Utilizing word embeddings for part-of-speech tagging
In this paper, we illustrate the power of distributed word representations for the part-of-speech tagging of Hungarian texts. We trained CRF models for POS-tagging that made use of features derived from the sparse coding of the word embeddings of Hungarian words as signals. We show that relying on s...
Elmentve itt :
| Szerző: | |
|---|---|
| Testületi szerző: | |
| Dokumentumtípus: | Könyv része |
| Megjelent: |
2016
|
| Sorozat: | Magyar Számítógépes Nyelvészeti Konferencia
12 |
| Kulcsszavak: | Nyelvészet - számítógép alkalmazása |
| Online Access: | http://acta.bibl.u-szeged.hu/58962 |
| Tartalmi kivonat: | In this paper, we illustrate the power of distributed word representations for the part-of-speech tagging of Hungarian texts. We trained CRF models for POS-tagging that made use of features derived from the sparse coding of the word embeddings of Hungarian words as signals. We show that relying on such a representation, it is possible to avoid the creation of language specific features for achieving reliable performance. We evaluated our models on all the subsections of the Szeged Treebank both using MSD and universal morphology tag sets. Furthermore, we also report results for inter-subcorpora experiments. |
|---|---|
| Terjedelem/Fizikai jellemzők: | 59-67 |
| ISBN: | 978-963-306-450-4 |