POS Tagging for Historical Texts with Sparse Training Data

What it is about

The paper tags historical German (1400–1770) in two steps. First, the Norma tool normalises the spelling to modern forms. Second, a tagger trained on modern German tags the normalised text. The normaliser needs only a few hundred hand-normalised tokens per text. Before normalisation, the text matches the gold normalisation for 23% and 39% of tokens in the two Anselm texts, and for 72% to 83% in the GerManC texts. With 100 training tokens, normalisation reaches 69% on the Anselm texts. With 250 tokens, tagging accuracy on a 15th-century manuscript rises from 28.65% to 74.89%.

Ideas

I1. en/src/morph.rs: learn a small rewrite table from about 250 hand-normalised tokens of an old English lesson, and apply it before en/src/ptag.rs. Measure tagging accuracy on the held-out part of tools/gutenberg lessons with and without the table.
I2. Learning curve: report normalisation accuracy at 100, 250 and 1000 training tokens, as the paper does, so the hand work for a new historical book is known in advance.
I3. Baseline first: for each historical test set, report the share of tokens that are already identical to their gold normalisation. The paper's 23% to 83% range shows that this share varies a lot between texts.

For Mova

The chapters are 06. Spelling rules: inflection, derivation, compounds and national standards and 09. Early Modern and archaic forms. The code is en/src/morph.rs and en/src/ptag.rs. The paper is about German and NC, so it gives no English figures. Measure first the identity baseline on the chosen lesson, twice.

Topics: Part-of-speech tagging

Cite

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

@inproceedings{bollmann2013pos,
  title = {{POS Tagging for Historical Texts with Sparse Training Data}},
  author = {Bollmann, Marcel},
  booktitle = {Proceedings of the 7th Linguistic Annotation Workshop and Interoperability with Discourse},
  year = {2013},
  url = {https://aclanthology.org/W13-2302/}
}

References

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

Alistair Baron and Paul Rayson. 2008. VARD 2: A tool for dealing with spelling variation in historical corpora. In Proceedings of the Postgraduate Conference in Corpus Linguistics. find our card
Marcel Bollmann, Stefanie Dipper, Julia Krasselt, and Florian Petran. 2012. Manual and semi-automatic normalization of historical spelling – Case studies from Early New High German. In Proceedings of KONVENS 2012 (LThist 2012 workshop), pages 342–350, Vienna, Austria. find
Marcel Bollmann. 2012. (Semi-)automatic normalization of historical texts using distance measures and the Norma tool. In Proceedings of the Second Workshop on Annotation of Corpora for Research in the Humanities (ACRH-2), Lisbon, Portugal. find
Sabine Brants, Stefanie Dipper, Silvia Hansen, Wolfgang Lezius, and George Smith. 2002. The TIGER treebank. In Erhard Hinrichs and Kiril Simov, editors, Proceedings of the First Workshop on Treebanks and Linguistic Theories (TLT 2002), Sozopol, Bulgaria. find
Thorsten Brants. 2000. TnT — a statistical part-of-speech tagger. In Proceedings of ANLP 2000, pages 224–231, Seattle, USA. find our card
Stefanie Dipper and Simone Schultz-Balluff. 2013. The Anselm corpus: Methods and perspectives of a parallel aligned corpus. In Proceedings of the NODALIDA Workshop on Computational Historical Linguistics. find
Stefanie Dipper. 2010. POS-tagging of historical language data: First experiments. In Proceedings of KONVENS 2010, pages 117–121, Saarbrücken, Germany. find
Iris Hendrickx and Rita Marquilhas. 2011. From old texts to modern spellings: an experiment in automatic normalisation. JLCL, 26(2):65–76. find
Bryan Jurish. 2010. More than words: using token context to improve canonicalization of historical German. Journal for Language Technology and Computational Linguistics, 25(1):23–39. find
Paul Rayson, Dawn Archer, and Nicholas Smith. 2005. VARD versus Word. A comparison of the UCREL variant detector and modern spell checkers on english historical corpora. In Proceedings of Corpus Linguistics 2005, Birmingham, UK. find
Paul Rayson, Dawn Archer, Alistair Baron, Jonathan Culpeper, and Nicholas Smith. 2007. Tagging the bard: Evaluating the accuracy of a modern POS tagger on Early Modern English corpora. In Proceedings of Corpus Linguistics 2007, University of Birmingham, UK. find
Cristina Sánchez-Marco, Gemma Boleda, Josep Maria Fontana, and Judith Domingo. 2010. Annotation and representation of a diachronic corpus of Spanish. In Proceedings of the Seventh Conference on International Language Resources and Evaluation, pages 2713–2718. find
Silke Scheible, Richard J. Whitt, Martin Durrell, and Paul Bennett. 2011a. Evaluating an ‘off-the-shelf’ POS-tagger on Early Modern German text. In Proceedings of the ACL-HLT 2011 Workshop on Language Technology for Cultural Heritage, Social Sciences, and Humanities (LaTeCH 2011), pages 19– 23, Port… find
Silke Scheible, Richard J. Whitt, Martin Durrell, and Paul Bennett. 2011b. A gold standard corpus of Early Modern German. In Proceedings of the ACL-HLT 2011 Linguistic Annotation Workshop (LAW V), pages 124–128, Portland, Oregon, USA. find our card
Anne Schiller, Simone Teufel, Christine Stöckert, and Christine Thielen. 1999. Guidelines für das Tagging deutscher Textcorpora mit STTS. Technical report. find
Helmut Schmid and Florian Laws. 2008. Estimation of conditional probabilities with decision trees and an application to fine-grained POS tagging. In Proceedings of COLING ’08, Manchester, Great Britain. find
Heike Telljohann, Erhard Hinrichs, and Sandra Kübler. 2004. The Tüba-D/Z Treebank: Annotating German with a Context-Free Backbone. In Proceedings of the Fourth International Conference on Language Resources and Evaluation (LREC 2004), pages 2229–2235, Lisbon, Portugal. find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/aclbulk-2013-bollmann-pos-tagging-historical-texts-sparse/