Open AccessDetection of Misinformation about COVID-19 in Brazilian Portuguese WhatsApp Messages Using Deep Learning
Author(s) -
Aline Maria Araújo Martins,
Lucas Manoel da Silva Cabral,
Pedro Jorge Chaves Mourão,
Ângelo Brayner,
Javam C. Machado
Publication year - 2021
Language(s) - English
Resource type - Conference proceedings
DOI - 10.5753/sbbd.2021.17868
Subject(s) - misinformation , portuguese , computer science , pooling , covid-19 , artificial intelligence , social media , pandemic , internet privacy , world wide web , computer security , medicine , philosophy , linguistics , disease , pathology , infectious disease (medical specialty)
During the COVID-19 pandemic, the misinformation problem arose once again through social networks, like a harmful health advice and false solutions epidemic. In Brazil, as well as in many developing countries, one of the primary sources of misinformation is the messaging application WhatsApp. Thus, the automatic misinformation detection (MID) about COVID-19 in Brazilian Portuguese WhatsApp messages becomes a crucial challenge. Still, due to WhatsApp's private messaging nature, there are still few methods of misinformation detection developed specifically for the WhatsApp platform. In this paper, we propose a new approach, called MIDeepBR, based on BiLSTM neural networks, pooling operations and attention mechanism, which is able to automatically detect misinformation in Brazilian Portuguese WhatsApp messages. Experimental results evidence the suitability of the proposed approach to automatic misinformation detection. Our best results achieved an F1 score of 0.834, while in previous works, the best results achieved an F1 score of 0.778. Thus, MIDeepBR outperforms the previous works.