Elamur Research⌕
☰
home / papers / 2026 / Data Augmentation for Named Entity Recognition in Domain-Specific Scenarios in Portuguese

Data Augmentation for Named Entity Recognition in Domain-Specific Scenarios in Portuguese

Paper · 2026 · Higor Moreira, Patricia Ferreira da Silva, Luciana Bencke et al.

Summary

Portuguese NER in specialized domains such as oil & gas has little annotated data, and manual labeling is expensive. Augmentation is harder for NER than for text classification because altering tokens can break the alignment between words and their BIO labels. The authors compare seven augmentation strategies on four Portuguese domain-specific datasets, using training subsets of several sizes:

  • rule-based: synonym and mention replacement, plus a context-aware variant
  • MLM-based generation (MELM)
  • back-translation of entity-free context segments
  • LLM paraphrasing
  • retrieval-augmented LLM generation

Each augmented set is merged with the original sampled instances and used to fine-tune Transformer-based NER models, which are then scored on the test split. Almost every technique beats the non-augmented baseline when few original instances are available. LLM paraphrasing (PP-LLM) performs best, followed by SR and MR. A domain expert's blind review of PetroGeoNER outputs shows that keeping entity labels correct matters more than fluent text. The study shows that cheap heuristics remain competitive for low-resource, domain-specific NER, and the augmented datasets are released publicly.

Key points

  • Techniques compared: SR, MR, CAMR, MELM, BT, PP-LLM and RAG-DA. RAG-DA was run only on PetroGeoNER, the only dataset with a segmented in-domain document collection.
  • Ranking by micro-F1: PP-LLM first, then SR, then MR.
  • In the human evaluation, 88% of SR outputs received the top span-quality score, even though their fluency was weak.
  • MELM had the poorest entity spans: only 26% received the top score and 64% fell into the two lowest categories.
  • RAG-DA was the most fluent method, with 96% at the top score and higher than the original sentences. However, 48% of its outputs contained mislabeled or hallucinated entities.
  • The augmented data is available on Hugging Face as hmoreira/da-datasets-propor2026.

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

 Fields, topics and concepts

📚 Shelves 2The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 works

🧭 Topics 1Research subjects and languages the work is about.Named entity recognition · 200 works

Cite

Cite as Higor Moreira, Patricia Ferreira da Silva, Luciana Bencke and Viviane Moreira. 2026. Data Augmentation for Named Entity Recognition in Domain-Specific Scenarios in Portuguese. Proceedings of the 17th International Conference on Computational Processing of Portuguese (PROPOR 2026) - Vol. 1.

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

@inproceedings{moreira2026data,
  title = {{Data Augmentation for Named Entity Recognition in Domain-Specific Scenarios in Portuguese}},
  author = {Higor Moreira, Patricia Ferreira {da Silva}, Luciana Bencke and Viviane Moreira},
  booktitle = {Proceedings of the 17th International Conference on Computational Processing of Portuguese (PROPOR 2026) - Vol. 1},
  year = {2026},
  url = {https://aclanthology.org/2026.propor-1.25/}
}

