Elamur Research⌕
☰
home / papers / 2026 / Topic Modeling in Brazilian Portuguese Documents on Antimicrobial Resistance

Topic Modeling in Brazilian Portuguese Documents on Antimicrobial Resistance

Paper · 2026 · Enrique Reis Susin and Lilian Berton

Summary

This study examines how different parts of Brazilian society talk about antimicrobial resistance. The goal is to give epidemiological surveillance and health policy a computational view of public and institutional discourse. The authors built three Portuguese-language collections with a shared keyword list (disease names, drug names and general resistance concepts):

  • tweets gathered via TwitterAPI.io
  • articles scraped from the science and health sections of the G1 news portal
  • official texts from .gov.br sites

After a Portuguese-specific cleaning pipeline, they compared two short-text topic modeling approaches: the probabilistic GSDMM and the embedding-based BERTopic. The analysis focused on two themes, bacterial resistance and tuberculosis.

The three sources play distinct roles:

  • Twitter mixes personal experience, humor and misinformation.
  • G1 translates science for a wider audience, sometimes in a sensational way.
  • Gov.br is normative, emphasizing clinical protocols and free treatment through SUS.

All sources share a biomedical core of prevention, diagnosis and treatment. Antimicrobial resistance and HIV coinfection recur in every dataset. The authors argue that such overlaps could anchor future automated monitoring of public health information.

Key points

  • Corpus: 70,103 documents in total, made up of 64,225 tweets (2008–2025), 4,363 G1 articles and 1,515 .gov.br publications.
  • Preprocessing: duplicate removal, LangDetect filtering to keep only Portuguese ("pt"), noise cleaning (mentions, URLs, emojis, laughter strings like "kkkkk"), and spaCy Portuguese tokenization, stop-word removal and lemmatization.
  • Models: GSDMM served as the probabilistic baseline. BERTopic used BERTimbau Base embeddings via sentence-transformers, with HDBSCAN for clustering. LDA was set aside as unsuitable for sparse short texts.
  • Tuberculosis on Twitter: seven refinement rounds produced four topic dimensions: clinical/preventive, coinfections, social/sanitary, and technical-scientific.
  • Future work: LLM-based augmentation of short texts into pseudo-documents, prompt-based topic inference, and human-in-the-loop evaluation such as word intrusion tasks and expert labeling.

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

💡 Concepts 1Key ideas and methods explained on our concept pages.Topic modeling · 47 works

Cite

Cite as Enrique Reis Susin and Lilian Berton. 2026. Topic Modeling in Brazilian Portuguese Documents on Antimicrobial Resistance. 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{susin2026topic,
  title = {{Topic Modeling in Brazilian Portuguese Documents on Antimicrobial Resistance}},
  author = {Enrique Reis Susin and Lilian Berton},
  booktitle = {Proceedings of the 17th International Conference on Computational Processing of Portuguese (PROPOR 2026) - Vol. 1},
  year = {2026},
  url = {https://aclanthology.org/2026.propor-1.10/}
}

