Covid-19 Sentiment Analysis using Bidirectional Encoder Representations from Transformers

Abstract

Corona coronavirus (COVID-19) is a progressive pandemic that is being recognized worldwide. However, spreading false news on social media platforms such as Twitter creates unnecessary concern about the disease. The motto of this study analyzes tweets by Indian netizens during the closure of COVID-19. The data included tweets collected between 23 March 2020 and 15 July 2020 and the text was written as fear, sadness, anger and happiness. Data analysis was performed by the Bidirectional Encoder Representations from Transformers (BERT) model, which is a new in-depth study model for text analysis and performance and was compared with three other models such as logistic regression (LR), vector support (SVM). The accuracy of all the words was calculated separately. The BERT model produced 86% accuracy. Our findings point to a significant increase in keywords and related names among Indian tweets during the COVID-19 era. In addition, this work clarifies public opinion on epidemics and leads public health authorities to a better society.