A Multilingual Neural Machine Translation Model for Biomedical Data

TL;DR

This paper introduces a multilingual neural machine translation model for biomedical data, capable of translating five languages into English, trained on extensive generic and biomedical datasets, and achieving near state-of-the-art performance.

Contribution

The paper presents a new multilingual NMT model for biomedical translation across five languages, with improved performance and a dedicated COVID-19 biomedical test set.

Findings

Performs near state-of-the-art on news and biomedical datasets

Outperforms existing publicly available models

Provides a new Korean-English biomedical test set

Abstract

We release a multilingual neural machine translation model, which can be used to translate text in the biomedical domain. The model can translate from 5 languages (French, German, Italian, Korean and Spanish) into English. It is trained with large amounts of generic and biomedical data, using domain tags. Our benchmarks show that it performs near state-of-the-art both on news (generic domain) and biomedical test sets, and that it outperforms the existing publicly released models. We believe that this release will help the large-scale multilingual analysis of the digital content of the COVID-19 crisis and of its effects on society, economy, and healthcare policies. We also release a test set of biomedical text for Korean-English. It consists of 758 sentences from official guidelines and recent papers, all about COVID-19.