Abstract
This paper describes our submission to the MT4All Shared Task in unsupervised machine translation from English to Ukrainian, Kazakh and Georgian in the legal domain. In addition to the standard pipeline for unsupervised training (pretraining followed by denoising and back-translation), we used supervised training on a pseudo-parallel corpus retrieved from the provided mono-lingual corpora. Our system scored significantly higher than the baseline hybrid unsupervised MT system.- Anthology ID:
- 2022.sigul-1.10
- Volume:
- Proceedings of the 1st Annual Meeting of the ELRA/ISCA Special Interest Group on Under-Resourced Languages
- Month:
- June
- Year:
- 2022
- Address:
- Marseille, France
- Editors:
- Maite Melero, Sakriani Sakti, Claudia Soria
- Venue:
- SIGUL
- SIG:
- SIGUL
- Publisher:
- European Language Resources Association
- Note:
- Pages:
- 78–82
- Language:
- URL:
- https://aclanthology.org/2022.sigul-1.10
- DOI:
- Cite (ACL):
- Ivana Kvapilíková and Ondrej Bojar. 2022. CUNI Submission to MT4All Shared Task. In Proceedings of the 1st Annual Meeting of the ELRA/ISCA Special Interest Group on Under-Resourced Languages, pages 78–82, Marseille, France. European Language Resources Association.
- Cite (Informal):
- CUNI Submission to MT4All Shared Task (Kvapilíková & Bojar, SIGUL 2022)
- PDF:
- https://preview.aclanthology.org/nschneid-patch-4/2022.sigul-1.10.pdf