Elamur Research⌕
☰
home / papers / 2026 / Exploring Knowledge Graphs for Automatic Fake News Detection in Portuguese

Exploring Knowledge Graphs for Automatic Fake News Detection in Portuguese

Paper · 2026 · Lucas dos Santos, Manoel Rodrigues Euclides Santos, Yuri Silva Souza et al.

Summary

Most Portuguese fake news detectors rely on writing style or social-network propagation signals. Fluent falsehoods, including those produced by LLMs, can slip past such cues. This paper takes a fact-centred route instead. Each article is reduced to a set of events encoded as subject–predicate–object triples, which are then checked against a reference Knowledge Graph called KG-base. The graph is built only from the true articles of three public Brazilian benchmarks: Fake.Br, FakeRecogna and FakeTrueBR. False articles are reserved for evaluation. Triples are produced by a deterministic, rule-based extractor working on spaCy dependency parses. A verification algorithm then estimates veracity from how well the article's triples are connected in the graph. On the merged collection the approach reaches F1 = 0.8725. Its main advantage is explainability: each verdict can be traced to specific claims that are supported or unsupported. Error analysis points to Portuguese Open Information Extraction and limited graph coverage as the main bottlenecks.

Key points

  • KG-base is built from 11,342 true articles: 3,600 from Fake.Br, 1,791 from FakeTrueBR and 5,951 from FakeRecogna. The corresponding dataset totals are 7,200, 3,582 and 11,902.
  • Triple extraction uses spaCy pt_core_news_lg. Subjects come from nsubj, predicates from the lemmatized main verb, and objects from obj. A triple is kept only if all three parts are found, which favours precision over recall. NER and coreference resolution were deliberately left out.
  • Articles scoring roughly 0.40–0.59 are treated as uncertain. At prediction time the system abstains on them, while for two-class scoring they are assigned to the negative label. This is a conservative choice that lowers precision and accuracy.
  • False negatives mostly come from contextual fakes, where true facts are mixed with fabricated conclusions. The authors suggest narrative-level aggregation or contradiction detection as remedies.
  • The code is released at github.com/FAKTUS-BR/propor26 as part of the FAKTUS-BR project, and a web demo interface was also built.

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.Knowledge graphs & ontologies · 192 works

Cite

Cite as Lucas dos Santos, Manoel Rodrigues Euclides Santos, Yuri Silva Souza, João Pedro Holanda Sousa and Roney Lira de Sales Santos. 2026. Exploring Knowledge Graphs for Automatic Fake News Detection 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{santos2026exploring,
  title = {{Exploring Knowledge Graphs for Automatic Fake News Detection in Portuguese}},
  author = {Lucas {dos Santos}, Manoel Rodrigues Euclides Santos, Yuri Silva Souza, João Pedro Holanda Sousa and Roney Lira {de Sales Santos}},
  booktitle = {Proceedings of the 17th International Conference on Computational Processing of Portuguese (PROPOR 2026) - Vol. 1},
  year = {2026},
  url = {https://aclanthology.org/2026.propor-1.44/}
}