References

27 references, 3 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Daniela Oliveira Ferreira Amaral. 2017. Reconhecimento de entidades nomeadas na Área da geologia: bacias sedimentares brasileiras. Ph.D. thesis, Pontifícia Universidade Católica do Rio Grande do Sul. find
  2. Markus Bayer, Marc-André Kaufhold, and Christian Reuter. 2022. A survey on data augmentation for text classification. ACM Computing Surveys, 55(7):1–39. find
  3. Jianlyu Chen, Shitao Xiao, Peitian Zhang, Kun Luo, Defu Lian, and Zheng Liu. 2024. M3-embedding: Multi-linguality, multi-functionality, multi-granularity text embeddings through self-knowledge distillation. In Findings of the association for computational linguistics: ACL 2024, pages 2318–2335.
  4. Alexis Conneau, Kartikay Khandelwal, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzmán, Edouard Grave, Myle Ott, Luke Zettlemoyer, and Veselin Stoyanov. 2019. Unsupervised cross-lingual representation learning at scale. CoRR, abs/1911.02116. find
  5. Rosimeire Costa, Hidelberg Oliveira Albuquerque, Gabriel Silvestre, Nádia Félix F. Silva, Ellen Souza, Douglas Vitório, Augusto Nunes, Felipe Siqueira, Eduard Hovy, Mitchell Marcus, Martha Palmer, Lance Ramshaw, and Ralph Weischedel. 2006. OntoNotes: The 90% solution. In Proceedings of the Human Lan… find
  6. Xuming Hu, Yong Jiang, Aiwei Liu, Zhongqiang Huang, Pengjun Xie, Fei Huang, Lijie Wen, and Philip S Yu. 2022. Entda: Entity-to-text based data augmentation approach for named entity recognition tasks. arXiv preprint arXiv:2210.10343. — [arXiv:2210.10343](https://arxiv.org/abs/2210.10343) find
  7. Aaron Hurst, Adam Lerer, Adam P. Goucher, Adam Perelman, and 1 others. 2024. Gpt-4o system card. Preprint, arXiv:2410.21276.
  8. Hang Jiang, Yining Hua, Doug Beeferman, and Deb Roy. 2022. Annotating the tweebank corpus on named entity recognition and building nlp models for social media analysis. In Proceedings of the Language Resources and Evaluation Conference, pages 7199–7208. find
  9. Daniel Jurafsky and James H. Martin. 2026. Speech and Language Processing: An Introduction to Natural Language Processing, Computational Linguistics, and Speech Recognition with Language Models, 3rd edition. Online manuscript released January 12, 2025. find
  10. Imed Keraghel, Stanislas Morbieu, and Mohamed Nadif. 2024. Recent Advances in Named Entity Recognition: A Comprehensive Survey and Comparative Study. Working paper or preprint. find
  11. Hyunjae Kim, Jaehyo Yoo, Seunghyun Yoon, Jinhyuk Lee, and Jaewoo Kang. 2022. Simple questions generate named entity recognition datasets. In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pages 6220–6236. find
  12. Lucas Lima de Oliveira, Regis Kruel Romeu, and Viviane Pereira Moreira. 2021. Regis: A test collection for geoscientific documents in portuguese. In Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 2363–2368. find
  13. Pedro H. Luz de Araujo, Teófilo E. de Campos, Renato R. R. de Oliveira, Matheus Stauffer, Samuel Couto, and Paulo Bermejo. 2018. LeNER-Br: a dataset for named entity recognition in Brazilian legal text. In International Conference on the Computational Processing of Portuguese (PROPOR), Lecture Notes… find
  14. Higor Moreira, Patricia Ferreira da Silva, Renata Vieira, and Viviane Moreira. 2025. Petrogeoner: A refined and unified dataset for ner in the oil & gas domain. In Simpósio Brasileiro de Tecnologia da Informação e da Linguagem Humana (STIL), pages 259–271. SBC. find
  15. Artur Nowakowski, Gabriela Pałka, Kamil Guttmann, and Mikołaj Pokrywka. 2022. Adam mickiewicz university at wmt 2022: Ner-assisted and quality-aware neural machine translation. In Proceedings of the Seventh Conference on Machine Translation (WMT), pages 326–334. find
  16. Rafael Oleques Nunes, Dennis Giovani Balreira, André Suslik Spritzer, and Carla Maria Dal Sasso Freitas. 2024a. A named entity recognition approach for Portuguese legislative texts using self-learning. In Proceedings of the 16th International Conference on Computational Processing of Portuguese - Vo… find
  17. Rafael Oleques Nunes, André Susliz Spritzer, Carla Maria Dal Sasso Freitas, and Dennis Giovani Balreira. 2024b. Reconhecimento de entidades nomeadas e vazamento de dados em textos legislativos: Uma reavaliação da literatura. Linguamática, 16. find
  18. Nadeesha Perera, Matthias Dehmer, and Frank Emmert-Streib. 2020. Named entity recognition and relation detection for biomedical information extraction. Frontiers in cell and developmental biology, 8:673. find
  19. Saket Sharma, Aviral Joshi, Yiyun Zhao, Namrata Mukhija, Hanoz Bhathena, Prateek Singh, and Sashank Santhanam. 2023. When and how to paraphrase for named entity recognition? In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 7052… find
  20. Priscilla Silva, Arthur Franco, Thiago Santos, Mozar Brito, and Denilson Pereira. 2023. Cachacaner: a dataset for named entity recognition in texts about the cachaça beverage. Lang. Resour. Eval., 58(4):1315–1333. find
  21. Sihan Song, Furao Shen, and Jian Zhao. 2024. Ropda: robust prompt-based data augmentation for low-resource named entity recognition. In Proceedings of the AAAI conference on artificial intelligence, volume 38, pages 19017–19025. find
  22. Erik F. Tjong Kim Sang and Fien De Meulder. 2003. Introduction to the CoNLL-2003 shared task: Language-independent named entity recognition. In Proceedings of the Seventh Conference on Natural Language Learning at HLT-NAACL 2003, pages 142– 147.
  23. Arthur Elwing Torres, Edleno Silva de Moura, Altigran Soares da Silva, Mario A Nascimento, and Filipe Mesquita. 2024. An experimental study on data augmentation techniques for named entity recognition on low-resource domains. arXiv preprint arXiv:2411.14551. — [arXiv:2411.14551](https://arxiv.org/abs/2411.14551) find
  24. Usama Yaseen and Stefan Langer. 2021. Data augmentation for low-resource named entity recognition using backtranslation. In Proceedings of the 18th International Conference on Natural Language Processing (ICON), pages 352–358. find
  25. Junjie Ye, Nuo Xu, Yikun Wang, Jie Zhou, Qi Zhang, Tao Gui, and Xuanjing Huang. 2024. Llm-da: Data augmentation via large language models for few-shot named entity recognition. Preprint, arXiv:2402.14568. — [arXiv:2402.14568](https://arxiv.org/abs/2402.14568) find
  26. Ran Zhou, Xin Li, Ruidan He, Lidong Bing, Erik Cambria, Luo Si, and Chunyan Miao. 2022. MELM: Data augmentation with masked entity language modeling for low-resource NER. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 2251–22… find
  27. Elisa Terumi Rubel Schneider, João Vitor Andrioli de Souza, Julien Knafou, Lucas Emanuel Silva e Oliveira, Jenny Copara, Yohan Bonescki Gumiel, Lucas Ferro Antunes de Oliveira, Emerson Cabrera Paraiso, Douglas Teodoro, and Cláudia Maria Cabral Moro Barra. 2020. Biobertpt-a portuguese neural language… find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/moreira-2026-data-augmentation-named-entity-recognition/