Elamur Research⌕
☰
home / papers / 2022 / Universal Grammatical Dependencies for Portuguese with CINTIL Data, LX Processing and CLARIN support

Universal Grammatical Dependencies for Portuguese with CINTIL Data, LX Processing and CLARIN support

Summary

Portuguese UD resources were comparatively small and scattered. The NLX group at the University of Lisbon addresses this with one coherent, openly licensed package: annotated data, trained tools, web services and expert helpdesk support. All data come from the manually annotated CINTIL corpus. CINTIL-UPos (about 675K tokens of news and fiction, without speech transcriptions) was created by automatically mapping the CINTIL tagset onto UPOS. CINTIL-UDep Treebank (37,780 sentences) merges two expert-validated dependency banks and converts their dependencies to UD with hand-written tregex rules. The first bank was built by manually disambiguating parses from the HPSG grammar LXGram, with double-blind annotation and adjudication. The second was annotated by hand in WebAnno. The authors trained two tools on this data. LX-UTagger, a fine-tuned BERTimbau model, reaches 99.01% mean accuracy in 10-fold cross-validation. LX-UDParser is the companion UD parser. The treebank holds over three times more text than previously published Portuguese UD treebanks, which gives considerably more training material. Everything is available through PORTULAN CLARIN and GitHub.

Key points

  • CINTIL-UDep Treebank: 37,780 sentences and 473,929 tokens. About 93% comes from the grammar-supported CINTIL DependencyBank, and about 7% from the hand-drawn CINTIL DependencyBank Premium, which covers phenomena LXGram cannot parse.
  • CINTIL-UPos: roughly 675K tokens with a UPOS layer derived by tagset mapping. Every CINTIL tag except UM maps to exactly one UPOS tag, so exhaustive manual checking was unnecessary.
  • LX-UTagger: built on bert-base-portuguese-cased (BERTimbau) using Hugging Face BertForTokenClassification, with 20 output labels. Compound tags for contractions (e.g., do = ADP+DET) are split into separate tokens and tags at tagging time. The released model comes from fold 3, which scored 99.06%. For context, the best published Portuguese UPOS score is 98.04% (UDPipe), though it was measured on different data.
  • LX-UDParser: described as top-performing. Its LAS results fall in the part of the paper not included in the provided excerpt.
  • Access and support: both tools run as local stand-alone software and as web services. Helpdesk support is provided by CLARIN's knowledge centre dedicated to Portuguese language science and technology.

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: Linguistics & languages

Topics: Treebanks

Cite

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

@inproceedings{branco2022universal,
  title = {{Universal Grammatical Dependencies for Portuguese with CINTIL Data, LX Processing and CLARIN support}},
  author = {Branco, António and Silva, João Ricardo and Gomes, Luís and Rodrigues, João António},
  booktitle = {Proceedings of the Thirteenth Language Resources and Evaluation Conference},
  year = {2022},
  doi = {10.63317/2n5gy7raxukx},
  url = {https://doi.org/10.63317/2n5gy7raxukx}
}

