Neural Tagger for Czech Language: Capturing Linguistic Phenomena in Web Corpora

Investor logo

Warning

This publication doesn't include Faculty of Arts. It includes Faculty of Informatics. Official publication website can be found on muni.cz.
Authors

NEVĚŘILOVÁ Zuzana STARÁ Marie

Year of publication 2019
Type Article in Proceedings
Conference Proceedings of the Thirteenth Workshop on Recent Advances in Slavonic Natural Language Processing, RASLAN 2019
MU Faculty or unit

Faculty of Informatics

Citation
Web https://nlp.fi.muni.cz/raslan/2019/paper10-neverilova.pdf
Keywords Czech Tagger; Multi-word Expressions; Pretrained WordEmbeddings
Description We propose a new tagger for the Czech language and particu-larly for the tagset used for annotation of corpora of the TenTen family.The tagger is based on neural networks with pretrained word embed-dings. We selected the newest Czech Web corpus of the TenTen familyas training data, but we removed sentences with phenomena that wereoften annotated incorrectly. We let the tagger to learn the annotation ofthese phenomena on its own. We also experimented with the recognitionof multi-word expressions since this information can support the correcttagging.We evaluated the tagger on 6,950 sentences (84,023 tokens) from thecstenten17corpus and achieved 75.25% accuracy when compared bytags. When compared by attributes, we achieved 91.62% accuracy; theaccuracy of POS tag prediction is 96.5%.
Related projects:

You are running an old browser version. We recommend updating your browser to its latest version.