Detecting the Impact of COVID-19 on Social Media using BERT-Based Model

In recent years, news about the COVID-19 pandemic has inundated both the real and virtual worlds. News and awareness of COVID-19 spread as quickly as the pandemic itself, with a deluge of posts, messages, videos, and tweets. In addition to the health risk posed by COVID-19, mass panic is a cause for concern. Unsurprisingly, media hysteria ensued due to misconceptions, insufficient information, and misleading information about COVID-19 and its implications. Therefore, it is important to ensure that published news on social media platforms is reliable and accurate due to its significant effect on people's sentiments. In this paper, two datasets of COVID-19 tweets are used to conduct sentiment analysis and fake news detection experiments, using different machine learning algorithms (ML) and the BERT model. The results showed that the BERT model outperformed the other models with an accuracy of 97.59.