References

38 references, 7 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Barreto, F., Branco, A., Mendes, A., Bacelar do Nascimento, M. F., and Silva, J. R. (2005). CINTIL corpus internacional do Português: Convenções de etiquetação. Barreto, F., Branco, A., Ferreira, E., Mendes, A., Bacelar do Nascimento, M. F., Nunes, F., and Silva, J. R. (2006). Open resources an… find
  2. Branco, A. and Costa, F. (2008). A computational grammar for deep linguistic processing of portuguese: Lxgram. find
  3. Branco, A. and Henriques, T. (2003). Aspects of verbal inflection and lemmatization: Generalizations and algorithms. In In Proceedings of XVIII Annual Meeting of the Portuguese Association of Linguistics (APL), pages 201–210. find
  4. Branco, A. and Nunes, F. (2012). Verb analysis in a highly inflective language with an MFF algorithm. In In Proceedings of the 11th International Conference on the Computational Processing of Portuguese (PROPOR), number 7243 in Lecture Notes in Artificial Intelligence, pages 1–11. find
  5. Springer. Branco, A. and Silva, J. (2004). Evaluating solutions for the rapid development of state-of-the-art POS taggers for Portuguese. In Proceedings of the 4th International Conference on Language Resources and Evaluation (LREC 2004), pages 507–510. find
  6. Branco, A. and Silva, J. (2006). A suite of shallow processing tools for Portuguese: LX-Suite. In In Proceedings of the 11th European Chapter of the Association for Computational Linguistics (EACL), pages 179–182. Branco, A., Costa, F., Silva, J., Silveira, S., Castro, S., Avelãs, M., Pinto, C., an… find
  7. Branco, A., Rodrigues, J., Silva, J., Costa, F., and Vaz, R. (2014a). Assessing automatic text classification for interactive language learning. In In Proceedings of the IEEE International Conference on Information Society (iSociety), pages 72–80. find
  8. Branco, A., Rodrigues, J., Costa, F., Silva, J. R., and Vaz, R. (2014b). Rolling out text categorization for language learning assessment supported by language technology. In Lecture Notes in Artificial Intelligence, 8775, pages 256–261. Branco, A., Silva, J., Gonçalves, P., Costa, F., Silveira, S.… find
  9. Branco, A., Silva, J., Querido, A., and de Carvalho, R. (2015). CINTIL DependencyBank PREMIUM handbook: Design options for the representation of grammatical dependencies. Technical report, University of Lisbon, Faculty of Sciences, Department of Informatics. TR-2015-05, DOI:10451/20226. find
  10. Branco, A., Mendes, A., Quaresma, P., Gomes, L., Silva, J., and Teixeira, A. (2020). Infrastructure for the science and technology of language PORTULAN CLARIN. In Proceedings, 1st International Workshop on Language Technology Platforms (IWLTP 2020), pages 1–7. European Language Resources Association… find
  11. Chen, D. and Manning, C. (2014). A fast and accurate dependency parser using neural networks. In Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 740–750, Doha, Qatar, October. Association for Computational Linguistics. find our card
  12. Choi, J. D. and McCallum, A. (2013). Transition-based dependency parsing with selectional branching. In Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1052–1062. find
  13. Copestake, A., Flickinger, D., Pollard, C., and Sag, I. A. (2005). Minimal recursion semantics: An introduction. Research on language and computation, 3(2):281–332. find
  14. Costa, F. and Branco, A. (2010). Lxgram: A deep linguistic processing grammar for Portuguese. In International Conference on Computational Processing of the Portuguese Language, pages 86–89. find
  15. Springer. Costa, F. and Branco, A. (2012). Aspectual type and temporal relation classification. In In Proceedings of the 13th Conference of the European Chapter of the Association for Computational Linguistics (EACL), pages 266–275. find
  16. Cruz, A. F., Rocha, G., and Cardoso, H. L. (2018). Exploring Spanish corpora for Portuguese coreference resolution. In Fifth International Conference on Social Networks Analysis, Management and Security (SNAMS), pages 290–295, Oct. de Carvalho, R., Querido, A., Campos, M., Pereira, R., Silva, J., an… find
  17. Devlin, J., Chang, M.-W., Lee, K., and Toutanova, K. (2019). Bert: Pre-training of deep bidirectional transformers for language understanding. find
  18. Dozat, T., Qi, P., and Manning, C. D. (2017). Stanford’s graph-based neural dependency parser at the CoNLL 2017 shared task. In Proceedings of the CoNLL 2017 Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies, pages 20–30, Vancouver, Canada, August. Association for Computation… find our card
  19. Freitas, C., Rocha, P., and Bick, E. (2008). Floresta sintá(c)tica: bigger, thicker and easier. In Proceedings of International Conference on Computational Processing of the Portuguese Language (PROPOR 2008), pages 216–219. Springer. Gomes, L., Apolónia, F., Branco, R., Silva, J. R., and Branco, A… find
