Experimenting with different machine translation models in medium-resource settings
Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review
Standard
Experimenting with different machine translation models in medium-resource settings. / Jónsson, Haukur Páll; Símonarson, Haukur Barri; Snæbjarnarson, Vésteinn; Steingrímsson, Steinþór; Loftsson, Hrafn.
Text, Speech, and Dialogue - 23rd International Conference, TSD 2020, Proceedings. ed. / Petr Sojka; Ivan Kopecek; Karel Pala; Aleš Horák. Springer, 2020. p. 95-103 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol. 12284 LNAI).Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review
Harvard
APA
Vancouver
Author
Bibtex
}
RIS
TY - GEN
T1 - Experimenting with different machine translation models in medium-resource settings
AU - Jónsson, Haukur Páll
AU - Símonarson, Haukur Barri
AU - Snæbjarnarson, Vésteinn
AU - Steingrímsson, Steinþór
AU - Loftsson, Hrafn
N1 - Funding Information: Acknowledgments. This project was funded by the Language Technology Programme for Icelandic 2019–2023. The programme, which is managed and coordinated by Almannarómur, is funded by the Icelandic Ministry of Education, Science and Culture. Publisher Copyright: © Springer Nature Switzerland AG 2020.
PY - 2020
Y1 - 2020
N2 - State-of-the-art machine translation (MT) systems rely on the availability of large parallel corpora, containing millions of sentence pairs. For the Icelandic language, the parallel corpus ParIce exists, consisting of about 3.6 million English-Icelandic sentence pairs. Given that parallel corpora for low-resource languages typically contain sentence pairs in the tens or hundreds of thousands, we classify Icelandic as a medium-resource language for MT purposes. In this paper, we present on-going experiments with different MT models, both statistical and neural, for translating English to Icelandic based on ParIce. We describe the corpus and the filtering process used for removing noisy segments, the different models used for training, and the preliminary automatic and human evaluation. We find that, while using an aggressive filtering approach, the most recent neural MT system (Transformer) performs best, obtaining the highest BLEU score and the highest fluency and adequacy scores from human evaluation for in-domain translation. Our work could be beneficial to other languages for which a similar amount of parallel data is available.
AB - State-of-the-art machine translation (MT) systems rely on the availability of large parallel corpora, containing millions of sentence pairs. For the Icelandic language, the parallel corpus ParIce exists, consisting of about 3.6 million English-Icelandic sentence pairs. Given that parallel corpora for low-resource languages typically contain sentence pairs in the tens or hundreds of thousands, we classify Icelandic as a medium-resource language for MT purposes. In this paper, we present on-going experiments with different MT models, both statistical and neural, for translating English to Icelandic based on ParIce. We describe the corpus and the filtering process used for removing noisy segments, the different models used for training, and the preliminary automatic and human evaluation. We find that, while using an aggressive filtering approach, the most recent neural MT system (Transformer) performs best, obtaining the highest BLEU score and the highest fluency and adequacy scores from human evaluation for in-domain translation. Our work could be beneficial to other languages for which a similar amount of parallel data is available.
KW - Evaluation
KW - Machine translation
KW - Parallel data
UR - http://www.scopus.com/inward/record.url?scp=85091177513&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-58323-1_10
DO - 10.1007/978-3-030-58323-1_10
M3 - Article in proceedings
AN - SCOPUS:85091177513
SN - 9783030583224
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 95
EP - 103
BT - Text, Speech, and Dialogue - 23rd International Conference, TSD 2020, Proceedings
A2 - Sojka, Petr
A2 - Kopecek, Ivan
A2 - Pala, Karel
A2 - Horák, Aleš
PB - Springer
T2 - 23rd International Conference on Text, Speech, and Dialogue, TSD 2020
Y2 - 8 September 2020 through 11 September 2020
ER -
ID: 371185063