+Roger+G%2E+Mark%2C+and+Gari+D%2E+Clifford%2E+2008%2E+Automated+)
11. Luyu Gao, Aman Madaan, Shuyan Zhou, Uri Alon, Pengfei Liu, Yiming Yang, Jamie Callan, and Graham Neubig. 2023. PAL: Program-aided language models. Preprint, arxiv:2211.10435 [cs]. [find](https://scholar.google.com/scholar?q=Luyu+Gao%2C+Aman+Madaan%2C+Shuyan+Zhou%2C+Uri+Alon%2C+Pengfei+Liu%2C+Yiming+Yang%2C+Jamie+Callan%2C+and+Graham+Neubig%2E+2023%2E+PAL%3A+Program%2Daided+language+models%2E+Preprint%2C+arxiv%3A2211%2E10435+%5Bcs%5D%2E)
12. Timo Schick, Jane Dwivedi-Yu, Roberto Dessì, Roberta Raileanu, Maria Lomeli, Luke Zettlemoyer, Nicola Cancedda, and Thomas Scialom. 2023. [Toolformer: Language models can teach themselves to use tools](https://md.research.elamur.ai/free/en/papers/schick-2023-toolformer-language-models-can-teach/). Preprint, arxiv:2302.04761 [cs].
13. Xinyi Hou, Yanjie Zhao, Shenao Wang, and Haoyu Wang. 2025. Model context protocol (MCP): Landscape, security threats, and future research directions. Preprint, arxiv:2503.23278 [cs]. — [arXiv:2503.23278](https://arxiv.org/abs/2503.23278) [find](https://scholar.google.com/scholar?q=Xinyi+Hou%2C+Yanjie+Zhao%2C+Shenao+Wang%2C+and+Haoyu+Wang%2E+2025%2E+Model+context+protocol+%28MCP%29%3A+Landscape%2C+security+threats%2C+and+future+research+directions%2E+Preprint%2C+arxiv%3A2503%2E23278+%5Bcs%5D%2E+%E2%80%94+%5BarXiv%3A2503%2E232)
14. Kexin Huang, Jaan Altosaar, and Rajesh Ranganath. 2020. [ClinicalBERT: Modeling clinical notes and predicting hospital readmission](https://md.research.elamur.ai/free/en/papers/huang-2019-clinicalbert-modeling-clinical-notes-predicting/). Preprint, arxiv:1904.05342 [cs].
15. Alistair E. W. Johnson, Lucas Bulgarelli, and Tom J. Pollard. 2020. Deidentification of free-text medical records using pre-trained bidirectional transformers. In Proceedings of the ACM Conference on Health, Inference, and Learning, CHIL ’20, pages 214–221. Association for Computing Machinery. [find](https://scholar.google.com/scholar?q=Alistair+E%2E+W%2E+Johnson%2C+Lucas+Bulgarelli%2C+and+Tom+J%2E+Pollard%2E+2020%2E+Deidentification+of+free%2Dtext+medical+records+using+pre%2Dtrained+bidirectional+transformers%2E+In+Proceedings+of+the+ACM+Conference+on+)
16. Simon A. Lee, Anthony Wu, and Jeffrey N. Chiang. 2025. Clinical ModernBERT: An efficient and Nils Reimers and Iryna Gurevych. 2019. [Sentence-BERT: Sentence embeddings using siamese BERT-networks](https://md.research.elamur.ai/free/en/papers/reimers-2019-sentence-bert-sentence-embeddings-siamese/). Preprint, arxiv:1908.10084 [cs].
17. Yongliang Shen, Kaitao Song, Xu Tan, Dongsheng Li, Weiming Lu, and Yueting Zhuang. 2023. [Hugging-GPT: Solving AI tasks with ChatGPT and its friends in hugging face](https://md.research.elamur.ai/free/en/papers/shen-2023-hugginggpt-solving-ai-tasks-chatgpt/). Preprint, arxiv:2303.17580 [cs].
