Current events

The UOC is partnering with Vall d'Hebron University Hospital in the translation of scientific articles on COVID-19

  hospital

Photo: Javier Matheu / unsplash.com

The researchers have developed a neural machine translation system specifically for the medical field

Researchers from Mexico and Switzerland are taking part in the initiative to translate the abstracts of more than 15,000 coronavirus-related scientific articles

The scientific community is devoting significant effort to combating the new coronavirus and finding the vaccine that it is hoped will bring the pandemic to an end. As an illustration of the intensity of this effort, the World Health Organization's database alone contains about  40,000 studies on the subject of COVID-19. To cover the high volume of scientific literature on the disease, a team of researchers from the Universitat Oberta de Catalunya (UOC) has developed technology for translating the abstracts of 15,000 scientific articles into Spanish. The goal is to facilitate access to this information for the medical staff at Vall d'Hebron University Hospital in Barcelona, among other professionals.

 

Training the translation system

"Our goal is to make relevant information about COVID-19 available to medical professionals, as not all of them have a sufficient command of English. Our technology makes it easier for them to access this information," said Antoni Oliver, a member of the Linguistic Applications Inter-University Research Group (GRIAL) at the UOC's Faculty of Arts and Humanities. The technology will translate abstracts and scientific articles about COVID-19 from English to Spanish. The University's collaboration in this project centres on training and use of a  neural machine translation system that has been specifically developed for medical translation and adapted to the subject areas related with the coronavirus. In addition, the UOC team is providing technological support in setting up and operating an online computer-assisted translation system.

The time that the system takes to translate depends on the computer on which the translation engine is run. It can range between 100 and 10,000 words per second, Oliver confirmed, although the important factor is the total time of human involvement. "The entire process (machine translation plus post-editing) can be four or five times faster than human translation alone, with very similar quality levels," the UOC researcher explained.

 

Detecting relevant articles

For this project, other research teams taking part in this project use artificial intelligence techniques to detect relevant scientific articles about COVID-19. "The first step was a manual classification of a large group of documents, which the system then used to learn to classify the documents," Oliver explained. Their system retrieves scientific articles from the Science Direct compilation developed by the Dutch scientific publisher Elsevier and from repositories such as LitCOVID, an open resources portal run by the US National Library of Medicine, which provides access to one of the world's largest collections of research papers on the new disease, currently listing more than 28,000 publications.

 

International cooperation

This project is a direct product of the pandemic. "During the early days of lockdown, a number of contacts were made to find ways to help using translation technologies," said Oliver. The intention is to continue with this collaboration for as long as COVID-19 remains a hot research topic.

With international collaborations, the initiative goes beyond the local sphere and one of its goals is to help create a repository of scientific articles on the disease at the National Autonomous University of Mexico (UNAM), together with a similar platform hosted at Vall d'Hebron University Hospital. In addition to the UOC and the hospital, other participants include the Agency for Health Quality and Assessment of Catalonia (AQuAS), the UNAM's Center for Genome Sciences, two researchers (one from the National Institute of Respiratory Diseases in Mexico and another from the Dalle Molle Institute for Artificial Intelligence Research in Switzerland), and a doctoral student from the Universitat Politcnica de Catalunya (UPC).

Oliver highlighted that it was not an institutionally managed initiative and that "the important work is all done by volunteers". In addition, the UOC encourages students carrying out their internships for the  Bachelor's Degree in Translation, Interpreting and Applied Languages and the  University Master's Degree in Translation and Technologies to participate.

 

Related article

Antoni Oliver (2020). "MTUOC: easy and free integration of NMT systems in professional translation environments". Proceedings of the 22nd Annual Conference of the European Association for Machine Translation: http://hdl.handle.net/10609/119706

 

UOC R&I

The UOC's research and innovation (R&I) contribute to solving the challenges facing the global societies of the 21st century by studying ICTs' interactions with human activity, with a specific focus on e-learning and e-health. Over 400 researchers and 48 research groups work among the University's seven faculties and three research centres: the Internet Interdisciplinary Institute (IN3), the eLearn Center (eLC) and the eHealth Center (eHC).

The United Nations 2030 Agenda's Sustainable Development Goals and open knowledge provide strategic pillars on which the UOC's teaching, research and innovation are built. More information: research.uoc.edu.

#UOCexperts

Photograph of Antoni Oliver Gonzlez

Antoni Oliver Gonzlez

Expert in: Automatic translation; assisted translation; terminology extraction; machine learning.

Knowledge area: Computational linguistics.

View file