Lemmatisation as a Tagging Task

What it is about

The authors recast lemmatization as a tagging task. A word-to-lemma transformation is encoded as one label, and the set of labels is inferred from the training data of each language. Any supervised tagger can then learn the labels, and it can use the context of the word and its predicted part of speech. The method is tested on eight languages with different morphological complexity, among them Czech, English, Estonian, Hungarian, Polish, Romanian, Serbian and Slovene. Using the context and the predicted tag gives a relative error reduction over the baseline from 28.1% for Hungarian to 66.7% for English. On unseen words the accuracy is 85% or higher for all languages except Estonian, which is 78.5%.

Ideas

• I1. en/src/morph.rs: encode each suffix-based lemma change of the dictionary as one label, and let the averaged perceptron of en/src/ptag.rs learn the labels with the tag and the neighbouring words as features. Measure lemma accuracy on EWT test, and the accuracy on forms that the dictionary does not contain.
• I2. Keep the paper's split of results: report accuracy with the predicted tag and with the gold tag, so the gain from the context is not hidden by the tagger error.
• I3. Use the label set as a check on the dictionary: count the labels that occur only once in data/forms.tsv, and list them for review, since a rare label is often a dictionary error.

For Mova

The code is en/src/morph.rs and en/src/ptag.rs, with the commands en morph-eval and en tag-eval. The chapter is 06. Spelling rules: inflection, derivation, compounds and national standards for the change rules and 13. Verbs for verbs. The paper's languages are mostly not English, and English is only one of eight, so use its method and measure on EWT test only. Two-fold check (en folds) before any claim.

Topics: Part-of-speech tagging

Cite

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

@inproceedings{gesmundo2012lemmatisation,
  title = {{Lemmatisation as a Tagging Task}},
  author = {Gesmundo, Andrea and Samardžić, Tanja},
  booktitle = {Proceedings of the 50th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)},
  year = {2012},
  url = {https://aclanthology.org/P12-2072/}
}

References

10 references, 2 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Grzegorz Chrupała. 2006. Simple data-driven context-sensitive lemmatization. In Proceedings of the Sociedad Española para el Procesamiento del Lenguaje Natural, volume 37, page 121131, Zaragoza, Spain. find
2. Tomaž Erjavec and Sašo Džeroski. 2004. Machine learning of morphosyntactic structure: lemmatizing unknown Slovene words. Applied Artificial Intelligence, 18:17–41. find
3. Tomaž Erjavec. 2010. Multext-east version 4: Multilingual morphosyntactic specifications, lexicons and corpora. In Nicoletta Calzolari (Conference Chair), Khalid Choukri, Bente Maegaard, Joseph Mariani, Jan Odijk, Stelios Piperidis, Mike Rosner, and Daniel Tapias, editors, Proceedings of the Sevent… find
4. Andrea Gesmundo. 2011. Bidirectional sequence classification for tagging tasks with guided learning. In Proceedings of TALN 2011, Montpellier, France. find
5. Dan Gusfield. 1997. Algorithms on Strings, Trees, and Sequences - Computer Science and Computational Biology. Cambridge University Press. find
6. Jan Hajič. 2000. Morphological tagging: data vs. dictionaries. In Proceedings of the 1st North American chapter of the Association for Computational Linguistics conference, pages 94–101, Seattle, Washington. Association for Computational Linguistics. find
7. Bart Jongejan and Hercules Dalianis. 2009. Automatic training of lemmatization rules that handle morphological changes in pre-, in- and suffixes alike. In Proceedings of the Joint Conference of the 47th Annual Meeting of the ACL and the 4th International Joint Conference on Natural Language Processi… find our card
8. Matjaž Juršič, Igor Mozetič, Tomaž Erjavec, and Nada Lavrač. 2010. LemmaGen: Multilingual lemmatisation with induced ripple-down rules. Journal of Universal Computer Science, 16(9):1190–1214. find
9. Libin Shen, Giorgio Satta, and Aravind Joshi. 2007. Guided learning for bidirectional sequence classification. In Proceedings of the 45th Annual Meeting of the Association of Computational Linguistics, pages 760– 767, Prague, Czech Republic. Association for Computational Linguistics. find
10. Kristina Toutanova and Colin Cherry. 2009. A global model for joint lemmatization and part-of-speech prediction. In Proceedings of the 47th Annual Meeting of the ACL and the 4th IJCNLP of the AFNLP, page 486494, Suntec, Singapore. Association for Computational Linguistics. find our card

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/aclbulk-2012-gesmundo-lemmatisation-as-tagging-task/