Constructing Parallel Corpora from COVID-19 News using MediSys Metadata

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 1068–1072 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 1068 Constructing Parallel Corpora from COVID-19 News using MediSys Metadata Dimitris Roussis, Vassilis Papavassiliou, Sokratis Sofianopoulos, Prokopis Prokopidis, Stelios Piperidis Institute for Language and Speech Processing (ILSP) Athena Research and Innovation Center, Athens, Greece {dimitris.roussis, vpapa, s_sofian, prokopis, spip}@athenarc.gr Abstract This paper presents a collection of parallel corpora generated by exploiting the COVID-19 related dataset of metadata created with the Europe Media Monitor (EMM) / Medical Information System (MediSys) processing chain of news articles. We describe how we constructed comparable monolingual corpora of news articles related to the current pandemic and used them to mine about 11.2 million segment alignments in 26 EN-X language pairs, covering most official EU languages plus Albanian, Arabic, Icelandic, Macedonian, and Norwegian. Subsets of this collection have been used in shared tasks (e.g. Multilingual Semantic Search, Machine Translation) aimed at accelerating the creation of resources and tools needed to facilitate access to information in the COVID-19 emergency situation. Keywords: COVID-19, Parallel Corpus, Machine Translation, Emergency Situations 1. Introduction In December 2019, the first cases of the COVID-19 disease caused by the SARS-CoV-2 virus were identified in Wuhan, Hubei, China. On 11 March 2020, the WHO (World Health Organization) declared the novel coronavirus outbreak a pandemic and it was only a matter of time before every country in the world would be affected by it to a smaller or larger extent. The COVID-19 pandemic, however, unfolded as something more than a disease; it became the catalyst of -quite simultaneous- global changes in everyday life and public discourse with a tremendous impact on financial stability, healthcare systems, national policies, working conditions, individual freedoms, travel and movement. Due to the pandemic, we became increasingly dependent on mass media, social media, news websites, online communication services and in general, anything that involved internet and the “digital”. Thus, an abundance of COVID-19 related textual information was generated in many languages and platforms, and motivated researchers to gather and structure data in order to facilitate the use of statistical methods and Artificial Intelligence (AI) for a plethora of applications (Shuja et al., 2020). The emergence of multidisciplinary initiatives, such as the creation of the CORD-19 (Covid-19 Open Research Dataset) dataset1 and the shared tasks built around it, illustrated how data scientists, biomedical experts and policy makers can connect and cooperate in the battle against COVID-19 (Wang et al., 2020). In emergency situations, Machine Translation (MT) can be embedded in a rapid-response infrastructure and accelerate the speed in which relief can be provided (Lewis et al., 2011). The COVID-19 pandemic is such an emergency situation, albeit with different characteristics compared to other crises due to its global character and extended duration. Access to guidelines, news, announcements, high-quality articles etc. is vital to a wide range of stakeholders, from healthcare professionals, decision- makers and researchers to vulnerable groups, such 1https://www.semanticscholar.org/cord19 2https://www.who.int/health-topics/infodemic 3http://eval.covid19-mlia.eu/ as immigrants, marginalized communities …

Content truncated to 500 words