MAchine Translation for Open Science

Towards the translation of whole scientific documents

The ANR MaTOS project (Machine Translation for Open Science), led by ISIR (MLIA), Inria (ALMAnaCH), UniversitĂ© Paris-CitĂ© (CLILLAC-ARP) and CNRS (INIST), aims to develop new methods for machine translation (MT) of whole scientific documents, as well as automatic metrics to assess the quality of translations.

image info

We will mainly focus on the translation of scientific articles between French and English, where linguistic resources can be exploited to obtain more reliable translations, both for the purpose of aiding publication and for the purposes of reading and text mining.

The project aims to contribute in several ways to automatic scientific article processing: (a) by developing new open resources for specialised MT; (b) by improving the description of coherence markers for scientific articles through the study of terminological variation; (c) by studying new multilingual processing methods for these documents; (d) by proposing metrics dedicated to measuring progress in scientific document MT. The final result will, through improved translation, facilitate the flow and dissemination of scientific knowledge and understanding.