First experiments and results in English-Hungarian neural machine translation

Neural machine translation (NMT) has emerged recently as a promising alternative of standard rule-based or phrase-based statistical approaches especially for languages which have so far been considered challenging for the two paradigms. Since Hungarian has long been one of these challenging language...

Teljes leírás

Elmentve itt :
Bibliográfiai részletek
Szerzők: Tihanyi László
Oravecz Csaba
Testületi szerző: Magyar Számítógépes Nyelvészeti Konferencia (13.) (2017) (Szeged)
Dokumentumtípus: Könyv része
Megjelent: 2017
Sorozat:Magyar Számítógépes Nyelvészeti Konferencia 13
Kulcsszavak:Nyelvészet - számítógép alkalmazása
Online Access:http://acta.bibl.u-szeged.hu/59016
Leíró adatok
Tartalmi kivonat:Neural machine translation (NMT) has emerged recently as a promising alternative of standard rule-based or phrase-based statistical approaches especially for languages which have so far been considered challenging for the two paradigms. Since Hungarian has long been one of these challenging languages, it is a natural candidate for neural machine translation to explore whether this approach can bring some improvement in a task which translation models have so far been unable to cope with. The paper presents our first results of applying neural models to English to Hungarian translation and shows that with the right configuration and data preparation, publicly available NMT implementations can significantly outperform the previous state-of-the-art systems on standard benchmarks.
Terjedelem/Fizikai jellemzők:275-286
ISBN:978-963-306-518-1