Terminology-Constrained Translation from Monolingual Data Using GRPO

Javier Garcia Gilabert, Carlos Escolano, Xixian Liao, Maite Melero


Abstract
Terminology consistency is essential for high-quality machine translation, especially in domain-specific and professional contexts, where accurate term translation directly impacts usability. This paper presents the submission from the BSC team to the WMT25 Terminology-Aware Translation Task. We propose the use of GRPO (Group Relative Policy Optimization) to adapt translation models using monolingual data only, without requiring parallel corpora. Our reward function jointly optimizes for terminology adherence and overall translation quality, leveraging quality-estimation metrics. Experimental results demonstrate that our method consistently improves terminology translation across three language directions—English to Spanish, German, and Russian—by up to +0.36 Tₚ points across all evaluated models.
Anthology ID:
2025.wmt-1.111
Volume:
Proceedings of the Tenth Conference on Machine Translation
Month:
November
Year:
2025
Address:
Suzhou, China
Editors:
Barry Haddow, Tom Kocmi, Philipp Koehn, Christof Monz
Venue:
WMT
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
1335–1343
Language:
URL:
https://preview.aclanthology.org/ingest-emnlp/2025.wmt-1.111/
DOI:
Bibkey:
Cite (ACL):
Javier Garcia Gilabert, Carlos Escolano, Xixian Liao, and Maite Melero. 2025. Terminology-Constrained Translation from Monolingual Data Using GRPO. In Proceedings of the Tenth Conference on Machine Translation, pages 1335–1343, Suzhou, China. Association for Computational Linguistics.
Cite (Informal):
Terminology-Constrained Translation from Monolingual Data Using GRPO (Garcia Gilabert et al., WMT 2025)
Copy Citation:
PDF:
https://preview.aclanthology.org/ingest-emnlp/2025.wmt-1.111.pdf