Elamur Research⌕
☰
home / papers / 2026 / PHMartialLawNER: A Tagalog Named Entity Recognition Corpus for the Philippine Martial Law Era

PHMartialLawNER: A Tagalog Named Entity Recognition Corpus for the Philippine Martial Law Era

Paper · 2026 · Abdiel Clarence Tabuzo, Vladimir Gray Velazco, Cassandra Cabral et al.

Summary

Tagalog NER resources such as TLUnified-NER are built almost entirely from contemporary text. Models trained on them handle archival material poorly because of older spelling, obsolete titles, and actors or organizations specific to an earlier time. The authors target Tagalog writing from the Martial Law period under Marcos Sr. (1972–1986), much of which is undigitized or computationally unexplored. They collected opposition and underground newspapers from the eVols archive (University of Hawai‘i) and the Bantayog ng mga Bayani digital library, then ran OCR with manual cleanup. GLiNER in zero-shot mode produced silver labels, which three annotators corrected in Argilla over several rounds while refining the guidelines. The resulting gold-standard corpus, PHMartialLawNER, uses a fine-grained, historically motivated label set. Benchmarks with GLiNER variants, calamanCy pipelines and transformers show the task can be learned even with little data. calamanCy-trf reaches the best scores, while rare, fine-grained categories remain difficult. The resource supports Tagalog NLP and digital humanities work on Philippine archives.

Key points

  • Corpus size and agreement: about 13k sentence segments (≈362,000 tokens), merged into roughly 8k annotated spans. Annotators agree with a Cohen's κ of 0.86 when every token is counted and 0.72 when only tokens carrying a label are considered, with a pairwise F1 of 0.74.
  • Label schema: Person (Individual/Collective), Organization (Political, Government, Other), Event (Local/International), Production (Media, Government, Doctrine), Time, Numerical Statistics, Location, and Object (weapons). The per-entity evaluation table also contains an Organization-Military label.
  • Learning curve: the data are split 70:10:20 with stratified sampling. In few-shot training, calamanCy_trf rises from 49.84 overall F1 with 88 training samples to 83.09 with 5,602. Production-Media, Event and Object-Weapon are the weakest labels when data are scarce.
  • Statistical analysis: a two-way ANOVA finds that model, training-set size and their interaction all have significant effects (p < .001). Gains level off around 1,400–2,801 samples. GLiNER does better in low-data settings, while calamanCy-trf reaches the highest peak.
  • Main error source: metonymy, for example "Malacañang" or "US" read as either a government or a location, and "Batas Militar" sometimes tagged as an organization.

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

Also on the byline: University of Santo Tomas.

 Fields, topics and concepts

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksLiterature and Literary Theory · 519 works

🧭 Topics 2Research subjects and languages the work is about.Named entity recognition · 200 worksLegal AI · 134 works

Cite

Cite as Abdiel Clarence Tabuzo, Vladimir Gray Velazco, Cassandra Cabral, Moneah Shaila Lacsam and Charmaine Salvador Ponay. 2026. PHMartialLawNER: A Tagalog Named Entity Recognition Corpus for the Philippine Martial Law Era. Proceedings of the 6th International Conference on Natural Language Processing for the Digital Humanities.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.nlp4dh-1.16

@inproceedings{tabuzo2026phmartiallawner,
  title = {{PHMartialLawNER: A Tagalog Named Entity Recognition Corpus for the Philippine Martial Law Era}},
  author = {Abdiel Clarence Tabuzo, Vladimir Gray Velazco, Cassandra Cabral, Moneah Shaila Lacsam and Charmaine Salvador Ponay},
  booktitle = {Proceedings of the 6th International Conference on Natural Language Processing for the Digital Humanities},
  year = {2026},
  doi = {10.18653/v1/2026.nlp4dh-1.16},
  url = {https://doi.org/10.18653/v1/2026.nlp4dh-1.16}
}

