Partners
CrossLang
University of Ghent (LT3 team)
University of Rennes (Department of Translation and Interpretation)
Project Description
In 2020, the French Ministry of Higher Education and Research (MESR) launched the Translations and Open Science project in order to explore the opportunities offered by translation technologies to foster multilingualism in scholarly communication and thus help to remove language barriers according to Open Science principles. MESR commissioned the OPERAS Research Infrastructure to coordinate a series of preparatory studies in view of establishing a scientific translation service targeted towards a broad audience and combining relevant technologies, resources and human skills. These studies consist of defining the use cases of a translation service, collecting resources for training machine translation (MT) systems, fine-tuning MT systems using these resources, evaluating the MT output, and setting up a roadmap for running a translation service.
The fine-tuning and evaluation activities will be performed by CrossLang and its partners LT3 and University of Rennes:
- Fine-tune commercial and open-source MT engines for various scientific disciplines
- Perform automatic evaluation of MT output using reference translations
- Organise human evaluation by various profiles
Results
The MT fine-tuning and evaluation study ran in the first half of 2023. Its conclusions point towards data collection challenges, variations in MT use across profiles, and strong customisation needs. If you would like to obtain more information on the study or on MT customisation in other areas, feel free to contact us.
Related Articles
Translations and Open Science: Exploring how translation technologies can support multilingualism in scholary communication
Authors: S. Fiorini, A. Tezcan, T. Vanallemeersch, S. Szoc, K. Migdisi, L. Meeus, and L. Macken