Can Multilingual Transformers Fight the COVID-19 Infodemic?

Proceedings of Recent Advances in Natural Language Processing, pages 1432–1437 Sep 1–3, 2021. https://doi.org/10.26615/978-954-452-072-4_160 1432 Can Multilingual Transformers Fight the COVID-19 Infodemic? Lasitha Uyangodage‡, Tharindu Ranasinghe§, Hansi Hettiarachchi♥, ‡University of M¨unster, §University of Wolverhampton,♥Birmingham City University luyangod@uni-muenster.de Abstract The massive spread of false information on so- cial media has become a global risk especially in a global pandemic situation like COVID- 19. False information detection has thus be- come a surging research topic in recent months. In recent years, supervised machine learning models have been used to automatically iden- tify false information in social media. How- ever, most of these machine learning models focus only on the language they were trained on. Given the fact that social media platforms are being used in different languages, manag- ing machine learning models for each and ev- ery language separately would be chaotic. In this research, we experiment with multilingual models to identify false information in social media by using two recently released multilin- gual false information detection datasets. We show that multilingual models perform on par with the monolingual models and sometimes even better than the monolingual models to de- tect false information in social media making them more useful in real-world scenarios. 1 Introduction By June 2021, the coronavirus(COVID-19) pan- demic has affected 219 nations around the world with 176 million total cases and 3.81 million deaths. The nature of the virus caused many governments to implement lockdown in their countries. As a result, many people started spending more time at home during the pandemic and started using so- cial media more, providing an unexpected boost to engagement on these platforms (Hettiarachchi and Ranasinghe, 2020b). As a drawback of these exponential growths, so- cial media has become a conduit for spreading both rumours and deliberate misinformation, and many perpetrators are deploying sites such as Facebook, Twitter, YouTube, and WhatsApp to create a sense of panic and confusion. On the other hand, the general public can not completely ignore the in- formation seen in social media due to the fact that the Centers for Disease Control and Prevention, the World Health Organisation (WHO), numerous journals, government and other health care organi- sations are regularly posting guidance across a host of platforms. Therefore, rather than completely disregarding information seeing in social media, accurate identification of false information is cru- cial (Nguyen et al., 2020). Considering the high data generation in social media, manual approaches to filter false informa- tion require significant human efforts. Therefore an automated technique to tackle this problem will be invaluable to the community. In the light of this many shared task has been organised to tackle the false information detection in social media (Shaar et al., 2021; Nakov et al., 2021a) leading to im- plement various machine learning models which can identify false information automatically (Uyan- godage et al., 2021; Tziafas et al., 2021). How- ever, most of these approaches build language- specific models trained specifically on a particular language. Given the fact that most of the social media …

Content truncated to 500 words