UoS Participation in the WMT20 Translation of Biomedical Abstracts

Proceedings of the 5th Conference on Machine Translation (WMT), pages 870–874 Online, November 19–20, 2020. c⃝2020 Association for Computational Linguistics 870 UoS Participation in the WMT20 Translation of Biomedical Abstracts Felipe Soares University of Sheffield - NLP Group fs@felipesoares.net Delton de Andrade Vaz UFRGS University of Montpellier delton.vaz@gmail.comn Abstract This paper describes the machine transla- tion systems developed by the University of Sheffield (UoS) team for the biomed- ical translation shared task of WMT20. Our system is based on a Transformer model with TensorFlow Model Garden toolkit. We participated in ten translation directions for the English/Spanish, English/Portuguese, English/Russian, English/Italian, and En- glish/French language pairs. To create our training data, we concatenated several paral- lel corpora, both from in-domain and out-of- domain sources. 1 Introduction In this paper, we present the system developed by the University of Sheffield for the Biomedical Translation shared task in the Fifth Conference on Machine Translation (WMT20), which consists in translating scientific texts from the biological and health domain. Our participation in this task considered the English/Portuguese, English/Spanish, En- glish/Russian, English/Italian, and English/French language pairs with translations in both directions. For that matter, we developed a machine transla- tion (MT) system based on neural machine trans- lation (NMT), using Google’s TensorFlow Model Garden. 1 2 Related Works Previous participation in biomedical translation tasks include the works of Costa-Juss`a et al. (2016) which employed Moses Statistic Machine Translation (SMT) to perform automatic transla- tion integrated with a neural character-based re- current neural network for model re-ranking and bilingual word embeddings for out of vocabulary 1https://github.com/tensorflow/models (OOV) resolution. Given the 1000-best list of SMT translations, the RNN performs a re-scoring and selects the translation with the highest score. The OOV resolution module infers the word in the target language based on the bilingual word embedding trained on large monolingual corpora. Their reported results show that both approaches can improve BLEU scores, with the best results given by the combination of OOV resolution and RNN re-ranking. Similarly, Ive et al. (2016) also used the n-best output from Moses as input to a re- ranking model, which is based on a neural network that can handle vocabularies of arbitrary size. More recently, Tubay and Costa-Juss`a (2018) employed multi-source language translation us- ing romance languages to translate from Spanish, French, and Portuguese to English. They used data from SciELO and Medline abstracts to train a Transformer model with individual languages to English and also with all languages concatenated to English. In the last two WMT biomedical translation challenges (WMT18 and WMT19) (Neves et al., 2018; Bawden et al., 2019), the submissions that achieved the best BLEU scores for the ES/EN and PT/EN, in both directions (Soares and Becker, 2018; Tubay and Costa-Juss`a, 2018; Car- rino et al., 2019; Saunders et al., 2019; Soares and Krallinger, 2019), used the Transformer architec- ture with enhancements such as handling of termi- nology during tokenization (Carrino et al., 2019), multi-domain inference (Saunders et al., 2019) and exploitation of additional linguistic resources (Soares and Becker, 2018; …

Content truncated to 500 words