18. Praphul Singh, Charlotte Dzialo, Jangwon Kim, Sumana Srivatsa, Irfan Bulu, Sri Gadde, and Krishnaram Kenthapadi. 2025. RedactOR: An LLM-powered framework for automatic clinical data de-identification. Preprint, arxiv:2505.18380 [cs]. [find](https://scholar.google.com/scholar?q=Praphul+Singh%2C+Charlotte+Dzialo%2C+Jangwon+Kim%2C+Sumana+Srivatsa%2C+Irfan+Bulu%2C+Sri+Gadde%2C+and+Krishnaram+Kenthapadi%2E+2025%2E+RedactOR%3A+An+LLM%2Dpowered+framework+for+automatic+clinical+data+de%2Didentification%2E)
19. Thomas Sounack, Joshua Davis, Brigitte Durieux, Antoine Chaffin, Tom J. Pollard, Eric Lehman, Alistair E. W. Johnson, Matthew McDermott, Tristan Naumann, and Charlotta Lindvall. 2025. BioClinical ModernBERT: A state-of-the-art long-context encoder for biomedical and clinical NLP. Preprint, arxiv:250… — [arXiv:2506.10896](https://arxiv.org/abs/2506.10896) [find](https://scholar.google.com/scholar?q=Thomas+Sounack%2C+Joshua+Davis%2C+Brigitte+Durieux%2C+Antoine+Chaffin%2C+Tom+J%2E+Pollard%2C+Eric+Lehman%2C+Alistair+E%2E+W%2E+Johnson%2C+Matthew+McDermott%2C+Tristan+Naumann%2C+and+Charlotta+Lindvall%2E+2025%2E+BioClinical+Mode)
20. Amber Stubbs and Özlem Uzuner. 2015. Annotating longitudinal clinical narratives for de-identification: The 2014 i2b2/UTHealth corpus. Journal of Biomedical Informatics, 58 Suppl:S20–S29. [find](https://scholar.google.com/scholar?q=Amber+Stubbs+and+%C3%96zlem+Uzuner%2E+2015%2E+Annotating+longitudinal+clinical+narratives+for+de%2Didentification%3A+The+2014+i2b2%2FUTHealth+corpus%2E+Journal+of+Biomedical+Informatics%2C+58+Suppl%3AS20%E2%80%93S29%2E)
21. Simone Tedeschi, Valentino Maiorca, Niccolò Campolungo, Francesco Cecconi, and Roberto Navigli. 2021. WikiNEuRal: Combined neural and knowledge-based silver data creation for multilingual NER. In Findings of the Association for Computational Linguistics: EMNLP 2021, pages 2521–2533. Association for… [find](https://scholar.google.com/scholar?q=Simone+Tedeschi%2C+Valentino+Maiorca%2C+Niccol%C3%B2+Campolungo%2C+Francesco+Cecconi%2C+and+Roberto+Navigli%2E+2021%2E+WikiNEuRal%3A+Combined+neural+and+knowledge%2Dbased+silver+data+creation+for+multilingual+NER%2E+In+Find)
22. Ozlem Uzuner, Yuan Luo, and Peter Szolovits. 2007. Evaluating the state-of-the-art in automatic de-identification. Journal of the American Medical Informatics Association, 14(5):550–563. [find](https://scholar.google.com/scholar?q=Ozlem+Uzuner%2C+Yuan+Luo%2C+and+Peter+Szolovits%2E+2007%2E+Evaluating+the+state%2Dof%2Dthe%2Dart+in+automatic+de%2Didentification%2E+Journal+of+the+American+Medical+Informatics+Association%2C+14%285%29%3A550%E2%80%93563%2E)
23. Qingyun Wang, Xiaoman Pan, Lifu Huang, Boliang Zhang, Zhiying Jiang, Heng Ji, and Kevin Knight. 2018. Describing a knowledge base. In Proceedings of the 11th International Conference on Natural Language Generation, pages 10–21, Tilburg University, The Netherlands. Association for Computational Lingu… [find](https://scholar.google.com/scholar?q=Qingyun+Wang%2C+Xiaoman+Pan%2C+Lifu+Huang%2C+Boliang+Zhang%2C+Zhiying+Jiang%2C+Heng+Ji%2C+and+Kevin+Knight%2E+2018%2E+Describing+a+knowledge+base%2E+In+Proceedings+of+the+11th+International+Conference+on+Natural+Langua)
24. Hsiu-Wei Yang, Yanyan Zou, Peng Shi, Wei Lu, Jimmy Lin, and Xu Sun. 2019. Aligning cross-lingual entities with multi-aspect information. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing… [find](https://scholar.google.com/scholar?q=Hsiu%2DWei+Yang%2C+Yanyan+Zou%2C+Peng+Shi%2C+Wei+Lu%2C+Jimmy+Lin%2C+and+Xu+Sun%2E+2019%2E+Aligning+cross%2Dlingual+entities+with+multi%2Daspect+information%2E+In+Proceedings+of+the+2019+Conference+on+Empirical+Methods+in+N)
25. Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik Narasimhan, and Yuan Cao. 2023. [ReAct: Synergizing reasoning and acting in language models](https://md.research.elamur.ai/free/en/papers/yao-2022-react-synergizing-reasoning-acting-language/). Preprint, arxiv:2210.03629 [cs].
26. Barry Zhang, Keith Lazuka, and Mahesh Murag. 2025. Equipping agents for the real world with agent skills. Published October 16, 2025. Updated December 18, 2025 to announce Agent Skills as an open standard. [find](https://scholar.google.com/scholar?q=Barry+Zhang%2C+Keith+Lazuka%2C+and+Mahesh+Murag%2E+2025%2E+Equipping+agents+for+the+real+world+with+agent+skills%2E+Published+October+16%2C+2025%2E+Updated+December+18%2C+2025+to+announce+Agent+Skills+as+an+open+stan)
27. Guangcheng Zhu, Ruixuan Xiao, Haobo Wang, Zhen Zhu, Gengyu Lyu, and Junbo Zhao. 2025. Large margin representation learning for robust cross-lingual named entity recognition. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 4270… [find](https://scholar.google.com/scholar?q=Guangcheng+Zhu%2C+Ruixuan+Xiao%2C+Haobo+Wang%2C+Zhen+Zhu%2C+Gengyu+Lyu%2C+and+Junbo+Zhao%2E+2025%2E+Large+margin+representation+learning+for+robust+cross%2Dlingual+named+entity+recognition%2E+In+Proceedings+of+the+63rd)

*Updated 2 November 2026*