  20. Grilo, S., Bolrinha, M., Silva, J., Vaz, R., and Branco, A. (2020). The BDCamões collection of Portuguese literary documents: a research resource for digital humanities and language technology. In Proceedings of the 12th Language Resources and Evaluation Conference (LREC 2020), pages 849–854. find
  21. Honnibal, M. and Johnson, M. (2015). An improved non-monotonic transition system for dependency parsing. In Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, pages 1373–1378, Lisbon, Portugal, September. Association for Computational Linguistics. Klimaszewski, M… find our card
  22. Kondratyuk, D. and Straka, M. (2019). 75 languages, 1 model: Parsing Universal Dependencies universally. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pages 2779–2795… find our card
  23. Levy, R. and Andrew, G. (2006). Tregex and Tsurgeon: Tools for querying and manipulating tree data structures. In Proceedings of the 5th International Conference on Language Resources and Evaluation (LREC 2006), pages 2231–2234. Miranda, N., Raminhos, R., Seabra, P., Sequeira, J., Gonçalves, T., an… find
  24. European Language Resources Association. Oepen, S. and Flickinger, D. (1998). Towards systematic grammar profiling. test suite technology 10 years after. Journal of Computer Speech & Language, 12(4):411–436. find
  25. Pollard, C. and Sag, I. A. (1994). Head-Driven Phrase Structure Grammar. The University of Chicago Press, Chicago. find
  26. Project Jupyter, Matthias Bussonnier, Jessica Forde, Jeremy Freeman, Brian Granger, Tim Head, Chris Holdgraf, Kyle Kelley, Gladys Nalvarte, Andrew Osheroff, Pacer, M., Yuvi Panda, Fernando Perez, Benjamin Ragan Kelley, and Carol Willing. (2018). Binder 2.0 - Reproducible, interactive, sharable envir… find
  27. Rademaker, A., Chalub, F., Real, L., Freitas, C., Bick, E., and de Paiva, V. (2017). Universal Dependencies for Portuguese. In Proceedings of the Fourth International Conference on Dependency Linguistics (Depling 2017), pages 197–206, Pisa,Italy, September. Linköping University Electronic Press. Ro… find
  28. (2016). LX-DSemVectors: Distributional semantics models for the Portuguese language. In In Proceedings of the 12th International Conference on the Computational Processing of Portuguese (PRO-POR’16), pages 259–270. find
  29. Rodrigues, J., Costa, F., Silva, J., and Branco, A. (2020). Automatic syllabification of Portuguese. Revista da Associação Portuguesa de Linguı́stica, 1. find
  30. Santos, R., Silva, J., Branco, A., and Xiong, D. (2019). The direct path may not be the best: Portuguese-Chinese neural machine translation. In In Proceedings of the 19th Portuguese Conference on Artificial Intelligence (EPIA), pages 757–768. find
  31. Silva, J., Branco, A., Castro, S., and Reis, R. (2010). Out-of-the-box robust parsing of Portuguese. In Lecture Notes in Artificial Intelligence, 6001, pages 75–85. find
  32. Silveira, S. and Branco, A. (2012a). Combining a double clustering approach with sentence simplification to produce highly informative multi-document summaries. In Proceedings, IEEE IRI2012-International Conference on Information Reuse and Integration, pages 482–489. find
  33. Silveira, S. and Branco, A. (2012b). Enhancing multi-document summaries with sentence simplification. In Proceedings, ICAI2012-14th International Conference on Artificial Intelligence, pages 742–748. find
  34. Souza, F., Nogueira, R., and Lotufo, R. (2020). BERTimbau: pretrained BERT models for Brazilian Portuguese. In 9th Brazilian Conference on Intelligent Systems, BRACIS, Rio Grande do Sul, Brazil, October 20-23 (to appear). find
  35. Straka, M. (2018). UDPipe 2.0 prototype at CoNLL 2018 UD shared task. In Proceedings of the CoNLL 2018 Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies, pages 197–207, Brussels, Belgium, October. Association for Computational Linguistics. Veiga, A., Candeias, S., and Perdiga… find our card
  36. Wolf, T., Debut, L., Sanh, V., Chaumond, J., Delangue, C., Moi, A., Cistac, P., Rault, T., Louf, R., Funtow-icz, M., Davison, J., Shleifer, S., von Platen, P., Ma, C., Jernite, Y., Plu, J., Xu, C., Scao, T. L., Gugger, S., Drame, M., Lhoest, Q., and Rush, A. M. (2020). Huggingface’s transformers: St… find our card
  37. CoNLL 2017 shared task: Multilingual parsing from raw text to Universal Dependencies. In Proceedings of the CoNLL 2017 Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies, pages 1–19, Vancouver, Canada, August. Association for Computational Linguistics. find our card
  38. Zilio, L., Wilkens, R., and Fairon, C. (2018). Pass-Port: A dependency parsing model for Portuguese. In Proceedings of the International Conference on Computational Processing of the Portuguese Language (PROPOR), pages 479–489. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/papers/branco-2022-universal-grammatical-dependencies-portuguese-cintil/