Semantic tagging of named entities in The Lusiads
Original (Portuguese): Marcação semântica de entidades nomeadas em Os Lusíadas
Paper · 2026 · Adriane Maria de Oliveira Queiroz and Bruno Oliveira Maroneze
Summary
The paper addresses semantic annotation of named entities in Luís de Camões's epic Os Lusíadas. Off-the-shelf NER handles the poem poorly because of its dense mix of historical and mythological names, Latin-influenced syntax and the irregular spelling of the 1572 edition. Within the "Lusíadas Digital" project, the authors design a TEI P5 model. It encodes persons with <persName> (typed as historical or mythological), places with <placeName> (real or mythical), and epithets and antonomasia with <rs> linked through @ref. They process both the 1572 edition and a modernized Project Gutenberg text with a Python pipeline: an authority gazetteer, spaCy NER, and a regex routine that rejoins names split by line-break tags. Manual philological post-editing follows. spaCy, trained on news text, confuses deities with places, treats capitalized common nouns as persons and misses epithets, so the gazetteer and human curation proved more dependable. The resulting linked TEI edition enables entity-level queries, future social network analysis and digital mapping of the voyages.
Key points
- The pipeline has three stages: a gazetteer for frequent, unambiguous entities; spaCy
pt_core_news_lgfor uncatalogued names; and regex protection of names split by<lb/>(e.g. Lusi<lb/>tana) in the 1572 text. - The two inputs are the 1572 edition, which keeps archaic spelling plus
<lb/>and<fw>markup, and the normalized Project Gutenberg edition. - Observed errors include Marte tagged as a location, "Fama" and "Mar" tagged as persons, and epithets like "o grão Macedônio" ignored.
- In the "Lampetusa" (I, 46) case, NER suggests the island, while the curated encoding links it to the Heliades. This draws on the dictionaries of Peixoto & Pinto (1924) and Nogueira (1960) and on Silva Dias's commentary.
- The
<teiHeader>holds a prosopography and a gazetteer with uniquexml:ids, so variants such as "Lusitana" and "Lusa" resolve to#portugal. No quantitative evaluation scores are reported.
Fields, topics and concepts
📚 Shelves 2The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksArtificial Intelligence · 17,965 works
Cite
Cite as Adriane Maria de Oliveira Queiroz and Bruno Oliveira Maroneze. 2026. Marcação semântica de entidades nomeadas em Os Lusíadas. Proceedings of the 17th International Conference on Computational Processing of Portuguese (PROPOR 2026) - Vol. 2.
@inproceedings{queiroz2026marcacao,
title = {{Marcação semântica de entidades nomeadas em Os Lusíadas}},
author = {Adriane Maria {de Oliveira Queiroz} and Bruno Oliveira Maroneze},
booktitle = {Proceedings of the 17th International Conference on Computational Processing of Portuguese (PROPOR 2026) - Vol. 2},
year = {2026},
url = {https://aclanthology.org/2026.propor-2.38/}
}References
- Vítor Manuel de Aguiar e Silva. 2010. Camões: Labir-intos e Fascínios. Cotovia, Lisboa. find
- Luís de Camões. 1916. Os Lusíadas. Companhia Por-tugueza Editora, Porto. Comentados por Augusto Epiphanio da Silva Dias. 2. ed. melhorada. Tomo I. find
- Algirdas Julien Greimas. 1973. Semântica estrutural: pesquisa de método. Cultrix, São Paulo. find
- Franco Moretti. 2011. Network theory, plot analysis. Pamphlet 2, Stanford Literary Lab. find
- Júlio Nogueira. 1960. Dicionário e Gramática de "Os Lusíadas". Livraria Freitas Bastos S.A., Rio de Janeiro. find
- Afranio Peixoto and Pedro A. Pinto. 1924. Dicionário d’Os Lusíadas de Luís de Camões. Livraria Francisco Alves - Casa de Paulo Azevedo e Cia., Rio de Janeiro. Acesso em: 21 maio 2025. find
- TEI Consortium. 2023. Tei p5 guidelines. find
Updated
This page in Markdown → https://md.research.elamur.ai/free/en/papers/queiroz-2026-marcacao-semantica-de-entidades-nomeadas/