References

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

  1. Brazilian Government. 2023. Gov.br open data portal. https://www.gov.br/. ↗
  2. Maarten Grootendorst. 2022. Bertopic: Neural topic modeling with a class-based tf-idf procedure. arXiv preprint arXiv:2203.05794.
  3. Grupo Globo. 2023. G1 news portal. https://g1.globo.com/. ↗
  4. Antônio Pereira De Souza Júnior, Pablo Cecilio, Felipe Viegas, Washington Cunha, Elisa Tuler De Alber-garia, and Leonardo Chaves Dutra Da Rocha. 2022. Evaluating topic modeling pre-processing pipelines for portuguese texts. In Proceedings of the Brazilian Symposium on Multimedia and the Web, pages 1… find
  5. Caitlin Doogan Poet Laureate, Wray Buntine, and Henry Linger. 2023. A systematic review of the use of topic models for short text social media analysis. Artificial Intelligence Review, 56(12):14223–14255. find
  6. Theme / Source Twitter G1 / Gov.br Bacterial Resistance Tuberculosis Informal debate; veterinary use; misinformation Campaigns; coinfections; social context Outbreak reports; microbiological research Clinical protocols; international health alerts *Table 3: Comparison of discourse patterns across da… find
  7. World Health Organization. 2024. Who bacterial priority pathogens list, 2024: Bacterial pathogens of public health importance to guide research, development and strategies to prevent and control antimicrobial resistance. Technical report, World Health Organization, Geneva. find
  8. Leland McInnes, John Healy, Steve Astels, and 1 others. 2017. hdbscan: Hierarchical density based clustering. J. Open Source Softw., 2(11):205.
  9. Xiaobao Wu, Thong Nguyen, and Anh Tuan Luu. 2024. A survey on neural topic models: methods, applications, and challenges. Artificial Intelligence Review, 57(2):18. find
  10. Christopher JL Murray, Kevin Shunji Ikuta, Fablina Sharara, Lucien Swetschinski, Gisela Robles Aguilar, Authia Gray, Chieh Han, Catherine Bisignano, Puja Rao, Eve Wool, and 1 others. 2022. Global burden of bacterial antimicrobial resistance in 2019: a systematic analysis. The lancet, 399(10325):629–… find
  11. Jim O’Neill. 2016. Tackling drug-resistant infections globally: final report and recommendations. Technical report, Review on Antimicrobial Resistance, London. Commissioned by the UK Government. find
  12. Jipeng Qiang, Zhenyu Qian, Yun Li, Yunhao Yuan, and Xindong Wu. 2020. Short text topic modeling techniques, applications, and performance: a survey. IEEE Transactions on Knowledge and Data Engineering, 34(3):1427–1445. find
  13. Xiaojun Quan, Chunyu Kit, Yong Ge, and Sinno Jialin Pan. 2015. Short and sparse text topic modeling via self-aggregation. In 24th international joint conference on artificial intelligence, IJCAI 2015, pages 2270–2276. AAAI Press/International Joint Conferences on Artificial Intelligence. find
  14. X Corp. 2025. X (formerly twitter). find
  15. Xiaohui Yan, Jiafeng Guo, Yanyan Lan, and Xueqi Cheng. 2013a. A biterm topic model for short texts. In Proceedings of the 22nd international conference on World Wide Web, pages 1445–1456. find
  16. Xiaohui Yan, Jiafeng Guo, Yanyan Lan, and Xueqi Cheng. 2013b. A biterm topic model for short texts. Proceedings of the 22nd International Conference on World Wide Web, pages 1445–1456. find
  17. Jianhua Yin and Jianyong Wang. 2014. A dirichlet multinomial mixture model-based approach for short text clustering. In Proceedings of the 20th ACM SIGKDD international conference on Knowledge discovery and data mining, pages 233–242. find
  18. Nils Reimers and Iryna Gurevych. 2019. Sentence-bert: Sentence embeddings using siamese bert-networks. arXiv preprint arXiv:1908.10084.
  19. Nádia FF da Silva, Marília Costa R Silva, Fabíola SF Pereira, João Pedro M Tarrega, João Vitor P Beinotti, Márcio Fonseca, Francisco Edmundo de Andrade, and André CP de LF de Carvalho. 2021. Evaluating topic models in portuguese political comments about bills from brazil’s chamber of deputies. In Br… find
  20. Fábio Souza, Rodrigo Nogueira, and Roberto Lotufo. 2020. Bertimbau: pretrained bert models for brazilian portuguese. In Brazilian conference on intelligent systems, pages 403–417. Springer. find
  21. United Nations General Assembly. 2015. Transforming our world: the 2030 agenda for sustainable development. Resolution adopted by the General Assembly on 25 September 2015 (A/RES/70/1). find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/susin-2026-topic-modeling-brazilian-portuguese-documents/