Abstrakt
Background: Biomedical language translation requires multi-lingual fluency as well as relevant domain knowledge. Such requirements make it challenging to train qualified translators and costly to generate high-quality translations.
Machine translation represents an effective alternative, but accurate machine translation requires large amounts of in-domain data. While such datasets are abundant in general domains, they are less accessible in the biomedical domain.
Chinese and English are two of the most widely spoken languages, yet to our knowledge, a parallel corpus does not exist for this language pair in the biomedical domain. Description: We developed an effective pipeline to acquire and process an English-Chinese parallel corpus from the New England Journal of Medicine (NEJM).
This corpus consists of about 100,000 sentence pairs and 3,000,000 tokens on each side. We showed that training on out-of-domain data and fine-tuning with as few as 4000 NEJM sentence pairs improve translation quality by 25.3 (13.4) BLEU for enRIGHTWARDS ARROW zh (zhRIGHTWARDS ARROW en) directions.
Translation quality continues to improve at a slower pace on larger in-domain data subsets, with a total increase of 33.0 (24.3) BLEU for enRIGHTWARDS ARROW zh (zhRIGHTWARDS ARROW en) directions on the full dataset. Conclusions: The code and data are available at https://github.com/boxiangliu/ParaMed.