References

25 references, 4 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. N. Abadie, E. Carlinet, J. Chazalon, and B. Duménieu. 2022. A Benchmark of Named Entity Recognition Approaches in Historical Documents Application to 19th Century French Directories. In Document Analysis Systems. find
  2. DAS 2022., number 13237 in Document Analysis Systems. find
  3. DAS 2022., La Rochelle, France. Springer, Cham. find
  4. Ron Artstein. 2017. Inter-annotator Agreement, pages 297–313. Springer Netherlands, Dordrecht. find
  5. Joseph Baumgartner. 1981. Newspapers as historical sources. Philippine Quarterly of Culture and Society, 9(3):256–258. find
  6. Alexis Conneau, Kartikay Khandelwal, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzmán, Edouard Grave, Myle Ott, Luke Zettlemoyer, and Veselin Stoyanov. 2020. Unsupervised cross-lingual representation learning at scale. In Proceedings of the 58th Annual Meeting of the Association fo… find
  7. Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. BERT: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies,…
  8. Maud Ehrmann, Ahmed Hamdi, Elvys Linhares Pontes, Matteo Romanello, and Antoine Doucet. 2023. Named entity recognition and classification in historical documents: A survey. ACM Computing Surveys, 56(2):1–47. find
  9. Frédéric Kaplan and Isabella di Lenardo. 2017. Big data of the past. Frontiers in Digital Humanities, Volume 4 - 2017. find
  10. Xuezhe Ma and Eduard Hovy. 2016. End-to-end sequence labeling via bi-directional lstm-cnns-crf. Preprint, arXiv:1603.01354.
  11. Gloria E. Melencio. 2023. The fall and rise of the marcoses: From mosquito press to troll farms. UP Los Baños Journal. find
  12. Lester James V. Miranda. 2023a. calamanCy: A Tagalog natural language processing toolkit. In Proceedings of the 3rd Workshop for Natural Language Processing Open Source Software (NLP-OSS 2023), pages 1–7, Singapore. Association for Computational Linguistics. find
  13. Lester James V. Miranda. 2023b. Developing a named entity recognition dataset for Tagalog. In Proceedings of the First Workshop in South East Asian Language Processing, pages 13–20, Nusa Dua, Bali, Indonesia. Association for Computational Linguistics. find
  14. A. Lin Neumann. 2005. The philippines: Amid troubles, a rich press tradition. Committee to Protect Journalists. find
  15. Sarah Oberbichler. 2024. Large-scale research with historical newspapers: A turning point through generative ai. DH Lab, Leibniz Institute of European History (IEG). find
  16. James Ald Teves, Ray Daniel Cal, Josh Magdiel Vil-laluz, Jean Malolos, Mico Magtira, Ramon Rodriguez, Mideth Abisado, and Joseph Marvin Imperial. 2025. Hiligayner: A baseline named entity recognition model for hiligaynon. Preprint, arXiv:2510.10776. find
  17. 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.
  18. Konstantin Todorov and Giovanni Colavizza. 2022. An assessment of the impact of ocr noise on language models. Preprint, arXiv:2202.00470. — [arXiv:2202.00470](https://arxiv.org/abs/2202.00470) find
  19. University of Hawai'i at Mānoa Philippine Collection. 2023. Papers of the underground movement during the marcos regime (philippine radical papers of the marcos regime). Philippine Studies Digital Collection. find
  20. Urchade Zaratiana, Nadi Tomeh, Pierre Holat, and Thierry Charnois. 2024. GLiNER: Generalist model for named entity recognition using bidirectional transformer. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technol…
  21. Ronalyn V. Olea. 2012. Underground press during martial law: Piercing the veil of darkness imposed by the dictatorship. Bulatlat. find
  22. Ma. Beatrice Emanuela Pilar, Dane Dedoroy, El-lyza Mari Papas, Mary Loise Buenaventura, Myron Darrel Montefalcon, Jay Rhald Padilla, Joseph Marvin Imperial, Mideth Abisado, and Lany Maceda. 2023. CebuaNER: A new baseline Cebuano named entity recognition model. In Proceedings of the 37th Pacific Asia… find
  23. Ricardo L. Punzalan. 2006. Archives of the new possession: Spanish colonial records and the american creation of a ‘national’ archives for the philippines. Archival Science, 6(3):381–392. find
  24. Shruti Rijhwani and Daniel Preotiuc-Pietro. 2020. Temporally-informed analysis of named entity recognition. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 7605–7617, Online. Association for Computational Linguistics. find
  25. Florangel Rosario-Braid and Ramon R. Tuazon. 1986. Communication media in the philippines: 1521– 1986. Philippine Social Science Council. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/tabuzo-2026-phmartiallawner-tagalog-named-entity-recognition/