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. Gabor Angeli, Melvin Jose Johnson Premkumar, and Christopher D. Manning. 2015. Leveraging linguistic structure for open domain information extraction. In Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics (ACL).
  2. Sören Auer, Christian Bizer, Georgi Kobilarov, Jens Lehmann, Richard Cyganiak, and Zachary Ives. 2007. Dbpedia: A nucleus for a web of open data. In Proceedings of the 6th International Semantic Web Conference (ISWC). find
  3. Michele Banko, Michael J. Cafarella, Stephen Soderland, Matt Broadhead, and Oren Etzioni. 2007. Open information extraction from the web. In Proceedings of the 20th International Joint Conference on Artificial Intelligence (IJCAI), pages 2670–2676.
  4. Juan Pablo Chavarro, Jonata Tyska Carvalho, Tarlis Tortelli Portela, and Jonathan Cardoso Silva. 2023. Faketruebr: um corpus brasileiro de notícias falsas. In Anais da Escola Regional de Banco de Dados (ERBD), Palmas, PR. find
  5. Giovanni Luca Ciampaglia, Prashant Shiralkar, Luis M. Rocha, Johan Bollen, Filippo Menczer, and Alessandro Flammini. 2015. Computational fact checking from knowledge networks. PLOS ONE, 10(6):e0128193. find
  6. Luciano Del Corro and Rainer Gemulla. 2013. Clausie: Clause-based open information extraction. In Proceedings of the 22nd International Conference on World Wide Web (WWW), pages 355–366. find
  7. Oren Etzioni, Anthony Fader, Janara Christensen, Stephen Soderland, and Mausam. 2011. Open information extraction: The second generation. AI Magazine, 32(3):3–13. find
  8. Gabriel Lino Garcia, Pedro Henrique Paiola, Danilo Samuel Jodas, Luis Afonso Sugi, and João Paulo Papa. 2024a. Text summarization and temporal learning models applied to portuguese fake news detection in a novel brazilian corpus dataset. In Proceedings of the 16th International Conference on Computa… find
  9. Marcos P. S. Gôlo, Adriel L. V. Mori, William G. Oliveira, Jacson R. Barbosa, Valdemar V. Graciano-Neto, Eliomar A. de Lima, and Ricardo M. Marcacini. 2024. On the use of large language models to detect brazilian politics fake news. In Proceedings of the Encontro Nacional de Inteligência Artificial… find
  10. Marcos P. S. Gôlo, Adriel L. V. Mori, William G. Oliveira, Jacson R. Barbosa, Valdemar V. Graciano-Neto, Eliomar A. Lima, and Ricardo M. Marcacini. 2025. Evaluating large language models for brazilian political fake news detection. In Proceedings of the Encontro Nacional de Inteligência Artificial e… find
  11. Aric Hagberg, Pieter J. Swart, and Daniel A. Schult. 2008. Exploring network structure, dynamics, and function using networkx. Technical report, Los Alamos National Laboratory. find
  12. Matthew Honnibal, Ines Montani, Sofie Van Lan-deghem, and Adriane Boyd. 2020. spacy: Industrial-strength natural language processing in python. find
  13. David M. J. Lazer, Matthew A. Baum, Yochai Benkler, Adam J. Berinsky, Kelly M. Greenhill, Filippo Menczer, Miriam J. Metzger, Brendan Nyhan, Gordon Pennycook, David Rothschild, Michael Schudson, Steven A. Sloman, Cass R. Sunstein, Emily A. Thorson, Duncan J. Watts, and Jonathan L. Zittrain. 2018. Th… find
  14. Rafael A. Monteiro, Roney L. S. Santos, Thiago A. S. Pardo, Tiago A. Almeida, Evandro E. S. Ruiz, and Oto A. Vale. 2018. Contributions to the study of fake news in portuguese: new corpus and automatic detection results. In Computational Processing of the Portuguese Language, Lecture Notes in Artific… find
  15. Leandro de Oliveira and Diego B. Claro. 2019. Dptoie: a portuguese open information extraction system based on dependency analysis. Disponível no Repositório da UFBA. find
  16. Gustavo L. Garcia, Lucas C. Afonso, and João P. Papa. 2022. Fakerecogna: a new brazilian corpus for fake news detection. In Computational Processing of the Portuguese Language, Lecture Notes in Artificial Intelligence, pages 57–67. Springer. find
  17. Gustavo L. Garcia, Lucas C. Afonso, and João P. Papa. 2024b. Fakerecogna 2.0: Advances in brazilian fake news detection. In PROPOR. find
  18. Rafael Glauber, Diego B. Claro, and Leandro S. Oliveira. 2019. Dependency parser on open information extraction for portuguese texts: Dptoie and dependentie on iberlef. In Proceedings of IberLEF, pages 442– 448. find
  19. Jeff Z. Pan, Siyana Pavlova, Chenxi Li, Ning Li, Yang Li, and Jie Liu. 2018. Content-based fake news detection using knowledge graphs. In The Semantic Web – ISWC 2018, pages 669–683. Springer. find
  20. Nils Reimers and Iryna Gurevych. 2019. Sentence-bert: Sentence embeddings using siamese bert-networks. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing (EMNLP).
  21. Victoria L. Rubin, Yimin Chen, and Niall J. Conroy. 2015. Deception detection for news: Three types of fakes. In Proceedings of the Association for Information Science and Technology, pages 1–4. Wiley. find
  22. Lucas Santos, Yuri Silva Souza, Manoel Santos, and Roney Santos. 2025. Análise e uso de grafos de conhecimento para detecção de eventos em notícias falsas. In Congresso UFBA 2025: Imaginar futuros – desafios da emergência climática, pages 214– 214, Salvador, Brazil. Universidade Federal da Bahia. Ca… find
  23. Roney L. S. Santos and Thiago A. S. Pardo. 2020. Fact-checking for portuguese: knowledge graph and google search-based methods. In Computational Processing of the Portuguese Language, Lecture Notes in Artificial Intelligence, pages 195–205. Springer. find
  24. Renato M. Silva, Roney L. Santos, Tiago A. Almeida, and Thiago A. Pardo. 2020. Towards automatically filtering fake news in portuguese. Expert Systems with Applications, 146:113199. find
  25. Fábio Souza, Rodrigo Nogueira, and Roberto Lotufo. 2020. Bertimbau: Pretrained bert models for brazilian portuguese. In Proceedings of the Brazilian Conference on Intelligent Systems (BRACIS). Springer. find
  26. Yuri Silva Souza, Lucas Santos, Manoel Santos, and Roney Santos. 2025. Fakecheck 2.0: Aplicativo para detecção de notícias falsas em português. In Congresso UFBA 2025: Imaginar futuros – desafios da emergência climática, pages 558–558, Salvador, Brazil. Universidade Federal da Bahia. Caderno de Resu… find
  27. Soroush Vosoughi, Deb Roy, and Sinan Aral. 2018. The spread of true and false news online. Science, 359(6380):1146–1151. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/santos-2026-exploring-knowledge-graphs-automatic-fake/